Barclays Warns $1 Trillion AI Revenue Gap Traps Stock Rally

巴克莱警告1万亿美元AI营收缺口恐困住股市涨势 股票多头面临急剧的战术性重置，因激增的AI基础设施债务和预计高达6000亿美元的技术营收缺口，加之企业股票供应转为净增加，恐将压垮市场流动性。 Alex Altmann深入剖析了战术性仓位、量化因子和资金流向如何在1至8周的周期内主导市场价格走势，而非受长期宏观叙事驱动。尽管自动化的401(k)缴费和韧性十足的散户逢低买盘形成了持久的基准买单，但多策略基金平台杠杆的膨胀以及创纪录的因子拥挤度，使得个股异常脆弱、极易遭受平仓冲击。Altmann警告称，面对庞大的科技资本支出承诺、不断走宽的数据中心债券利差，以及预计到2028年AI营收将出现约6000亿美元的缺口，市场多头正日益深陷被动。随着英特尔和亚马逊等科技风向标在盘后财报公布后股价急挫，Altmann提出了标普500指数或将向其200日均线发生战术性重置的依据。 巴克莱的战术模型揭示了迫在眉睫的6000亿美元AI营收缺口，并警告称超大规模云服务商到2028年底将需要实现1万亿美元的年化AI营收，才能证明当前的资本支出趋势是合理的，而目前预期的年化营收仅为4000亿美元左右。 (60:39) 美国散户家庭净资产中配置于公开上市股票的比例已攀升至35%的历史新高，这受到66%的家庭持股率以及高额房贷利率将年轻投资者挤出房地产市场的推动。 (23:04) 巴克莱的预测表明，由于巨额的科技资本支出对企业股票回购计划构成威胁，2026年市场将扭转长达15年的股票供应缩减趋势，转为股票净发行。 (43:54) 单B级与CCC级企业债之间的利差已扩大至近8%，这一极端的信贷分化自1996年以来仅出现过四次，且此前每次都伴随着标普500指数至少12%的跌幅。 (54:39)

Alex Altmann, head of equity tactical strategies at Barclays, argues that the stock market faces an asymmetric downside risk as massive artificial intelligence infrastructure spending collides with widening credit spreads, surging capital issuance, and crowded positioning. Tactical Modeling and Baseline Odds (9:26) Altmann frames tactical market strategy around baseline mathematical probabilities rather than long-term forecasting. Historical data shows that buying the S&P 500 on a random day and holding for 42 trading days produces a positive return roughly 73% of the time, with an average gain of about 190 basis points. Barclays' quantitative framework, the Barclays Equity Timing Indicator (BETI), aggregates market metrics to seek forward return hit rates between 80% and 90% across tactical horizons spanning one to eight weeks. Altmann emphasizes that quantitative signals must be cross-referenced with qualitative market sentiment to identify dissonance. While financial media and analysts routinely invent retroactive macro explanations for price moves—such as intraday swings in crude oil or AI sentiment—desk positioning and actual flow data often reveal a contrary reality. For instance, while public narratives suggested institutional investors had broadly turned bearish on AI over debt concerns, an internal Barclays survey of 200 institutional investors showed that only 60% viewed data-center financing as a genuine bottleneck. Retail Dominance and the Automated Bid (19:18) Retail investors have structurally outperformed professional institutional hedge funds during this cycle, largely because retail traders face neither tight institutional drawdown limits nor allocator committee scrutiny. Following the adoption of zero-commission trading in October 2019 and the post-pandemic account boom, retail volume has stabilized at around 25% of market share. Concurrently, US households have allocated an unprecedented 35% of their total net worth to equities, while rising mortgage rates and unaffordable home prices have redirected prospective real estate down payments into stock purchases. This retail presence interacts with a multi-trillion-dollar institutional 'relentless bid.' Altmann and Josh Brown point to non-discretionary capital entering equities mechanically via $800 billion to $900 billion in annual automatic 401(k) allocations, approximately $750 billion in baseline corporate buybacks, programmatic advisor ETF rebalancing, and volatility-selling funds. However, Altmann warns this extreme financialization makes retail exposure a fragile macroeconomic vulnerability, as real-time smartphone portfolio monitoring creates an immediate negative wealth effect on consumer spending during severe downturns. Market-Neutral Pods and Factor Crowding (32:36) Beneath calm headline index volatility lies extreme single-stock dispersion: approximately 40% of the 124,750 possible company pairs in the S&P 500 now exhibit negative correlation on any given day. Altmann attributes this dynamic to the expansion of multi-manager relative-value (RV) hedge fund platforms, which grew from $300 billion in assets around 2018–2019 to roughly $1 trillion by 2024. These funds run market-neutral, high-gross and low-net exposure books designed to isolate idiosyncratic alpha while parking unencumbered cash in short-term paper yielding above 5%. Because competing pod shops apply near-identical risk models across an investable universe that has shrunk from roughly 10,000 public equities to 3,000, crowding has been pushed into underlying factors and single-stock pairs. When market-neutral books unwind simultaneously, individual names experience violent swings, dampening headline index volatility while single-stock and factor variance spike. The AI Capex Gap and Capital Issuance (45:22) Altmann cautions that the AI infrastructure buildout faces severe capital constraints. Hyperscaler capex projections are on track to hit $1.2 trillion in 2025 and $1.4 trillion in 2026, creating a requirement for roughly $1 trillion in annual AI-derived downstream revenues by the end of 2028 to justify current spending. However, projected software run rates reach only about $400 billion, leaving an unfunded $600 billion shortfall. This monetization lag coincides with a fundamental reversal in share supply: Barclays projects 2026 will flip from multi-year share buyback retirements into net positive equity issuance, driven by IPOs, convertibles, and at-the-money offerings, alongside $1.2 trillion in net corporate credit issuance. Credit markets are flashing cautionary warnings ahead of equities. The spread between single-B and triple-C rated corporate debt has expanded to nearly 8%—a threshold reached only four times since 1996, with the S&P 500 down at least 12% from peaks in every prior instance, compared to near all-time highs today. Data center bond spreads benchmarked against the 20-year Treasury are also steadily widening, showing that debt markets are beginning to price rising risk premia and pushing leveraged neoclouds like CoreWeave into distress. Altmann views equity bulls as trapped by locked-in capex commitments, a looming corporate competition for debt and equity capital, and stretched systematic positioning across volatility-control and quantitative funds. He sees an eventual downward repricing toward the S&P 500's 200-day moving average—around the 7,100 to 7,200 level, or roughly 17.5 times next-twelve-months forward earnings—as a necessary valuation reset before the secular enterprise technology cycle can sustainably resume.

Barclays 股票战术策略主管 Alex Altmann 指出，巨额人工智能资本支出与下游变现之间存在巨大缺口，叠加多经理平台拥挤头寸与企业发债需求激增，多头资金目前陷入困局，市场或需向 200 日均线回调以重新校准估值。 战术策略定位与短期概率基准 (9:26) 战术交易策略通常聚焦于 1 至 8 周的快速资金流动与仓位变化，而非卖方传统 6 至 12 个月的结构性研究。Altmann 提出，若随机买入标普 500 指数并持有 42 个交易日，历史胜率基准约为 73%，平均回报接近 190 个基点；战术量化模型（如 Barclays 的 BETI）核心在于识别胜率偏离该基准并达到 80% 至 90% 的不对称交易窗口。市场走势往往先于叙事出现，价格波动迫使评论员事后寻找逻辑予以合理化。 散户与系统性资金构筑结构性买盘 (18:34) 美国家庭财富向股市的倾斜达到前所未有的程度，约三分之二的美国家庭持有股票，股票占净资产比重升至 35% 的历史峰值，而高利率导致的高房价促使年轻群体将首付储蓄转向购买碎股。叠加 401(k) 每年约 8000 亿至 9000 亿美元的自动化扣款，以及历史上每年达数千亿美元的企业回购，形成了持续吸收市场回调的“无情买盘”。然而，随着家庭资产负债表与股市高度绑定，以及手机实时推送盈亏，股市大幅下跌将直接通过财富效应冲击终端消费。 多经理平台扩张与因子中性陷阱 (32:36) 多经理对冲基金（Pod Shops）资产规模自疫情前的约 3000 亿美元暴增至约 1 万亿美元。这类机构普遍运行高总头寸、低净头寸的市场中性多空策略，并将约 70% 的未质押现金存放于无风险收益率超 5% 的短期票据中。由于各机构使用趋同的风险模型对行业、规模和风格因子进行完全中性化，加之美股可交易股票池由过去近万只缩减至约 3000 只，导致标普 500 成分股中多达 40% 的股票对在日常呈现负相关性，指数表面的低波动掩盖了个股与因子层面剧烈的反转和踩踏风险。 AI 资本开支与万亿营收缺口 (60:39) 科技巨头在 AI 基础设施上的资本开支预计在 2025 年达到 1.2 万亿美元，2026 年升至 1.4 万亿美元。Altmann 测算，要支撑这一量级的资本开支，到 2028 年底行业需产生约 1 万亿美元的 AI 年化收入，但目前预估的年化营收仅约 4000 亿美元，留下高达 6000 亿美元的资金缺口。作为对比，苹果与微软自 1990 年以来经通胀调整后的累计总营收才达到 9.8 万亿美元，寄望生成式 AI 在短期内凭空创造万亿美元高利润收入脱离了宏观现实。 融资供给逆转与估值均值回归 (45:22) 过去 15 年每年通过回购注销 4000 亿至 6000 亿美元股票的供应红利即将终结，Barclays 预测 2026 年美股将转为净发行，与投资级和高收益企业债约 1.2 万亿美元的净发行以及巨额美国国债供给正面竞争有限的资本池。信贷市场中 B 级与 CCC 级企业债利差已走阔至近 8%，历史上该利差出现时标普 500 指数均自高点下跌至少 12%。Altmann 认为目前无需特定催化剂，多头已处于被动承压状态，标普 500 指数或需向 7100 至 7200 点（即 200 日均线附近、对应约 17.5 倍远期市盈率）回调以重新出清风险。 Altmann 强调其看跌观点严格属于战术层面的仓位出清，并不否定人工智能长期的企业级转型潜能。未来关键在于基础设施融资成本走高与信用利差扩大能否倒逼科技巨头放缓资本开支，或者下游应用能否在估值全面下修前展现出实质性的商业化盈利能力。

## Background 本期内容由 YouTube 频道 The Compound 于 2026 年 10 月 30 日发布，为播客节目《The Compound and Friends》第 266 期。节目由 Downtown Josh Brown 与 Michael Batnick 共同主持，特邀嘉宾为巴克莱银行（Barclays）股票战术策略董事总经理 Alex Altmann。根据官方简介，本期核心议题涵盖战术交易定义、华尔街与散户对 AI 的真实仓位态度、美国家庭股票持有比例、多经理人基金（Pod Shops）的负相关策略、微软发债与数据中心利差，以及高达 1 万亿美元的 AI 商业化收入鸿沟。 ## TL;DR 巴克莱战术策略主管 Alex Altmann 深入拆解了当前美股看似坚韧背后的结构性隐患。随着 401(k) 资金定投和散户因高房价转向全仓股票，美股形成了强大的“无情买盘”，但也让宏观经济对股市回调极度敏感，呈现出“股市即经济”的脆弱性。与此同时，多策略平台型基金（Pod Shops）推崇的市场中性配对交易导致指数内部 40% 的个股对呈现异常的负相关，掩盖了个股与因子层面的剧烈波动。核心矛盾在于，科技巨头锁定了万亿美元级别的资本支出，但行业到 2028 年难以补齐 1 万亿美元的 AI 实际变现收入，加之数据中心债券利差走阔与回购萎缩转为净增发，多头正陷入拥挤陷阱，市场亟待向 200 日均线进行估值重估。 ## Atmosphere & style 整场对话在曼哈顿式的幽默风趣与对冲基金交易台的冷峻量化洞察之间自如切换。主持团队与嘉宾熟络契合，不仅穿插着关于曼哈顿西区房租与外号由来的逗趣调侃，更在盘后科技巨头暴跌之际展现出敏锐而严肃的专业审视。现场氛围既有撕开华尔街“皇帝新衣”的畅快淋漓，又弥漫着对狂热市场结构性脆弱的深层警示，给听众带来如同亲临机构交易室复盘一般的真实紧迫感。 ## Bulletin - [45:22](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2722s) 巴克莱模型预测美股将在2026年迎来罕见的“净发行正增长”，告别过去每年4000亿至6000亿美元的股票净回购注销时代；叠加预计高达1.2万亿美元的企业债净发行与巨额美债供给，资本成本显著推高，使得数据中心长期债券利差加速走阔。 - [23:04](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1384s) 美国家庭股票资产在净财富中占比攀升至35%的历史最高峰，高昂的购房门槛倒逼年轻一代全仓押注股市，401(k)定投、股票回购及投顾被动配置每年向标普500指数输送超1.5万亿美元的刚性买盘，同时也令宏观经济面临更直接的“负财富效应”威胁。 - [01:00:40](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3640s) AI产业变现出现巨大结构性脱节：预计到2028年底全行业需达到1万亿美元的年收入才能覆盖资本支出，而当前预期运行率仅约4000亿美元，巨额缺口导致财报季重仓多头承压被困。 - [32:36](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1956s) 管理规模超1万亿美元的多经理人对冲基金（Pod Shops）普遍执行高毛敞口、低净敞口策略，将微观个股与因子中性化并赚取5%的无风险现金利息，直接引发标普500成分股中高达40%的股票对呈现负相关的罕见割裂格局。 ## Key takeaways - [03:30](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=210s) 巴克莱股票战术策略核心聚焦未来1至8周的“快钱”交易窗口，通过对冲基金PB仓位、资金流向、衍生品数据及BETI模型捕捉前瞻胜率达80%至90%的不对称盈亏比。 - [09:26](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=566s) 在任意随机一天买入标普500指数并持有42个交易日的基准胜率为73%，战术择时的关键不是盲目进出，而是只在胜率显著偏离基准时才选择出手。 - [15:28](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=928s) 市场普遍误以为华尔街已对AI极度悲观，但巴克莱对200家机构的调研显示绝大多数并不看空，对冲基金低净敞口更多反映了多策略平台的中性策略属性而非方向性做空。 - [18:34](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1114s) 散户投资组合在当前周期跑赢机构，主要得益于其无需向投资委员会汇报且没有回撤停摆限制的长期持股定力，逼迫散户认赔离场的往往只有失业或追缴保证金。 - [23:04](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1384s) 美国家庭资产中股票占比达35%的历史峰值，高利率冻结了房地产市场，迫使Y世代与Z世代将首付储蓄转投流动性极佳的股市碎股交易。 - [24:35](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1475s) 401(k)每年8000亿至9000亿美元的自动定投、企业万亿美元级股票回购以及财富管理全面转向ETF被动配置，共同构筑了美股难以被利空撼动的“无情买盘”。 - [28:33](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1713s) 现代经济已高度金融化，全社会养老体系与核心雇员薪酬深度绑定二级市场，这使得美联储在制定货币政策时高度紧盯股市均线底线（如2018年平安夜的紧急转向）。 - [32:36](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1956s) 多策略对冲基金规模膨胀至1万亿美元，通过多空配对消除市场贝塔并享受5%的闲置现金无风险利息，导致大盘指数低波动而成分股内部负相关性达到40%的异常水平。 - [36:18](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2178s) 美股上市公司标的从约10,000家萎缩至3,000家，量化模型在有限资产池内的高度趋同造成了严重的隐性拥挤，导致非财报日几乎全由算法与机器互搏主导。 - [40:55](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2455s) 科技巨头演变为宏观中枢，标普500指数2027年一致预期盈利增长中预计有75%集中在半导体与大型云服务商，使得个股财报剧震直接左右宏观走势。 - [44:12](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2652s) 微软等科技巨头即便交出增速重回30%的无瑕财报，巨额资本开支迫使其寻求250亿美元发债融资，市场定价逻辑已从营收狂欢转向对其现金流代价的严厉拷问。 - [46:53](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2813s) 2026年美股净发行量预计逆转转正，叠加超万亿美元的企业债融资与美债供给，平均久期约17年的数据中心债券利差加速走阔，AI无限低成本融资的幻觉面临现实考验。 - [50:00](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3000s) 依靠高杠杆的新型算力云服务商（Neoclouds）融资压力剧增，市场寄希望于英伟达兜底是危险假设，类似次贷危机中AAA级资产在流动性风暴中同样难逃牵连。 - [56:42](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3402s) 高收益债利差达800个基点在历史上极罕见且以往均伴随标普500超12%的回撤，当前回购执行率下滑叠加大体量IPO井喷，美股供给端正承受类似2000年互联网泡沫前夕的抛压。 - [01:03:14](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3794s) AI产业到2028年底面临1万亿美元资本支出与4000亿美元实际收入之间的巨大预期鸿沟，多头资金已然在拥挤仓位中被困，市场不需要特定催化剂即可发生自发性出清。 - [01:05:21](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3921s) 巴克莱策略团队认为健康的估值重估需要标普500指数向7,100至7,200点（贴近200日均线、对应远期市盈率约17.5倍）调整，以出清拥挤仓位并为下一阶段技术突破蓄力。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=0s) 开场闲聊：曼哈顿西区与商业地产考量 节目在一阵轻松的闲聊中开场。主持人与嘉宾聊起了对纽约曼哈顿各街区的偏好，对比了东区（East Side）与西区（West Side）的居住体验。嘉宾对西区沿哈德逊河（Hudson River）的整条水岸线赞不绝口，指出从炮台公园（Battery Park）、翠贝卡（Tribeca）、苏豪区（SoHo）、肉类加工区（Meatpacking District）、西村（West Village），一路往北穿过切尔西（Chelsea）、哈德逊城市广场（Hudson Yards）、地狱厨房（Hell's Kitchen），直至林肯广场（Lincoln Square）和上西区（Upper West Side），整段风景非常宜人。嘉宾自己就住在河滨大道（Riverside Boulevard），并提到滨河公园现在已经一路向南修通到了第59街，沿河跑步非常舒服，90多街附近还有著名的船坞餐厅（Boat Basin）。 随后话题转到了办公地点与商业地产成本。谈到哈德逊城市广场（Hudson Yards）时，Josh Brown 指出那里的办公楼租金极度高昂：“要碰任何楼层稍微高一点的物业，每平方英尺租金就高达250美元。”（“it's $250 a square foot to touch anything on a decently high floor.”）他坦言自己公司并不属于对冲基金那种“2%管理费加20%业绩提成”（two-and-twenty）的暴利行业，曼哈顿的写字楼动辄需要签署长达十年的租约，因此他尽量推迟敲定新办公室的计划。嘉宾也笑称大家彼此彼此，并提醒未来18个月宏观变数很多，在下届大选前经济很有可能陷入衰退，届时若有一批企业撑不住倒闭，整座城市的商业地产或许会迎来以极具吸引力价格抄底的绝佳机会。 ### [03:30](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=210s) 嘉宾 Alex Altmann 登场与巴克莱战术策略定义 节目正式开场后，Josh Brown 介绍了本期的新嘉宾——巴克莱银行（Barclays）股票战术策略（Equities Tactical Strategies）董事总经理兼全球主管 Alex Altmann（业内常称其为“Alti”）。Alex 领导的战术策略团队负责将交易台的一线动态提炼成面向机构客户的研究成果；在此之前，他曾担任千禧年基金（Millennium）的宏观投资组合经理，并在花旗集团（Citi）掌管了十年的股票交易策略，对买方与卖方的视角均有极深的洞察。 在被问及传统卖方研究与战术策略的差异时，Alex 先是幽默地打趣称所谓的“战术”（tactical）就像军装一样带迷彩、装武器、身上全是多余的口袋，随后他给出了极为严肃专业的界定：传统研究团队通常关注6至12个月乃至更长周期的基本面，而战术团队则主要面向“快钱”（fast money）社区，紧盯1周、2周、4周至最多8周的短线时间跨度。虽然团队也有维持数月甚至数年的结构性宏观主题，但其核心使命是研判未来4到8周内：“市场到底想往哪里走？哪里存在最佳的不对称性？” 此外，战术策略团队深入挖掘大量传统卖方研报极少覆盖的核心垂直领域： - 仓位结构（Positioning）与资金流向（Flow） - 衍生品（Derivatives）数据异动 - 量化因子（Quant styles）与策略偏好 - 对冲基金的实际交易动向 ```mermaid flowchart TD accTitle: 传统研究与战术策略的对比 accDescr: 对比传统卖方研究的长周期视角与巴克莱战术策略的短周期多维度分析 subgraph Traditional["传统卖方研究 (Traditional Sell-Side)"] T1["关注时间跨度: 6-12个月或更长"] T2["核心方法: 单纯基本面与估值分析"] end subgraph Tactical["巴克莱战术策略 (Tactical Strategies)"] TC1["关注时间跨度: 1-8周超短线交易窗口"] TC2["核心垂直领域: 仓位结构/衍生品/量化因子"] TC3["核心目标: 捕捉不对称盈亏比 (Asymmetry)"] end Traditional -.->|"时间跨度与覆盖盲区"| Tactical ``` ### [06:41](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=401s) 仓位分析（Positioning）的本质与不对称交易逻辑 讨论深入到了机构投资者为何如此痴迷于探寻“同行的持仓”。Josh Brown 与 Michael Batnick 调侃道，每当市场出现无法用常理解释的剧烈波动时，大家总会一口咬定是“仓位问题”（positioning）。专业基金经理迫切想知道同行在押注什么，而传统按月发布的投资经理调查往往沦为群体思维（groupthink），参考价值有限。 Alex Altmann 阐明了仓位分析对于研判价格与市场叙事的内在逻辑架构：资金流向驱动价格波动，而价格的走势最终重塑市场叙事（Flow dictates price, and price dictates narrative）。如果市场在某一方向上积聚了极度极端的仓位不对称，这种失衡随时可能以不对称的烈度推动价格反转，进而彻底颠覆整个市场原本深信不疑的宏观叙事。Alex Altmann 强调：“只要你知道市场上积聚了某种仓位不对称性，它就有可能以极不对称的方式改变价格，并在此过程中重塑市场对某一叙事的认知。”（“if there's a positioning asymmetry that has been built in the market, how does it have the propensity to change price in a very asymmetric way and in doing so can shift the market's perception of a narrative”） 在跟踪维度上，Alex 表示虽然团队有能力监控个股仓位，但日常操作中更倾向于聚合到宏观主题、行业板块、量化风格以及全市场大盘层面： - 对冲基金（Hedge Funds）的净敞口与总敞口 - 公募互惠基金（Mutual Funds）的仓位偏好 - 商品交易顾问（CTAs）的趋势跟踪动量 - 波动率控制（Vol Control）与量化投资策略（QIS） - 风险平价策略（Risk Parity） - 散户投资者（Retail Investors）的资金动向 针对发现的极端拥挤仓位，Alex 澄清说这并不意味着无脑进行逆向交易（contrarian bet）。战术团队会将所有的仓位报告输入量化模型，与相关标的随后的前向历史回报进行系统性回测。很多时候，回测结果会推翻直觉——即便某类资产的持仓极度拥挤，历史数据显示拥挤度对后续走势毫无负面影响，甚至会进一步强化原有的上涨动能。仓位数据必须与严格的回测结合，才能成为具有真正指导意义的情报工具。 ### [09:26](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=566s) 战术交易与标普 500 的胜率基准 对话从短期预测究竟是徒劳还是合理展开。Josh Brown 直言，对于大多数普通散户而言，试图预测短期走势完全是在浪费时间；但在如今变化极快的投资世界里，指望对一年后的情况保持清晰认知同样荒谬，因此如果真要做战术判断，0 到 30 天的窗口反而是最合理的。 巴克莱（Barclays）的 Alex Altmann 用一组基准统计数据印证了这一点：如果在任意随机的一天买入标普 500 指数并持有 42 个交易日（约两个月），赚钱的概率高达 73%，平均回报率略低于 200 个基点（约 1.9%）。这远胜于去赌场玩任何游戏的胜率。Alex Altmann 解释称，这个 73% 就是基准线，被动持有在绝大多数时候都能赢，而战术择时的核心，就是判断当前的胜率是否已经显著偏离了这 73%。 Michael Batnick 借用赌徒与棒球的比喻指出：普通赌徒在周末 20 场比赛里会盲目押注 11 场，根本没有优势；而真正的专业交易者不必对每一个球都挥棒，80% 的时间属于噪音、什么都不做，只在数据明确显示有 54% 击中概率的 20% 时机里出手。Alex Altmann 赞同并补充道，在巴克莱的模型中，他们通常寻找的是命中率高达 80%、85% 甚至 90% 的场景，无论是针对标普 500 指数、特定风格因子还是行业板块。 - 标普 500 指数任意随机买入并持有 42 个交易日（约 2 个月）的赚钱概率为 73%，平均回报率约 190 个基点。 - 战术交易的本质不是每天进出，而是在模型胜率显著偏离 73% 基准、达到 80%–90% 极值时才选择“挥棒”。 ### [12:26](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=746s) 科学与艺术的结合：巴克莱 BETI 模型与叙事噪音 Alex Altmann 介绍了巴克莱用于判断多空的核心量化框架——“BETI”（Barclays Equity Timing Indicator，巴克莱股票择时指标）。他笑着透露，原本团队打算将其命名为“BETTING”（博彩），但不出意料地被合规部门否决了。他坦言，市场上所有的数据都是公开的，没有任何秘密配方，BETI 只是将市场上现有的海量指标整合在一起。当足够多的指标同时闪烁做多或做空信号时，从统计学上看未来标普指数的走势概率便一目了然。 但纯粹的量化只是“科学”部分，真正的“艺术”在于如何结合主观定性分析。Alex Altmann 强调，这正是 AI 目前尚未弄明白的部分——如何将量化数据、持仓情况、客户交流以及市场主流叙事拼接在一起，决定何时全力出击。 Josh Brown 提出了对“市场叙事”形成机制的疑问：市场上往往是价格走势先出来，随后大家找个由头把它当成公认的叙事，比如周四有五个人异口同声说出同一种解释，而这种反向用价格倒推逻辑的现象几乎随处可见。Alex Altmann 表示完全赞同，并指出这本质上就是噪音： > “人们会用价格来合理化叙事，或者因为价格走出某种形态而在事后去解释它，而不是真正提供背后的情报并去质疑这种观点。” 他举例说，在当下的原油市场尤其如此，油价随便波动一两美元，大家就会把原因归咎于某个不知名人士发布的一条无足轻重的新闻标题，虽然可能存在算法自动抓取标题交易的情况，但绝大多数时候纯粹是市场噪音。 ```mermaid flowchart TD accTitle: BETI 决策流程 accDescr: 展示巴克莱 BETI 指标如何结合量化信号与定性叙事做出战术交易决策 subgraph Science ["科学层：量化框架 (BETI)"] A["公开市场多维度数据"] --> B["聚合指标多空信号"] B --> C["统计学前瞻胜率 (80%-90%)"] end subgraph Art ["艺术层：定性与仓位修正"] D["客户交流与市场情绪"] --> E["排查价格倒推的叙事噪音"] F["对冲基金 PB 真实仓位"] --> E end C --> G{"信号与叙事产生背离？"} E --> G G -->|"是：存在认知错配"| H["果断出击 (Take a swing)"] G -->|"否：共识过度拥挤"| I["保持观望 (Do nothing)"] ``` ### [15:28](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=928s) 华尔街真的对 AI 悲观吗？认知错配与对冲基金持仓真相 针对 Michael Batnick 提出的“既然大家看的是同一套数据，定性判断到底占多大比重”这一问题，Alex Altmann 明确指出，绝不能 100% 依赖量化。完全封闭在黑匣子里的量化或许有人做得很成功，但他自己的管钱经历证明，走入市场群体、感受时代思潮（Zeitgeist），并观察定性反馈与量化框架是否出现背离（dissonance），才是超额收益的最大来源。 当前的 AI 交易就是一个经典范例。市场上普遍充斥着一种说法，认为由于近期的融资瓶颈，所有人突然都对 AI 极度悲观。然而巴克莱刚刚对 200 位华尔街专业机构投资者进行的调研显示，结果绝大多数人根本不看空 AI： > “只有 60% 的受访者认为我们面临融资问题。我就在想，这怎么可能？过去两周我们刚刚经历了一场融资困境，这个数字本该是 100% 才对。” Alex Altmann 指出，这种调研结果与客户口头反馈之间的断层，恰恰意味着市场存在认知偏差。更深一层的错觉来自对冲基金的主经纪商（PB）持仓数据。一些竞争对手的大行研报经常根据 PB 数据惊呼“对冲基金净敞口降至极低，全员看空并抛售了科技股”。但他指出，这种分析忽视了一个关键事实：许多对冲基金净敞口的波动区间其实非常狭窄，往往只是在 48% 到 55% 之间微调，随后又回到 48%。造成这种现象的原因在于，大量机构本质上是相对价值（RV）的多策略平台型基金（Pod Shops），它们执行的是严格的市场中性策略，其净敞口的微小变化根本不代表对科技股或 AI 方向性的真实看空。 ### [18:34](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1114s) 散户是“愚蠢的钱”还是聪明钱？ Alex Altmann 解释了巴克莱如何追踪市场资金定位（positioning）。虽然量化对冲基金（Pod shops）的贝塔敞口不高，但散户群体是美股绝对的第一大持有者，其次是共同基金、杠杆 ETF、波动率控制基金（Vol-control funds）、量化投资策略（QIS）以及商品交易顾问（CTA）。当这些维度的量化画像建立起来后，就能与当前盛行的市场叙事以及客户的实际反馈进行交叉比对。 针对 Michael Batnick 提问散户在当前周期中频繁逢低买入（buy the dip）究竟是盲目的“愚蠢资金”（dumb money）还是聪明钱，Alex Altmann 毫不客气地指出事实：绝大多数散户投资组合的表现实际上都跑赢了机构投资组合。Alex Altmann 强调，散户拥有的最大优势就是更强的持股定力（holding power）： > “绝大多数散户投资组合都会跑赢机构投资组合，这告诉你了什么？” 对冲基金面临极为严苛的单月或季度最大回撤限制，而且随时处于外界的监控之下；而普通散户根本不需要向投资委员会解释自己为什么超配半导体股票，也不用为账户净值回撤 400 个基点写检讨，他们唯一的观众只有自己和伴侣，甚至还热衷于在晚餐派对上向朋友吹嘘自己如何抄底。Josh Brown 总结道，能迫使散户认赔离场的真正原因只有两个：要么是经纪账户遭遇追缴保证金（margin call），要么是自己或邻居彻底失业。 回顾散户资金的历史，Alex Altmann 澄清了一个常见误区：零售交易真正的分水岭并不是 2020 年疫情，而是 2019 年 10 月。当时嘉信理财（Charles Schwab）和富达（Fidelity）等传统巨头相继宣布将交易佣金降至零，加之 Robinhood 等主打游戏化界面的平台崛起，交易摩擦成本被降到了近乎为零。随之而来的疫情封闭、巨额政府纾困支票以及全民闭门不出，引爆了 18 个月内新增 4000 万个经纪账户的狂潮。虽然在 2022 年美联储激进加息期间，英伟达（NVIDIA）和 Meta 等明星股重挫 75%，SPAC 投机潮破裂导致散户损失惨重，但散户并未彻底撤离市场，其交易量占比始终维持在 25% 左右的高位，并持续在随后的反弹中逢低买入。 --- ### [23:04](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1384s) 破纪录的家庭持股与冷冻的房地产市场 当讨论到当前散户数据与历史（如 1990 年代末互联网泡沫）的对比时，Alex Altmann 直言如今的市场参与度早已彻底打破了 1990 年代的记录。目前全美约有三分之二的家庭直接或间接持有股票；更震撼的数据在于，美国家庭净资产中股票资产的占比已达到 35%，创下历史最高纪录。作为传统财富蓄水池的房产净值（home equity）虽位列第二，但股票与房产在财富构成中的差距已经扩大到了 1960 年代以来的历史最高水平。 面对这种分化，Michael Batnick 与 Josh Brown 认为因果关系显而易见：过去三年由于利率飙升，全美按揭贷款利率居高不下，二手房与新房交易陷入冰封，整个房地产市场陷入僵尸般的滞胀萧条，房价增长停滞；与此同时，美股大盘却直接翻倍。对于手头攒下了储蓄的 Y 世代（千禧一代）和 Z 世代年轻人而言，高昂的购房门槛将他们拒之门外，他们只能放弃买房的首付计划，把钱持续砸进流动性更好、造富效应更显著的股票市场中。 --- ### [24:35](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1475s) 401(k)与万亿美元级别的“无情买盘” 这种向股市集中的财富转移，演化成了一股对估值完全不敏感的自动化买盘。Alex Altmann 详细拆解了支撑美股底层的刚性资金流向：原本每年可能流向房屋首付的 2000 亿至 3000 亿美元资金转入股市；同时，美国家庭的 401(k) 退休金定投每年源源不断地输送 8000 亿至 9000 亿美元；再加上美股上市公司每年高达 7500 亿至 1 万亿美元的股票回购，市场上存在着超过 1.5 万亿至接近 2 万亿美元的刚性买入力量。 ```mermaid flowchart TD accTitle: Structural Inflows Supporting US Equities accDescr: Diagram illustrating the three automated pillar flows driving the relentless bid in US equities. subgraph CapitalSources["核心刚性资金来源"] A["401(k) 自动定投<br/>年均 8,000 - 9,000 亿美元"] B["上市公司股票回购<br/>年均 7,500 亿 - 1 万亿美元"] C["年轻人置业转向资金<br/>年均 2,000 - 3,000 亿美元"] end subgraph DeliveryMechanism["被动与财富管理渠道"] D["目标日期基金与定投规则<br/>（对估值不敏感、自动化划扣）"] E["理财顾问与 ETF 资产配置<br/>（买入标普500大盘权重股）"] end subgraph MarketImpact["市场宏观效应"] F["构筑‘无情买盘’ (The Relentless Bid)"] G["美股波动率平抑器<br/>（VIX 恐慌指数快速消退）"] end A --> D B --> F C --> E D --> F E --> F F --> G ``` Alex Altmann 说明了 401(k) 数据的测算逻辑：全美家庭中扣除无力缴纳的部分，按美国家庭收入中位数 88,000 美元以及平均 12% 的薪资缴存比例计算，保守扣除后每年流入美股的资金稳定在 8000 亿至 9000 亿美元之间。Alex Altmann 强调这种流入就像天气一样确定： > “401(k) 资金就像天气一样确定，只要大家还有工作，它就会源源不断地涌进来。” Josh Brown 补充道，他在十几年前（2014 年前后）就曾在博客中将这种现象命名为“无情买盘”（The Relentless Bid）。当时标普 500 刚在 2013 年突破历史新高，市场面对欧债危机等利空时虽有震荡，但总能迅速收复失地并再创新高。Josh Brown 指出三大核心动力共同构成了这一机制： - **股票回购缩减流通盘**：上市公司从过去的派息转向万亿美元级别的股票回购，持续收缩市场流通股总股数。 - **401(k) 的被动注入**：与退休目标日期挂钩的薪资自动扣款，不看估值、不看牛熊，按月定投买入。 - **财富管理商业模式的根本转型**：华尔街从传统的股票经纪佣金制（broker trading）全面转向按资产管理规模收费（fee-based）的财富规划模式（以摩根士丹利为代表）。理财顾问不再频繁炒单，而是将数万亿美元长期配置进宽基指数 ETF。 Josh Brown 认为，这种模式将巨额资金持续注入标普 500 指数的权重股中，充当了市场的“波动率平抑器”（volatility dampener）。这也是为什么近十年来每次恐慌指数 VIX 冲高，往往两周后就会迅速被涌入的 401(k) 和被动买盘浇灭，使美股每一次回调都变成转瞬即逝的买入机会。 ### [27:30](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1650s) 散户全仓股票：买不起房与碎股投资的结构性转变 Alex Altmann 指出，当前市场除了有规律的定期再平衡资金和投顾资金入市外，还充斥着备兑看涨期权 ETF（covered call overwriting ETFs）以及刻意做空波动率的产品，这些机制向市场持续供应 Gamma 和 Vega，进一步压低了波动率。然而，真正令市场底色发生深刻转变的，是散户投资者几乎“全仓（all-in）”压注股市的现实。Alex Altmann 解释道，这种激进配置在某种程度上是被迫的，因为年轻一代对拥有住房的梦想已经失去信心。面对美国中位数在 30 万至 40 万美元的房价，年轻人很难一次性拿出 5 万或 6 万美元的购房首付；相比之下，碎股交易（fractional shares）极其便捷且门槛极低，他们随时可以买入一股英伟达（NVIDIA）或一股 SPY 然后抛在脑后。 这种将全部财富押注在股票上的资产结构，为整体经济埋下了巨大的脆弱性。当被问及究竟什么最让人夜不能寐时，Alex Altmann 坦言，既不是杠杆 ETF，也不是地缘政治或 AI 泡沫，真正令人担忧的是散户的脆弱性：“It's retail is the scary one in the sense that because we've got such a financialized economy, right?” 他回顾了此前关税风波期间标普 500 指数下跌 20% 的情景，随后的两个月内，全美消费出现了极其明显的滑坡。Michael Batnick 随即指出，这正是经典的“负财富效应”（negative wealth effect）。Alex Altmann 补充说，如今这种冲击是实时的，与 2000 年互联网泡沫破裂时截然不同——当年家庭虽然也持有大量股票，但投资者往往需要等待每月或每季度的券商纸质对账单，而现在的散户手机每天都会弹出几十条资产变动推送，随时承受财富净值起伏的心理冲击。 - **购房门槛倒逼**：美国房屋中位数约在 30 万至 40 万美元，首付需要 5 万至 6 万美元；而股市碎股交易让年轻人可以零门槛建仓优质资产。 - **波动率压制结构**：做空波动率策略与覆盖看涨期权 ETF 持续向衍生品市场提供 Gamma 与 Vega，压制了市场的自然波动。 - **即时财富反馈**：相较于互联网泡沫时期的月度/季度对账单，移动端全天候推送让散户资产的“负财富效应”传导速度大幅加快。 --- ### [28:33](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1713s) “股市现在就是经济”：实时财富效应与美联储的隐形底线 谈及股市与宏观经济的关系，Josh Brown 直截了当地打破了华尔街流传多年的传统教条：“The old adage was stock market is not the economy. No, it is the economy now.” 他强调，过去人们常常将股市视为衡量公司经营状况的粗糙代表，质疑 IBM 或可口可乐的股价与平民大众的日常生活究竟有何关联；但如今，整个社会的养老体系（401(k)）完全建立在股票市场之上，所有高收入家庭都持有巨额退休账户，企业员工的薪酬体系更是与股票深度绑定，新生代百万富翁几乎全部来自即将上市、刚上市或拥有高估值股票的科技企业。在这种结构下，现代经济根本不可能在承受长达两年的持续熊市的同时还不发生严重衰退。 Alex Altmann 完全赞同这一判断，并指出绝大多数顶尖宏观经济学家的计量经济模型中，竟然连一个专门追踪标普 500 指数的变量单元都没有。然而，真正的监管决策者却时刻盯着盘面。Josh Brown 回顾了 2018 年美联储的典型转向案例：当年 12 月初美联储还坚称利率离中性水平“还很遥远”，但随后标普 500 指数一路暴跌至 21%，在技术上跌入熊市；就在那年平安夜，美联储的态度发生了一百八十度大转弯，一个月后不仅宣布利率已处于中性水平，甚至直接开启了降息周期。这一幕充分表明美联储实际上在紧盯 200 日均线或 10 个月移动平均线来感知政策极限。Josh Brown 与 Alex Altmann 还调侃道，2008 年金融危机期间华盛顿的美联储总部总共只有一台彭博终端（Bloomberg Terminal），如今美联储和财政部不仅终端遍地，甚至常年与对冲基金保持实时沟通。 ```mermaid flowchart LR accTitle: 股市对现代宏观经济的全面传导机制 accDescr: 描绘股市波动如何通过退休金、高管薪酬与美联储反应函数直接驱动宏观经济 subgraph WealthChannel["财富与消费链条"] direction TB A["401(k) 养老账户<br/>与股权薪酬绑定"] --> B["实时资产估值变动<br/>（手机即时推送）"] B --> C["消费支出扩张 / 收缩<br/>（财富效应）"] end subgraph PolicyFeedback["政策反应链条"] direction TB C --> D["经济基本面反馈"] B --> E["股市跌入熊市边界<br/>（如 2018 年触及 -21%）"] E --> F["美联储态度逆转<br/>（暂停加息 / 开启降息）"] end WealthChannel --> PolicyFeedback ``` - **薪酬与财富高度金融化**：全美高收入群体的资产净值与企业薪酬完全由股票价值主导，使得经济抗风险能力与指数表现捆绑。 - **2018 年平安夜转向**：美联储在指数跌破 20% 熊市线后迅速改口，印证了所谓的“美联储看跌期权（Fed Put）”高度锚定技术均线。 - **监管机构的数据化演变**：美联储内部从 2008 年仅有一台彭博机，演变至如今全员紧盯二级市场流动性与金融状况指数（FCI）。 --- ### [32:36](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1956s) 标普 500 内部脱钩：40% 负相关与多经理人对冲基金的套利逻辑 Michael Batnick 提出了一张来自法国兴业银行（Société Générale）的异常图表：在标普 500 指数内部可能构成的 124,750 组股票对中，目前有高达 40% 的股票对呈现负相关关系。今年市场频繁上演指数微跌、但大量成分股反向大涨的分化行情。例如当天受科技巨头财报影响，标普 500 指数下跌 53 个基点，但标普等权重指数却逆势上涨 98 个基点。Michael Batnick 梳理了背后的阶段性驱动因素：年初是能源股逆势拉升带来奇异的负贝塔（Negative Beta），随后利率攀升打击了必需消费品、公用事业与房地产投资信托（REITs），而以苹果为首的巨头岿然不动，再叠加 AI 软件股反反复复的资金切换。 对于这种深度脱钩的微观机理，Alex Altmann 给出了来自机构交易视角的底层逻辑：这背后全是由多策略/多经理人平台（Multi-Manager Pod Shops，如 Citadel、Millennium 等）驱动的相对价值（Relative Value, RV）策略。在 2018 至 2019 年疫情前，这类多策略机构的管理资产规模（AUM）仅约为 3000 亿美元，而如今已膨胀至约 1 万亿美元。这类平台要求旗下投资组合经理运行“高毛敞口、低净敞口（High Gross, Low Net Exposure）”，其使命是剥离系统性贝塔、专注于榨取微观个股的特质阿尔法（Idiosyncratic Alpha）。高利率环境更让这一策略成为躺赢的完美生意：“In a high interest rate environment RV strategies are absolute no-brainers right because I deploy a lot of growth on my long side load of gross on my short side.” 经理人在做多与做空两端放大毛敞口以实现市场中性，实际上并不占用太多净资本；在向主经纪商（Prime Broker）缴纳约 15% 到 30% 的风险抵押保证金后，剩余高达 70% 的闲置现金可以直接存入安全资产，稳稳赚取 5% 甚至更高的无风险基准利率。这种数千亿美元级别的中性多空配置，直接加剧了板块与个股之间的负相关割裂。 - **标普 500 内部对冲脱钩**：法国兴业银行统计显示，当前指数内部 40% 的个股对呈现负相关，远超历史常态。 - **多策略资管规模爆发**：多经理人平台管理规模从 2019 年的 3000 亿美元暴增三倍至目前的 1 万亿美元。 - **高利率下的 RV 套利闭环**：通过极致的多空配对冲销市场风险，多策略基金不仅赚取微观价差阿尔法，还能将大比例闲置资金配置在 5% 的短期高息资产上。 ### [36:18](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2178s) 对冲基金的“中性陷阱”：拥挤交易与负相关波动 Alex Altmann 解释了多策略平台型基金（Pod Shops）的运作逻辑。这些机构为了追求低净敞口（Low Net），会用严格匹配的多头与空头进行对冲，将未配置的闲置资金用来吃无风险利率。整个体系运行极为严苛，基金经理必须严格遵守风险限额，本质上是在组织层面扮演着高效的资金管理（Treasury）功能。其核心目标非常纯粹：剔除一切宏观和风格波动，仅提取纯粹的个股超额收益（Idiosyncratic Alpha）。 然而，这一模式在全市场普及后催生了巨大的系统性隐患。当所有 Pod Shop 都在使用同一套风险量化模型，将价值、质量、市值规模、成长等风格因子以及行业、宏观敞口全部中性化时，市场的可交易股票池却在急剧缩水。Josh Brown 指出，美股上市公司数量已从他刚入行时的约 10,000 家骤降至如今的大约 3,000 家；加上近年来 IPO 稀少而企业股票回购激增，市场形成了事实上的“负净发行”。当规模庞大的资产管理规模（AUM）叠加主经纪商（Prime Brokerage）提供的杠杆，一齐涌入这个不断缩小的资产池时，灾难性的拥挤效应不可避免。 ```mermaid flowchart TD accTitle: Pod Shop 运作与拥挤解杠杆机制 accDescr: 展示多策略量化模型如何将风险推向未知角落并引发踩踏 subgraph FundMechanics ["Pod Shop 运作机制"] A["多空仓位严格中性化"] B["剥离风格/行业/市值因子"] C["提取纯粹个股 Alpha"] A --> B --> C end subgraph MarketConstraints ["市场环境制约"] D["美股标的从 10,000 缩减至 3,000 家"] E["回购激增 + IPO 枯竭 (负净发行)"] F["AUM 膨胀 + 经纪商高杠杆"] D & E & F --> G["资金极度拥挤"] end subgraph UnwindShock ["踩踏与分化"] C & G --> H["风险被挤压至未察觉的隐藏维度"] H --> I["单一因子同时被动平仓"] I --> J["大盘指数低波动 / 单股与因子剧烈暴跌"] end ``` 这种机制导致投资人自以为做了完全对冲，实际上只是将风险挤压到了未曾察觉的隐蔽角落。一旦出现被迫平仓，所有人同时抛售某一部分仓位，造成的冲击会呈倍数放大。Alex Altmann 极其精辟地总结了当前这种撕裂的市况：“这就是为什么大家看到指数几乎没有波动，但因子波动、个股波动以及负相关性却全面爆发。”（“...which is why you're seeing very little index vol and an absolute ton of factor V single stock V negative correlation.”） - 标的池缩水：美股可交易股票数量历史峰值约为 10,000 家，当前已缩减至约 3,000 家左右。 - 供需结构失衡：企业股票回购额远超 IPO 融资金额，形成美股市场整体的负净发行。 - 杠杆集中：主经纪商（PBS）在有限标的上叠加多倍杠杆，推高了剧烈平仓（Vicious Unwinds）的破坏力。 --- ### [38:17](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2297s) 机器博弈与“希腊神话”叙事：财报日与非财报日的流动性格局 由于量化模型在相同标的篮子中高频换手，许多业务关联度有限的股票在盘面上走出了完全一致的形态。Michael Batnick 举例称，Spotify 和 Netflix 在长达一年的时间里K线走势“完全一模一样”，就像从一个模子刻出来的。Alex Altmann 透露，这正是机器与机器博弈的结果：高频量化基金（Quant Funds）运行着实时统计套利（Stat-Arb），捕捉日内或单日的价差做均值回归；而主观阿尔法团队（Discretionary Alpha）则侧重 3 到 6 周的周期，常围绕财报日前后做单股对冲套利。 这种资金参与者的结构导致了明显的日度分水岭：在财报公布当天，单股交易流绝大多数由主观选股团队驱动；而在非财报日，绝大部分市场交易流完全由量化模型和散户接管。Josh Brown 对此提出了尖锐的文化观察：财经媒体面对这些复杂的统计套利和算法博弈结果时，由于普通大众无法理解机器互搏的冰冷逻辑，只能生搬硬套编造故事。这就像古希腊人无法解释日升月落、地震火山，便编造出阿波罗驾着太阳车、神灵发怒的神话一样；如今的媒体也总是执着于寻找一条新闻标题，煞有介事地解释“为什么市场今天对 AI 改变了看法”。 在谈到 Barclays 的业务时，Alex Altmann 介绍了自己带领的团队——Barclays 的股票战术策略团队（BETS 团队）。该团队在全球共由 11 人组成，其中纽约 3 人、伦敦 3 人，近期在亚洲新扩充了 5 人。Josh Brown 顺着话头打趣求证：“传闻你们团队是‘只准开枪击杀’（Allowed to kill）是真的吗？”Alex Altmann 笑着接梗幽默回应：“我们绝对是‘只杀不抓’，我们从不干活捉俘虏的营生。”（“We are kill only. Yeah, capture. Yeah, kill. Yeah, we're not we're not in the capture business.”） --- ### [40:55](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2455s) 财报雷区与宏观化的科技巨头：微观个股吞噬宏观格局 谈到财报季对战术交易的干扰时，Michael Batnick 提到 ServiceNow 在财报发布后股价单日重挫 27%，并询问团队在财报期是否会设定交易静默期。Josh Brown 直言财报季对传统策略而言简直就是“雷区”（Landmines），“根本没有任何战术结构可言，电话会议里随便一句评论，就能让一只股票瞬间从暴涨 10% 砸成暴跌 10%。” Alex Altmann 坦言，BETS 团队在宏观主题维度通常会在财报期保持观望，交由深耕行业的专职销售团队处理。但当下的美股已经发生了质的蜕变——股票资产本身正变得越来越宏观化（Equities have become more macro）。AI 已经成为驱动宏观经济的关键主题，超大规模云服务商（Hyperscalers）不再只是普通散户或机构眼中的单个软件代码，而是吸引了信用债交易员、宏观对冲基金以及散户等全市场各路资金死盯的宏观锚点。Altmann 凭借早年摩根大通个股分析师的职业老本行经验，带领团队利用自然语言处理（NLP）等工具对巨头电话会议的每一个措辞进行严密扫描。 随后，Alex Altmann 抛出了一个令人咋舌的宏观数据预测，揭示了指数对极少数巨头病态的依赖度：“明年也就是 2027 年，标普 500 指数一致预期的盈利增长中，预计有 75% 将完全来自半导体和超大规模云服务商。”（“So, next year, next year 2027, consensus S&P earnings, 75% of that is expected to come from semiconductors and hyperscalers.”） - BETS 团队规模：全球共 11 人（纽约 3 人、伦敦 3 人、亚洲 5 人）。 - 利润高度集中：标普 500 指数 2027 年一致预期盈利的 75% 押注在半导体与大型云服务商身上。 - 资本开支极端化：Meta 当前已将其总营收的 70% 砸入资本支出（CapEx）。 - 经济支柱动能：全美约三分之二的经济增长，动力直接或间接来自与 AI 相关的资本开支扩张。 --- ### [44:12](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2652s) 微软的增长悖论：无懈可击的财报与大跌的股价 讨论转向具体的财报表现与融资动态。Michael Batnick 调出图表，展示了微软云业务（Microsoft Cloud）令人难以置信的扩张节奏。数据显示，微软云业务的同比营收增速正在反身向上加速，飙升至 30%。这创下了自 2022 年第三季度以来的最高增速纪录。 更令人震惊的在于体量的剧变：2022 年第三季度该业务增速处于类似水平时，其季度营收规模为 230 亿美元；而如今，在基数膨胀到接近当年三倍的体量下，微软竟然依然能交出 30% 且加速的增长曲线。Batnick 惊呼这种局面是“无法建模、不可思议、无法预测且史无前例的”。然而，就在发布近乎完美季报的同时，微软宣布了 250 亿美元的债券融资发行计划，其股价随即在盘面承压下挫。 面对这种业绩超预期但股价下跌的背离局面，Alex Altmann 犀利地反问道，市场在经历了最初的狂欢之后，核心拷问已经发生了转移：“如果你们砸了这么多的钱，眼下固然换来了惊人的营收增长，但这一切未来究竟要以牺牲什么为代价？”（“...if you're spending that much money and right now, yeah, you you've got phenomenal revenue growth, what's it going to come at the cost of, right?”）巨额的资本支出若演变为无休止的债务扩张，即便业绩亮眼，市场也开始对巨头未来的资产回报率和现金流代价打上巨大的问号。 ### [45:22](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2722s) 巨额回购时代的终结与净发行逆转 节目讨论到微软（Microsoft）等巨头时，嘉宾提到其虽然股价接近历史最高点，但营运现金流依然为正，甚至可能是少数仍能维持强劲正现金流的公司。然而，市场的门槛已经被抬得极高。过去 15 年间，美股一直处于大规模股票回购的黄金期，如今这一趋势正在发生剧烈逆转。市场不再是单纯的注销股票，而是迎来了巨头公司的密集股票发行，这些公司在指数中占据着举足轻重的权重。 Alex Altmann 指出，巴克莱（Barclays）团队的模型预测显示，2026 年美股将转为“净发行正增长”（positive net issuance），这几乎是现代市场历史上极为罕见的转折点。过去数年，市场每年通常能通过回购净注销 4000 亿到 6000 亿美元的股票，如今这数千亿美元的资金缺口发生了逆转。更严峻的是资金竞争：不仅是股市在发行股票，2026 年投资级（IG）和高收益级（HY）企业债的净发行量预计也将达到 1.2 万亿美元，加上美国政府为了弥补巨额财政赤字也在大规模发债借钱。尽管市场仍在吸收这些供给，但风险溢价（risk premia）已经因此显著走阔。 - **净回购逆转**：美股从每年净减少 4000 亿至 6000 亿美元供给，转变为 2026 年的净发行正增长。 - **资金争夺战**：2026 年投资级与高收益级企业债净发行预计高达 1.2 万亿美元。 - **政府借贷压力**：美债发行量居高不下，与企业融资产生正面竞争。 --- ### [46:53](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2813s) 数据中心债券利差走阔与 AI 融资成本现实 Alex Altmann 引导大家查看图表 11。该图表追踪了一个等权重的数据中心债券利差指数，对比的是平均久期约 17 年的基准（采用 20 年期美债）。数据显示，这一利差过去一段时间在稳步走阔，如今走阔的速度甚至还在加速。这表明市场虽然仍在接纳这些债券，但资金正在要求更高的价格补偿。 当前市场的主流叙事往往认为“这笔 AI 交易无论多贵都讲得通”，无论是英伟达（Nvidia）、Meta 还是微软的高管在财报电话会上展示的营收数据，似乎都在强化这种认知。然而现实却截然不同：AI 融资是有成本上限的。市场目前正处于试探阶段，试图找出究竟在什么价格水平上，投资者会彻底拒绝买单。 ```mermaid flowchart LR accTitle: AI 融资流动与利差压力 accDescr: 展示 AI 基础设施投资从企业债市场融资导致利差走阔并反噬市场的传导链条。 subgraph 融资端["资本市场供给压力"] A["美国国债巨额赤字发行"] B["1.2 万亿美元企业债发行"] end subgraph 传导端["AI 数据中心融资"] C["数据中心债券集中承销"] D["久期匹配基准：20 年期美债"] end subgraph 结果端["市场定价修正"] E["数据中心债券利差加速走阔"] F["资本成本上升挤压高杠杆算力玩家"] end A --> C B --> C C --> D D --> E E --> F ``` 随后展示的图表 12 揭示了自 1995 年以来的资本竞争常态。在经典企业金融和经济学 101 中，资本成本（以两年期美债收益率为代表）越高，发行量占市场规模的比例理应越低。但眼下市场却把常识抛在脑后，仿佛 AI 可以无视重力。Alex Altmann 尖锐地反驳了这一幻想： > “数据中心债务图表正在告诉你，你无法忽视它，因为利差正在走阔，而且它们将继续走阔，直到有效地测试或击垮这个市场。” - **数据中心债券久期**：平均久期约为 17 年，对比 20 年期美债。 - **历史相关性反常**：自 1995 年以来，高利率通常压制资本发行，但当前 AI 叙事正在强行对抗这一规律。 - **裂痕显现**：利差飙升首先对高杠杆的远期算力押注产生实质性冲击，CoreWeave 和新型云服务商（Neoclouds）已率先感受到压力。 --- ### [50:00](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3000s) 资产负债表分化与高杠杆 Neoclouds 的困境 图表 13 进一步展示了股市底层正在发生的理性轮动——向优质资产靠拢（flight to quality）。Alex Altmann 解释，很多人只盯着标普 500 指数的点位，以为它只会一路上涨，却忽略了指数内部极高效率的定价修正。该图表通过多头优质股票篮子与空头脆弱股票篮子的相对价值交易（RV trade），完美映射了 CCC 级与 BB 级信用利差的走势。逻辑非常直观：当借贷成本变高时，依赖借债生存的公司所面临的危机远比现金储备雄厚的企业严重得多。如果债券市场对脆弱企业关上大门，它们就不得不被迫转向股权市场低价增发，导致股价受到直接重创。 主持人和嘉宾随即就新兴云算力公司（Neoclouds）的命运展开激烈讨论。市场目前对 CoreWeave 等玩家能否继续以过去的速度融资深存疑虑。Josh Brown 抛出一个关键问题：是否存在一种可能，即这些 Neoclouds 最终彻底崩溃，但危机完全不会蔓延到传统的超级云服务商（Hyperscalers，如微软、谷歌、亚马逊），甚至因为竞争对手出清而强化传统巨头的垄断优势？ Alex Altmann 将这称为“交易之后的交易”（the trade after the trade）。针对市场上关于 CoreWeave 市值回落后英伟达是否会出手兜底甚至全盘收购的传闻，Alex 态度冷峻。他指出，黄仁勋在公开表态中已经用企业级外交辞令让 CoreWeave “靠边站”（go pound sand），明确表达了无意伸出援手的立场。CoreWeave 目前股价仍在 IPO 发行价之上，即便英伟达要收购，也只想要其算力资产而非股权，完全可以在破产清算阶段去买打折的债券。 关于是否存在托底市场的“黄仁勋看跌期权”（Jensen Put），Alex 警告称，英伟达已经成为整个 AI 生态的红太阳，若发生严重的洗盘，即便巨头下场扫货，代价也可能是其自身股价的重挫。他用次贷危机（Subprime）时期的机构举例，警告投资者不要盲目相信巨头的神话： > “仅仅因为你假设这些企业拥有出色的营运现金流，就认为它们能在这场 AI 退潮中毫发无损，我认为这是一个非常不切实际的设想。” 当年次贷危机中，也有精明的交易员看准了次级抵押贷款的崩盘，选择做空权益层（Equity tranches）并做多最高等级的 AAA 评级证券，然而在全面的流动性风暴中，两者最终都被一网打尽。当前的 AI 供应链生态同样存在高度交织的连带风险。 ### [54:39](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3279s) AI因子的不可预测性与高收益债利差预警 Alex Altmann 指出，当前没有任何量化模型能够准确映射或捕捉“AI 因子”。他的团队反复尝试建立因子模型，但参数与贝塔系数变化太快，导致整个市场本质上只被简化为“做多 AI”或“做空 AI”。超大规模云厂商（Hyperscalers）显然全盘做多 AI，这意味着如果 AI 交易或 AI 因子出现长期平仓或去杠杆，整个市场“没有任何资产可以幸免”。 ```mermaid flowchart TD accTitle: AI交易因子的连锁风险 accDescr: 展示AI因子无法准确定价及其对超大规模云厂商和整体市场的潜在外溢影响 A["AI因子变动极快<br/>（Beta参数持续漂移）"] --> B["市场二元对立：<br/>做多AI vs 做空AI"] B --> C["超大规模云厂商<br/>全盘重度做多AI"] C --> D{"若AI因子发生<br/>持续平仓/去杠杆"} D -->|"系统性外溢"| E["没有任何资产能完全幸免"] ``` 即便如此，大盘与大型科技股迄今表现出了极强的韧性。Michael Batnick 引用 Daily Chartbook 的数据（图表 14）指出，目前评级为 B 级与 CCC 级的高收益信用债利差已接近 800 个基点（8%）。自 1996 年以来，该利差达到这一水平仅发生过 4 次，而在以往每次出现这种情况时，标普 500 指数都至少较其 52 周高点回撤了 12% 以上；然而当前标普 500 指数却依然处于历史高位附近。Alex Altmann 补充称，尽管超大规模科技股业绩表现极佳，但由于估值调整，市场的前向市盈率（Forward Multiple）已出现显著压缩，英伟达等核心标的估值甚至被压制在极低水平，这背后既有共同基金受限于 5% 单一持仓上限的“规模限制”，也有市场对整个 AI 生态系统相互依存度的担忧。 - **信贷利差历史极端值**：B 级与 CCC 级企业债利差达到近 8%，1996 年以来仅出现过 4 次，以往每次标普 500 指数均跌超 12%。 - **核心个股市盈率压缩**：部分硬件龙头（如英伟达）由于规模瓶颈与公募基金单一持股上限（多为 5%），前向估值受到压制。 --- ### [56:42](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3402s) 股票净发行量激增与回购萎缩的“点泡沫”既视感 关于美国股市的供需基本面，Josh Brown 强调美股“净发行量”（Net Issuance）正在发生颠覆性转变。过去投资者习惯了标普 100 指数成份股不论业绩好坏都大举回购股票的托底模式，如今这种“无脑回购托市”的格局已被打破。Alex Altmann 展示了团队数月前制作的美国股票净发行走势图（图表 16），该图涵盖了近 30 年的历史数据，并对 2026 年的供需进行预测。 ```mermaid flowchart LR accTitle: 股票净发行量激增驱动机制 accDescr: 资本开支挤压回购，叠加IPO与增发浪潮，导致股市供给端面临巨大抛压 subgraph 回购端收缩 A["巨额资本开支承诺"] --> B["回购执行率下滑<br/>（实际仅为授权的60%-70%）"] B --> C["回购总额停滞/萎缩"] end subgraph 供给端爆发 D["超大体量IPO<br/>（SpaceX / Anthropic 等）"] --> E["新股供给总额翻倍"] F["ATM市价增发 / 可转债"] --> E end C --> G{"供需格局逆转<br/>市场接盘承压"} E --> G ``` Alex 解释道，虽然市场经常宣扬企业批准了高达 1.2 万亿美元的回购计划，但“授权”与“实际执行”完全是两码事。企业实际执行率通常只有董事会授权规模的 60% 至 70%。鉴于当前众多科技巨头必须将现金流用于庞大的资本开支（Capex），其实际回购额很难进一步增长。与此同时，新股毛发行量（Gross Issuance）却几乎较上年翻倍，除了巨头的市价增发（ATM Offerings）和可转债外，超大规模 IPO 的集中涌现构成了巨大的增量供给。Alex 警告称，这一幕与 1999 至 2000 年互联网泡沫破裂前夕非常相似，当时并非仅仅因为负股权风险溢价（ERP），而是因为市场被毫无经济合理性的巨量新股发行所淹没： > “最终，如果你还记得当年真正击垮股市的是什么——我们经历过长时间的负股权风险溢价并维持了数年，但真正压垮市场的，是我们向市场上抛售了太多的新股票，而那些交易背后的商业逻辑根本站不住脚。”（Alex Altmann） - **回购执行率缺口**：企业实际完成的回购规模通常仅为董事会批准授权金额的 60% 至 70%。 - **供给端压力翻倍**：受可转债、市价增发及潜在超大规模 IPO（如 SpaceX、Anthropic）影响，美股新股毛发行量已接近上年同期的两倍。 --- ### [01:00:40](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3640s) 盘后科技巨头暴跌与 1 万亿美元的 AI 收入鸿沟 讨论进行期间，盘后公布财报的科技巨头集体遭遇抛售：英特尔（Intel）暴跌 25%，亚马逊（Amazon）重挫 12%，闪存芯片企业闪迪（SanDisk）跌超 9%，半导体板块整体走低约 4%。盘后行情印证了 Alex Altmann 几个月以来的防守立场。Alex 透露，巴克莱的量化风险模型“Betty”早在 5 月底就全面发出了减仓和规避风险的信号，成功避开了夏季的动量崩塌（Momentum Unwind）。 到了 8 月底，Alex 团队对基本面的评估变得更为悲观。他在拜访亚洲机构投资者后明确提出，科技巨头的资本开支即将见顶（预计明年达到 1.2 万亿美元，后年达 1.4 万亿美元），但产业界根本无法找到能够覆盖这笔支出的收入来源： > “我们根本不知道这些 AI 收入究竟要从哪里冒出来。到 2028 年底，整个行业基本上需要产生 1 万亿美元的收入，才能为眼下所有的资本开支买单……而按照目前的速度推算，届时的收入运行率大约只有 4000 亿美元左右。这中间的缺口我们根本补不上。”（Alex Altmann） Alex 进一步将这一期望与科技历史进行对比：标普 500 指数全体上市公司的年收入总计约 17 万亿美元，利润总额约 5.5 万亿美元。如果 AI 凭空变出 1 万亿美元收入，其体量将是不可思议的——苹果和微软这两家科技史上最伟大的企业，从 1990 年至今三十余年间经通胀调整后的累计总收入加起来也不过 9.8 万亿美元，而如今部分未上市公司（如 SpaceX）却声称其潜在市场规模（TAM）高达 30 万亿美元，暴露出市场对 AI 及科技商业化前景存在极度不切实际的幻觉。 - **盘后财报杀跌**：英特尔盘后暴跌 25%，亚马逊下挫 12%，闪迪跌 9%，半导体板块跌 4%。 - **AI 资本支出终值**：预计全行业资本开支将在 2025 年达到 1.2 万亿美元、2026 年达到 1.4 万亿美元的极限峰值。 - **万亿美元变现鸿沟**：为匹配资本开支，AI 到 2028 年需实现 1 万亿美元年收入，但当前行业预测的年化运行率（Run-rate）仅在 4000 亿美元左右。 - **历史收入参照**：苹果与微软自 1990 年以来的历史累计收入（经通胀调整）仅为 9.8 万亿美元。 ### [01:03:14](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3794s) AI 收入困局与多头被困：市场并不需要催化剂才会下跌 三人继续调侃所谓“数万亿”的 AI 市场规模预测，笑称如果算上冥王星、海王星甚至天王星的 GDP，这些天文数字或许才能合理化。随后，讨论切入严肃的核心议题：市场是否正在痛苦地意识到，根本没有足够的实际收入来支撑当前的资本开支估值？Alex Altmann 指出，这实际上是关于资本回报率（ROI）的反复拉锯。这种周期在 2024 年夏季就曾上演过——当时市场面临资本开支激增、日本央行（BOJ）加息以及疲软的非农就业数据（NFP），标普 500 指数瞬间回撤了 12% 到 14%；2025 年类似情况又与关税摩擦交织在一起。8 月底成为了一个明确的分水岭，促使 Alex 的团队因为收入兑现问题而坚定转向战术性看跌，并将这一逻辑与资本争夺战的宏观叙事相融合。 当被问及如果市场要向下调整、其“催化剂”会是什么时，Alex Altmann 直言当下根本不需要特定的催化剂，因为所有的负面要素早已聚齐：“在现阶段，我其实不需要一个催化剂；你所有的原料都已经摆在那里了。”（“At this point, I don't really need a catalyst, right? I'm kind of like, you've got all the ingredients. They're already there.”）融资端的裂痕正在自发演变为催化剂，而各大巨头的资本支出承诺已成定局，最早也要等到次年 1 月公布全年业绩时才有修正的可能。与此同时，Barclays 的量化模型发出了强烈的预警信号：实际利率高企，现货与波动率（spot-vol）的正相关性居高不下，波动率控制基金（vol-control）与量化投资策略（QIS）极度超配股票风险，散户资金也重仓多头。Downtown Josh Brown 借此总结道，多头目前已经彻底“被困”，除非市场预期降低或者价格完成重估。 ```mermaid flowchart TD accTitle: AI 市场压力与重估机制 accDescr: 展示 AI 资本开支过高、收入不足与量化多头拥挤如何引发市场价格重估的逻辑链条。 subgraph MacroPressure["宏观与资金面压力"] A["巨额资本开支 (CapEx)<br/>已锁定至明年 1 月"] B["实际利率高企<br/>融资成本上升"] C["量化与散户重仓<br/>(Vol-Control/QIS/Retail)"] end subgraph ValuationGap["基本面脱节"] D["实际 AI 业务收入兑现缓慢"] E["多头资金被困 (Bulls Trapped)"] end subgraph Resolution["市场出清路径"] F["风险出清价格下移<br/>(回撤至 200 日均线)"] G["估值消化重估<br/>(NTM P/E 降至 17.5 倍)"] H["开启新一轮健康创新与融资周期"] end A --> D B --> E C --> E D --> E E --> F F --> G G --> H ``` - 2024 年夏季市场曾因资本开支引发的 ROI 恐慌、日本央行行动及弱势 NFP 导致标普 500 指数下跌 12% 至 14%。 - 巨头们的资本开支数字具有极强刚性，在次年 1 月财报季之前几乎不可能主动调降。 - 量化模型警示：实际利率居高不下，现货与隐含波动率出现异常同向波动，系统化策略与散户群体的权益风险敞口处于极端饱和状态。 ### [01:05:21](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3921s) 重估的目标位、科技早期类比与长线视角 对于如何才能重新转为看多，Alex Altmann 表示最健康的方式莫过于让风险资产的出清价格实质性下行。2026 年因伊朗局势引发的最大回撤仅约 8.5%，这对于挤出泡沫来说远远不够。Alex 心理预期的技术与估值支撑位在标普 500 指数的 7,100 至 7,200 点之间，即 200 日移动平均线附近；该点位对应未来 12 个月预期市盈率（Forward NTM P/E）降至约 17.5 倍。唯有经过这种全盘重估并释放大量风险后，市场才能重新蓄力，开启下一轮芯片融资与技术创新浪潮。 Alex 借用初代 iPhone 的历史作出生动类比：当年 iPhone 面世时，最令人兴奋的爆款应用居然是用陀螺仪模拟“喝啤酒”的趣味软件，但这显然不是智能手机的巅峰形态。同理，人们今天根本无法想象几年后的“巅峰 AI”到底是什么样。Downtown Josh Brown 赞同这一观点，指出当前的看空只是战术性的仓位调整，而非全盘否定技术未来；随着时间推移到下一年度一季度，当生物科技公司宣布借助 AI 让三期临床试验提速 30%，或者实体机器人开始走入工作场所时，新的兴奋点又会涌现。但在当下，科技股财报公布后股价普遍遭遇抛售，市场对“大手笔资本开支”的故事已经彻底脱敏、产生审美疲劳，价格重估势在必行。 - 标普 500 指数潜在健康调整目标位：7,100 - 7,200 点区间（贴近 200 日均线）。 - 估值回撤目标：未来 12 个月预期市盈率降至 17.5 倍左右。 - 历史对照：初代 iPhone 上的“喝啤酒”应用证明了技术初期应用形态的局限性，不能将当前的收入瓶颈等同于 AI 潜力的终局。 ### [01:07:08](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=4028s) 亚洲路演的“皇帝新衣”与资本圈的群体心理 关于市场何时开始对资本开支神话产生动摇，Alex Altmann 透露，最先敲响警钟的其实是亚洲的机构投资界。9 月初他在中国香港和新加坡进行路演时，带着自己对 AI 收入缺口的审慎思考与各家机构交流。几乎在每一场会议的开头，当地投资者都会激烈反驳，坚称“半导体股票市盈率很便宜”、“技术前景无懈可击”。然而，当 Alex 把逻辑层层拆解剖析完之后，戏剧性的一幕发生了——90% 的人最终都会私下承认：“其实我们同意你的看法。” 这一现象揭示了市场深处的羊群心理：表面上人人都试图表现出积极乐观的态度，但内心深处都怀有同样的隐忧。Michael Batnick 指出，这就像硅谷西海岸的社交法则一样，没有人愿意成为那个公开唱衰 AI 的“怪人”或“失败者”。因为过去三年里，任何一个在任何阶段做空或抵触 AI 的人，要么显得像个小丑，要么直接丢掉了饭碗。Downtown Josh Brown 借此为 Alex 的观点定性：“也就是说，由于持仓结构的原因，你在战术上保持看空。”（“tactically bearish due to positioning.”）Alex 完全认同这一界定，强调自己并非长期的死空头，只是在当前拥挤的市场结构下保持高度审慎。 - 亚洲投资者在路演中展现了矛盾心理：口头上力挺半导体估值，但深入讨论后 90% 的机构私下认可 AI 变现困难的逻辑。 - 过去三年做空或怀疑 AI 的基金经理普遍面临巨大的职业风险甚至失业压力，导致全市场形成了不敢公开看空的从众壁垒。 ### [01:08:53](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=4133s) “Alty”外号的由来与尾声闲聊 在节目的尾声，录音室的气氛转向轻松幽默的私人闲聊。Downtown Josh Brown 问起为什么大家都叫 Alex 为“Alty”。Alex Altmann 解释称，这个外号始于他在伦敦读高中时。那所学校非常传统讲究，班里同时有五个 Alex、两个 Alistair、一个 Alec 和一个 Ali。由于重名太多，师生之间全凭姓氏称呼，整天“Altmann、Altmann”地喊，听起来就像随时被校长叫去办公室训话一样令人不适。2007 年刚步入职场时，为了避免继续被呼来喝去地叫姓氏，也为了与公司里其他叫 Alex 的同事区分开，他便主动让团队叫自己“Alty”，这个外号随后在业内沿用至今。 听完故事，Downtown Josh Brown 调侃说自己想到了更绝的外号，比如基于其量化策略背景叫他“Liquid Alt”（流动性另类投资），或者干脆叫“Ctrl+Alt+Delete”（控制键+删除键）。随后，主持团队向观众介绍了获取 Barclays 机构研究的渠道，并与 Alex 约定未来财报季与重估周期演进时再度做客节目深聊。 ### [01:10:55](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=4255s) 节目收尾与致谢 在节目接近尾声之际，主持人们意犹未尽地与嘉宾 Alex Altmann 开起了玩笑，甚至当场发出继续录制的邀请，打趣问道：“Are you busy tomorrow?” 对此，现场大家附和着表示干脆明天再来录，甚至直接提议：“Oh, yeah. Come back. Let's do it. Let's make it weekly.” 这种轻松幽默的互动不仅展现了对本期内容深度的认可，也体现了现场极其融洽的气氛。 随后，主持人正式向本期嘉宾表达感谢，并致谢了幕后支持团队：“Guys, Alex Altmann, shout out to Barclays and the BETS team. Much appreciated.” 节目在对观众与听众的常规道谢中圆满落幕，全场在掌声中结束了本期录制。 - 嘉宾与团队致谢：特别感谢嘉宾 Alex Altmann，以及来自 Barclays 及其 BETS（Barclays Equity Thematic Strategy）团队的支持。 - 节目结语：“Thanks for listening. Thanks for watching and we'll see you soon.” ## Notable quotes > [28:33](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1713s) The old adage was stock market is not the economy. No, it is the economy now. — Downtown Josh Brown > [24:35](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=1475s) 401(k) 资金就像天气一样确定，只要大家还有工作，它就会源源不断地涌进来。 — Alex Altmann > [46:53](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=2813s) 数据中心债务图表正在告诉你，你无法忽视它，因为利差正在走阔，而且它们将继续走阔，直到有效地测试或击垮这个市场。 — Alex Altmann > [50:00](https://www.youtube.com/watch?v=E5CTwZKN_OI&t=3000s) 仅仅因为你假设这些企业拥有出色的营运现金流，就认为它们能在这场 AI 退潮中毫发无损，我认为这是一个非常不切实际的设想。 — Alex Altmann ## Mentioned Alex Altmann, Downtown Josh Brown, Michael Batnick, Barclays, Millennium Management, Citi, Citadel, Société Générale, JPMorgan, Charles Schwab, Fidelity, Robinhood, Meta, NVIDIA, Microsoft, Apple, Amazon, Intel, SanDisk, ServiceNow, Spotify, Netflix, SpaceX, Anthropic, CoreWeave, Jensen Huang, Federal Reserve, S&P 500 ## Video description On episode 266 of The Compound and Friends, ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Downtown Josh Brown⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ and Michael Batnick are joined by ⁠Alex Altmann⁠ to discuss: what tactical trading really means, why hedge funds track each other's positioning, the odds of making money in the S&P 500, whether Wall Street is actually bearish on AI, why retail investors keep buying the dip, record household stock ownership versus housing, 401(k) flows and the relentless bid, whether the stock market is now the economy, pod shops and negatively correlated stocks, ServiceNow's earnings crash, Microsoft's $25 billion bond deal, the AI financing squeeze in data center bonds, CoreWeave and the neoclouds, shrinking buybacks and the IPO wave, Intel and Amazon's after-hours plunge, and the $1 trillion AI revenue gap. This episode is sponsored by Grayscale. To learn more, visit ⁠https://www.grayscale.com/ ►00:00 - Cold Open ►03:30 - Meet Alex Altmann of Barclays ►09:32 - S&P 500 Odds and the Barclays Timing Model ►16:51 - Is Wall Street Really Bearish on AI? ►19:17 - Retail Investors: Dumb Money or Smart Money? ►23:04 - Households All-In on Stocks, Not Homes ►28:33 - Is the Stock Market Now the Economy? ►32:36 - Pod Shops, Crowding and Negative Correlation ►40:55 - ServiceNow and AI Earnings Season ►44:12 - Microsoft's $25B Bond and Equity Issuance ►46:53 - AI Financing and Data Center Bond Spreads ►51:56 - CoreWeave, Neoclouds and the Jensen Put ►56:42 - Buybacks Shrink as the IPO Wave Hits ►1:00:39 - Intel, Amazon and the $1T AI Revenue Gap 👕 Check out The Compound shop: https://www.idontshop.com 🎙️Listen to our podcasts: TCAF: https://podcasts.thecompoundnews.com/show/TCAF/ Animal Spirits: https://podcasts.thecompoundnews.com/show/animalspirits/ Ask The Compound: https://podcasts.thecompoundnews.com/show/ATC/ The Compound Newsletter: https://www.thecompoundnews.com/subscribe Linkedin: https://www.linkedin.com/company/the-compound-media/ Instagram: https://instagram.com/thecompoundnews Twitter: https://twitter.com/thecompoundnews Tik Tok: https://www.tiktok.com/@thecompoundnews Investing involves the risk of loss. This podcast is for informational purposes only and should not be or regarded as personalized investment advice or relied upon for investment decisions. Josh Brown and Michael Batnick are employees of Ritholtz Wealth Management and may maintain positions in the securities discussed in this video. All opinions expressed by them are solely their own opinion and do not reflect the opinion of Ritholtz Wealth Management. See our disclosures here: https://ritholtzwealth.com/podcast-youtube-disclosures/ The Compound Media, Incorporated, an affiliate of Ritholtz Wealth Management, receives payment from various entities for advertisements in affiliated podcasts, blogs and emails. Inclusion of such advertisements does not constitute or imply endorsement, sponsorship or recommendation thereof, or any affiliation therewith, by the Content Creator or by Ritholtz Wealth Management or any of its employees. For additional advertisement disclaimers see here https://ritholtzwealth.com/advertising-disclaimers. #thecompound #investing #stockmarket #stocks [Source](https://www.youtube.com/watch?v=E5CTwZKN_OI)

[28:33] Downtown Josh Brown: The old adage was stock market is not the economy. No, it is the economy now. [24:35] Alex Altmann: 401(k) 资金就像天气一样确定，只要大家还有工作，它就会源源不断地涌进来。 [46:53] Alex Altmann: 数据中心债务图表正在告诉你，你无法忽视它，因为利差正在走阔，而且它们将继续走阔，直到有效地测试或击垮这个市场。 [50:00] Alex Altmann: 仅仅因为你假设这些企业拥有出色的营运现金流，就认为它们能在这场 AI 退潮中毫发无损，我认为这是一个非常不切实际的设想。

[6:40] Alex Altmann: Positioning can be instructive because obviously flow tends to dictate price, right, and price dictates narrative. [24:33] Downtown Josh Brown: There's more guaranteed money into the same 500 stocks every time people allocate to their wealth accounts. [65:34] Alex Altmann: I don't think we've yet even discovered or even imagined what peak AI is going to look like years from now. [63:14] Josh Brown: The bulls might be trapped at least for the time being until expectations change or prices change. [46:53] Alex Altmann: Spreads are widening and they're going to continue widening to the point where it's effectively going to either test or break the market and we're seeing cracks already.

[0:00] I sort of always like the east side [0:01] better than the west side. Just [0:03] personally, [0:03] >> it depends what for. See, I think the [0:06] west side doesn't skip a beat on the [0:08] water, right? Cuz you got [0:09] >> Oh, like Hudson Yards, [0:11] >> but just the whole strip, right? So, you [0:13] obviously have Battery Park, Tribeca, [0:15] then you got Soho, Meat Packing, [0:18] >> West Village. [0:19] >> We're not going to go that far from the [0:20] train station, [0:21] >> but you know what I mean. Like, and then [0:22] soon and you got Chelsea, Hudson Yards, [0:24] even Hell's Kitchen's kind of fun uh on [0:27] the, you know, the right environment. [0:29] And then you've got Upper West, Lincoln [0:31] Square, and then Upper West, which is [0:32] where I am. And like the whole stretch [0:34] is just beautiful. [0:35] >> I love the Upper West side. [0:36] >> See, I mean that's the only I lived in [0:38] Tribeca that when I was single. [0:40] >> Is this recording? [0:40] >> Yeah. So, also Alice, [0:42] >> it is okay. [0:43] >> When we when we start when the show [0:45] starts properly, you could drop the [0:46] Halloween British accent and talk [0:47] normally. [0:48] >> Yeah, exactly. [0:50] >> Uh you like on Riverside or [0:51] >> Yeah, I'm on Riverside Boulevard. [0:53] >> One of my favorite neighborhoods. [0:54] >> Oh, yeah. I love it. [0:55] >> You know it. um my be my best friend [0:57] after college I so I was upper east he [1:00] was all the way there so it's a pain in [1:02] the ass to go see him but like [1:04] >> I love that neighborhood [1:05] >> it's changed completely so they've they [1:07] they built the park all the way south to [1:09] 59th Street now all on on the water so [1:12] >> you can run all the way down [1:13] >> all the way down and you've got [1:14] obviously the [1:16] >> the [laughter] [1:18] boat basin [1:20] >> oh it's up on 90 something [1:22] >> okay cuz that's a whole thing also in [1:24] the summers Yeah. Yeah. I've got a [1:25] friend who owns uh one of those. [1:27] >> Have you ever been there? [1:28] >> No. [1:28] >> It's like a restaurant boat thing. [1:31] >> I I've I think the Upper West Side's uh [1:33] beautiful. I also I like Hudson Yards. I [1:36] would work there. [1:37] >> Hudson, but the but the office is [1:39] expenses there. It's like [1:40] >> it's $250 a square foot to touch [1:43] anything on a decently high floor. [1:45] >> Yeah. Forget it. [1:46] >> I don't It doesn't make sense for us. [1:48] We're not in the We're not in the two [1:49] and 20 world. So, um I [1:51] >> You and me both, mate. [1:52] >> Yeah. So, I I think we'll we'll figure [1:55] it out, but uh I'm delaying that project [1:57] as long as possible because it's it's a [2:00] big decision. You got to sign 10-year [2:01] leases in Manhattan. [2:03] >> Listen, like a lot could happen in the [2:04] next 18 months. We'll talk a bit about [2:06] it on this on this pod, I guess. But [2:08] like, [2:08] >> yes, [2:09] >> you know, [2:10] >> you you could see I mean, it's very [2:11] plausible that we see a recession before [2:13] the next election and you might actually [2:15] get a decent opportunity to scoop some [2:17] commercial real estate at like [2:19] attractive prices. A bunch of businesses [2:21] go pop. Oh, the whole city, God willing. [2:24] >> Lit literally [laughter] anything could [2:26] happen. [2:26] >> Anything for a good deal. All right, [2:28] let's go. [2:29] >> Yeah, boys. How we doing? [applause] [2:32] >> Ready? [2:34] >> All right. [2:34] >> It's the compound episode. [2:39] >> Curious about crypto and not sure where [2:41] to start? Start with Grayscale. [2:42] Grayscale is the world's largest [2:44] cryptofocused investment platform and [2:46] has been in crypto for over a decade. [2:49] That's a long time when you consider how [2:50] early we still are in crypto adoption. [2:53] In markets like these, education matters [2:55] more than ever. Gayscale provides [2:58] research, insights, and long-term [3:00] perspective to help investors better [3:02] understand digital assets and the [3:03] evolving crypto ecosystem. Whether [3:06] you're exploring crypto for the first [3:08] time or looking to deepen your [3:10] knowledge, let Gayscale be your guide. [3:19] >> [music] [3:24] >> Thank you, Thing One. We appreciate it. [3:27] Miss Nicole in Halloween regalia. Ladies [3:30] and gentlemen, welcome to the best [3:31] investing podcast in the world. [3:35] Four out of five dentists say the [3:36] compound and friends is where they get [3:38] all their financial news. We appreciate [3:41] that. [cheering] We have a new guest. [3:43] I'm so excited. I've been looking [3:44] forward [screaming] to this conversation [3:45] since we booked it. Uh I'm really pumped [3:48] for this. Our [music] guest today is [3:50] Alex Alman. People call you Alti, I've [3:54] been told. Is that a thing? [3:55] >> There's a story to that, but yeah, sure. [3:57] >> All right. You'll get into that in one [3:58] second. Alex is managing director and [4:01] global head of equities, tactical [4:04] strategies at Barclays, where he [4:06] generates thematic ideas and market [4:09] content for the bank's equities and [4:11] global markets clients. The Barclays [4:14] Equital Equity Tactical Strategies or [4:18] Bets team generates tactical ideas from [4:21] the trading desk and turns them into [4:23] research for clients. Before Barclays, [4:25] Alex was a macro portfolio manager at [4:27] Millennium and spent a decade running [4:30] equity trading strategy at City. So, [4:33] he's seen [music] the markets from both [4:34] the buy side and the sell side. Alex, [4:37] thank you for being here. We really [4:38] appreciate it. [4:39] >> Thank you for having me. [4:40] >> The crowd is going wild. Can you hear [4:41] that? It's unbelievable. It's an [4:43] exception. [4:46] uh for the benefit of the audience, how [4:48] do you define the term tactical and what [4:51] is different about the work you do at [4:53] Barclays versus other tactical trading [4:55] desks or research groups that refer to [4:58] themselves as tactical? [4:59] >> Sure. Okay. Well, again, thanks for [5:00] having me, guys. I mean, tactical is a [5:01] really easy answer. It just means it's [5:03] camouflaged and, you know, has has [5:05] weapons involved, right? [5:06] >> Many a lot of pockets. [5:07] >> Yeah, exactly. Just unnecessary pockets. [5:09] No, I mean like tactical in the in the [5:12] traditional sense of of of Wall Street [5:15] will mean that rather than thinking [5:17] about your typical research outfit that [5:19] will look at 6 12 even longer time [5:22] horizon 6 12 months uh we we'll we're [5:25] kind of trafficking in the shorter time [5:26] horizon more suited but not exclusively [5:29] to the fast money community. So we're [5:32] thinking one week, two week, four week [5:34] maybe you know eight weeks ahead not [5:37] really much beyond that. We have [5:38] thematic views. We've, you know, we have [5:40] views that have sort of been prevailing [5:41] for for months if not years in some [5:43] cases on some structural opinions. But [5:46] we're generally trying to think four to [5:47] eight weeks out and say, "All right, how [5:49] are we going to make money? [5:50] >> What what are markets are going to want [5:52] to do? Where's the best asymmetry?" And [5:54] then the other aspect about the tactical [5:56] consideration is we we traffic a lot in [5:59] verticals of the market that your [6:01] traditional research outfits tend to not [6:03] cover. So we'll think about positioning, [6:06] derivatives, quant, what are hedge funds [6:10] up to, you know, and and and these are [6:13] really important verticals or pillars of [6:15] the investment process that quite [6:18] frankly all of the buy side are doing, [6:20] but a lot of the sell side still kind of [6:22] don't cover from a traditional research [6:24] perspective. So we try to basically [6:26] match the desk investment process to [6:29] what the buy sides or the guys that we [6:31] typically talk to what they're doing and [6:33] that way we can have a pretty sort of [6:35] like Harry Pass head-to-head converge [6:38] funds want to know what other hedge [6:40] funds are up to. [6:41] >> Yeah. Okay. They do [6:42] >> and not purely like from a gossipy sense [6:45] like they want to probably but they want [6:48] to know. So positioning has become [6:50] something really important in the [6:52] market. Michael and I joke about it. [6:54] Anytime there's something going on in [6:55] the market and we can't explain it, we [6:57] go [6:57] >> positioning. [6:58] >> It's positioning [6:59] >> and it it is [7:00] >> but it's but people want to know people [7:02] that are managing money professionally [7:03] want to know what their peers are doing [7:05] and they I think they recognize the [7:07] limitations of like a traditional [7:09] portfolio manager survey which sort of [7:12] just comes out like group think. Um so [7:15] you so I guess you guys become a [7:17] repository for well here's what your [7:20] peers are here are the trades that [7:21] people are putting on and why that that [7:24] that's one aspect of it. [7:25] >> Yeah. [7:26] >> And then there's other aspects about [7:28] positioning. Okay. So positioning can be [7:30] instructive because obviously flow tends [7:33] to dictate price right and price [7:36] dictates narrative. So in theory you can [7:40] carve up and quite frankly I do follow [7:42] this mantra which is if you know if [7:44] there's a positioning asymmetry that has [7:46] been built in the market how does it [7:49] have the propensity to change price in a [7:51] very asymmetric way and in doing so can [7:54] shift the market's perception of a [7:56] narrative [7:56] >> and situational awareness was a perfect [7:58] example of that [7:59] >> perfect example of it right and you get [8:01] that periodically throughout any given [8:04] year you just need to have a framework [8:06] in place that allows you to kind of [8:08] monitor a lot of things simultaneously. [8:11] So are we monitoring single stock [8:12] positioning? I mean we can is the short [8:14] answer but we don't because generally if [8:17] you aggregate up to like theme sector [8:21] quant style [8:22] >> good enough [8:22] >> good enough and and of of course like [8:25] aggregate market. So you want to know [8:27] how are hedge funds positioned? How are [8:29] mutual funds positioned? How are CTAs [8:30] positioned? How are vault control and [8:32] QIS strategies positioned? How is risk [8:34] parity positioned? If you kind of get [8:37] that right. Oh, how's retail positioned? [8:39] That would be probably the biggest [8:40] >> is the idea. That's so that's an [8:42] opportunity to go the other way or not [8:45] necessarily bet betting against an over [8:48] an asymmetric positioning but maybe just [8:51] as like intelligence like where is the [8:53] money right now? [8:54] >> Correct. And so sometimes it will [8:56] actually provide an informed conclusion [8:57] about taking a a contrarian bet. Other [9:00] times that's not necessarily the case [9:02] and that applies pretty much [9:03] universally. So what we tend to do is [9:06] back test a lot of these positioning [9:08] reports that we run against what the [9:11] forward returns of whatever asset we're [9:14] interested in. And then in some cases it [9:16] can actually validate the view and [9:18] reinforce it. So i.e. this position is [9:20] crowded but actually based upon all of [9:22] our back testing the crowding means [9:24] nothing and actually the forward returns [9:26] are really good. And other times it will [9:28] say the opposite where you want to take [9:29] a contrarian bet and and sort of say [9:31] something controversial. I feel like [9:33] there is a school of thought that [9:35] short-term predictions are a total [9:38] fool's errand and I would say for most [9:40] for the casual investor of course it is [9:42] right trying to predict the market in [9:44] the short term for most people is a [9:45] total waste of their time do something [9:46] else [9:47] >> but the idea that you have clarity a [9:50] year out in today's investing world [9:53] sounds so insane given how fast things [9:56] are moving [9:57] >> that if you are going to make some sort [9:58] of market call tactical call is 0 to 30 [10:02] days probably is reasonable. [10:05] >> Yes. And so I'll give you a couple of [10:07] stats that can sort of help clarify that [10:09] a bit, right? So [10:12] if I told you that if you were to buy [10:14] the S, let's just use the S&P, it's [10:16] easy, right? So if you were to buy the [10:18] S&P on any given random day and you were [10:21] to hold that position for 42 trading [10:23] days, is equate equate to about 2 [10:25] months, the odds of you making money, [10:28] right, is about 73%. and your average [10:30] return would be around about just shy of [10:32] 200 basis points, 190 basis points. [10:35] That's pretty good. Right now, that's [10:38] better than better way better odds than [10:39] going to a casino and playing any game. [10:42] So, if you consider that your baseline, [10:45] you're you're effectively trying to [10:47] either outperform that or underperform [10:49] that. And I think that's where the [10:50] randomness comes in because if you're [10:53] going to just sort of buy the S&P and [10:54] hold it and just sort of take a kind of [10:56] a passive view on it, then you've you're [11:00] the odds are you're going to be right [11:02] more time than not. So really what [11:04] you're ever trying to do if you're going [11:05] to consider the tactical considerations [11:08] is you're trying to basically say, okay, [11:10] are the odds materially deviating from [11:13] that 73%. And we have a framework which [11:15] is actually pretty good. Is it perfect? [11:18] No. Because if it was, I wouldn't be [11:20] sitting here telling you about it or [11:22] talking to our customers about it. [11:23] >> You just run run it internally and keep [11:25] your mouth shut. [11:25] >> I keep my mouth shut from a from a [11:27] beach. [11:27] >> But you have the data. So the way that a [11:30] lot of gamblers lose, just a casual [11:32] gambler, there's uh whatever 20 games [11:35] over the weekend and they'll bet on 11 [11:37] of them and there's obviously no edge. [11:38] You don't have to swing at every pitch. [11:40] You don't have to bet every game. But [11:42] when there is reason, when the data [11:44] says, "Hey, uh whatever 80% of the time [11:47] do nothing, right? Right. Mostly noise, [11:49] but 20% of the time we have a 54% chance [11:53] of of making contact and we'll swing [11:55] when it makes sense to do so. [11:56] >> Yeah. And so in in our case obviously [11:58] we'll work on kind of the opposite where [12:00] we're generally looking for the 85s 80 [12:03] to 85 90s in some cases hit rates where [12:07] we'll look at kind of the forward return [12:09] profiles whether it's on the S&P whether [12:11] it's on a a particular style factor or [12:13] whether it's even on a sector and the [12:16] the hit rates look really attractive [12:18] based upon pretty robust not just back [12:21] tests but just very long sort of out of [12:24] sample and in sample periods. So good [12:26] case in point, right? So our main [12:28] framework that helps guide whether we [12:30] want to be bullish or bearish, we it's [12:32] called Betty. It's Barclay's equity [12:34] timing indicator. We were originally [12:35] going to call it betting, but then kind [12:37] of compliance had a bit of an issue with [12:39] that unsurprisingly. And it's a all it [12:41] is is that you say we have all the data. [12:44] Everyone has access to data. If I told [12:45] you what all of these inputs were, [12:47] >> you could build it yourself. It's [12:49] there's no secret source. It's just a [12:51] case of aggregating a bunch of stuff [12:53] that's already available in the market [12:55] and knowing that when a significant or a [12:58] sufficient number of them are flashing [12:59] either positively or negatively, you [13:02] know in your head and and on paper [13:05] statistically what the forward returns [13:07] of the S&P are more likely than not [13:08] going to do. We have the whole the whole [13:10] thing laid out and that's really our [13:12] guiding light with the from a [13:14] quantitative framework and then the [13:15] qualitative framework is the [13:17] positioning. It is talking to customers. [13:19] It's or it's knowing what the organized [13:22] narrative is in the market. And [13:24] hopefully, this is where it becomes more [13:25] of an art than a science. You've got the [13:27] science bit, right? The art bit is [13:29] pulling it all together. This is the bit [13:30] that AI hasn't figured out yet is like [13:32] pulling it all together in a way that [13:34] you can say, "Okay, all the ducks are in [13:36] a row. I think we got, you know, now we [13:39] to use your polls, we have to take a [13:40] swing." [13:41] >> Can you go back to what you said about [13:43] the order? How how you get to narrative? [13:47] It start like it starts with it starts [13:49] with what because I agree with the the [13:51] premise. Uh I just I sort of think like [13:55] um whatever Price is doing that becomes [13:58] the story that everybody accepts. It [14:01] just takes a little while for that to be [14:02] transmitted around the street. And then [14:04] it's it's funny like you hear you hear [14:07] people saying something on Thursday like [14:09] five people will say the same thing and [14:12] they arrive there because of whatever [14:14] Price just told them. It almost never [14:16] works the other way where somebody is [14:18] going against price and and people [14:21] accept that narrative. Does that make [14:23] sense? Like what I'm trying to ask you [14:25] >> it it it it does and and that's I would [14:28] mostly for for the most part I would [14:30] think I think of that as noise [14:32] >> is because people will use price to [14:35] justify [14:36] >> narrative or they'll explain something [14:40] expost because a price moved a certain [14:43] way rather than necessarily providing [14:46] real intelligence behind it and go [14:48] actually challenging the view. Does does [14:50] that make actual sense? Right. And so [14:54] again, that's not always the case, but [14:56] there's often times in the market where [14:59] things just move because maybe there's a [15:01] big print, someone had to put through a [15:04] certain trade and it was it was done in [15:05] a sloppy fashion, moves price, everyone [15:08] said, "Oh, what just happened?" That's [15:09] particularly uh prone like in the oil [15:11] market at the moment. You know, oil will [15:13] move around by a dollar or two and then [15:15] everyone jumps to conclusion that some [15:17] random headline from someone you've [15:19] never heard of. [15:19] >> That's the reason. [15:20] >> Yeah, that's the reason. And it's like, [15:22] okay, maybe there was an algo that [15:23] responds to a certain headline. That's [15:25] obviously the world we live in. But more [15:26] often than not, it's just noise. [15:28] >> Do you take every qualitative signal or [15:30] quantitative signal? And the reason I [15:32] ask is if you say it's just data and [15:34] everybody has the data, then there's got [15:36] to be something else. You need some sort [15:37] of I don't know, situational awareness [15:39] for lack of a better phrase here. Um, so [15:42] for example, the analytics on in [15:44] football might say go for and fourth and [15:45] one every time, but there could be well [15:47] it's like yeah, but we have the [15:48] momentum, our defense is strong, this [15:50] offense sucks. just punt it away and [15:51] live to see another day. So, how much of [15:53] the qualitative overlay dictates it? Or [15:55] do you just It's just 100% quantitative. [15:57] 82% chance we go no matter what. [15:59] >> No, no, no. The the qualitative aspect [16:01] is really important. And that that's [16:02] what I was referring to before. And [16:04] where there's art rather than just just [16:06] science. There are plenty of strategies [16:07] in the marketplace that are just [16:08] literally, you know, inputs and outputs. [16:11] And if it's good quality inputs, you get [16:13] good quality outputs and vice versa. But [16:15] I I'm a big believer that the [16:17] discretionary overlay is really [16:19] powerful. It's really important and you [16:21] know I kind of experienced that [16:22] firsthand when I was running money as [16:24] well in in so far that if you just rely [16:27] on a quantitive process and you inside [16:29] your own bubble I think it can work. [16:32] Clearly there's guys who have had [16:34] phenomenally successful careers running [16:35] those kind of strategies. But for me [16:37] personally, I believe there's real power [16:39] in in talking to the community and [16:42] getting a really good qualitative [16:43] assessment of what like the zeitgeist is [16:45] and then building the quantitive [16:47] framework alongside that to see if they [16:49] match up or if there's a discord, right? [16:51] And if there's some kind of dissonance [16:53] between the two, then that's to me is [16:55] usually where there's a real opportunity [16:57] because I'm a big believer personally [16:59] that and today in this environment is a [17:02] perfect example of that where you've got [17:04] this so-called prevailing narrative that [17:06] suddenly everyone is bearish on AI, [17:08] right? Because of what happened in the [17:10] financing space or or whatever it may [17:12] be. And you know, I conduct surveys. We [17:16] just sent out a survey just today uh [17:18] which we conducted yesterday. We had 200 [17:19] investors that came, which is a pretty [17:20] good sample size from from the Wall [17:22] Street community, and it's a high [17:24] quality survey, and the and the feedback [17:26] was overwhelmingly not bearish on on AI, [17:30] right? Only 60% of people that responded [17:32] thought that we had a financing problem. [17:34] I'm like, how can that be? We have a [17:35] financing problem because we've just [17:37] gone through one over the past two [17:38] weeks. So, that number should have been [17:39] 100%. [17:40] >> So, what does that tell you? [17:41] >> Well, it tells me so that that's the [17:43] start of a of a story, right? Because [17:45] you've got dissonance now. You've [17:46] already got okay customer feedback is [17:49] everyone's bearish AI survey says you're [17:52] wrong. Okay, let's check the positioning [17:54] data. So, traditional hedge fund PB [17:58] positioning data from some of my [18:00] competitors will say hedge fund nets are [18:02] super low, therefore everyone's bearish [18:04] and they've all sold tech. What they may [18:06] not necessarily tell you, and this is in [18:07] no way we no way me sort of like poo [18:10] pooing the the quality of work from [18:12] competitors, but what that data probably [18:14] doesn't show necessarily or people [18:15] overlook is that the range of net [18:18] exposure changes by hedge funds is [18:20] minimal. It will go to from 48% to 55% [18:24] back to 48%. Right. Yeah. [18:26] >> Why? Of course, because a lot of these [18:27] guys are RV pod shops and so they're [18:29] just running market neutral strategies. [18:31] So in reality that they have no or [18:34] little market sensitivity beyond their [18:36] beaters within the portfolio. So I don't [18:38] care what they're up to. And we again [18:40] we've run stats around this. So the [18:42] point being is is that we'll use the [18:43] positioning data where we think it's [18:45] valid. So what are retail positioned in? [18:47] Huge vertical. I mean the largest owner [18:49] of stocks is the retail community. Next [18:51] largest owner re mutual funds, right? [18:54] And levered ETFs. What are they up to? [18:56] Right? So these are the verticals. What [18:58] are V control funds are up to? Those [19:00] guys have significant equity length, QIS [19:02] strategies, ditto, CTA, ditto. So you [19:05] build up like a quantitive picture [19:06] across your framework and you say, okay, [19:09] what does the positioning metrics [19:10] actually look like? Now let's compare [19:13] that to what the prevailing narrative is [19:15] and now let's compare it to what our [19:17] customers are saying. [19:18] >> What do you see in retail? Is it was it [19:19] is it dumb money as people say or have [19:21] they been I feel like they've been [19:22] pretty damn right the past this this [19:24] cycle with buying dips. [19:25] >> Vast majority of retail would portfolios [19:27] would have outperformed institutional [19:28] portfolios. What does that tell you? [19:31] >> Tells us a few things. First and [19:32] foremost is they have more holding [19:34] power, right? So, a lot of the hedge [19:37] fund community are operating on much [19:39] tighter draw down limits because their [19:41] mandate on you know their [19:43] >> they're being observed. [19:45] >> Correct. [19:45] >> The retail account is not being observed [19:47] by anyone other than the holder and the [19:48] holder's wife. [19:49] >> Yes. And and and and [19:52] uh and it's something they want to brag [19:54] about to their their friends at at a [19:55] dinner party. [19:56] >> I bought the debt. [19:56] >> I bought the deb. But they're not [19:57] answering to a committee about why [19:59] they're uh overweight semiconductors. [20:02] They're just overweight semiconductors [20:03] cuz they want to be [20:04] >> and why they draw down 400 basis points. [20:06] >> Right. [20:06] >> Right. And the only thing that will [20:08] basically cause them to stop out is [20:10] either their brokerage account basically [20:12] margin calls them or they lose their [20:14] job. That's it. [20:16] >> Yeah. [20:16] >> Right. It really is all about people [20:18] always ask me, "Oh, what gets retail to [20:19] sell?" I'm like, "Lose your job [20:21] >> or when your neighbor loses their job." [20:22] So there's been there's been times where [20:24] retail has dumped stocks like obviously. [20:27] Is that is that a buying opportunity or [20:29] not necessarily? [20:30] >> It depends on the starting point and how [20:32] long it's been going on for. [20:34] >> So what if I told you we've got data on [20:35] retail back to around about 1995 like [20:38] good quality data data and what we do. [20:39] >> Is that relevant still in today's world? [20:41] >> Absolutely. [20:42] >> It's more relevant today than I think [20:43] it's ever been. [20:44] >> Wow, that's surprising. Why do you say [20:45] that? [20:46] >> Well, because you know the devices Okay, [20:50] I guess we have to kind of go back into [20:51] history a little bit. [20:53] 2019 was a real demarcation for retail [20:56] trading. Not 2020, 2019. People don't [20:59] really talk about it. And that's because [21:00] in October of that year, it's when zero [21:03] commission trading was introduced, [21:05] right? And it just so happened to [21:06] coincide with the launch of Robin Hood [21:09] and some of the, you know, the more [21:10] gamified. [21:11] >> This is when Schwab said zero, Fidelity [21:13] said zero, and zero became the standard. [21:15] >> Zero. Exactly. Yeah. Right. So there was [21:18] very, very low friction trading. Then [21:20] what happened? COVID, right? And then [21:23] what happened? [21:23] >> Everyone trades. [21:24] >> Bunch of stimulus checks, people sitting [21:25] at home. [21:26] >> 40 million new brokerage accounts open [21:27] in 18 months. [21:29] >> Exactly. So, but really, I mean, would [21:32] nonzero zero commission trading have [21:34] changed it? Probably not. But I do think [21:37] it's important because, you know, it [21:39] effectively lowered your transaction [21:40] friction to to to nothing. And then you [21:42] got a bunch of people with stimulus [21:43] checks and nothing to do and they're [21:45] basically, you know, let's face it, [21:46] we're all degenerate gamblers. So, a lot [21:49] of people basically got involved in that [21:52] and and that obviously coincided with a [21:54] significant short-term correction in in [21:56] equities. [21:57] >> They didn't leave. [21:59] >> They never left after the bare market. [22:01] They're still here. [22:01] >> They're still here. Oh, no. No. They [22:02] they that's not true. They did. So, [22:04] there was a degree of selling by retail [22:07] community in 2022. And that was a [22:09] deliberate financial tightening exercise [22:11] by the Fed, right? [22:12] >> But it was but it was shortlived. Like, [22:13] if you look at the percentage of overall [22:14] volume, it's still pretty it's still [22:16] high. Is it 25%? [22:18] >> The retail share of trading is still [22:20] where it was. [22:20] >> Oh, in terms of the share. Yeah. But [22:22] what we tend to track is the evolution [22:24] of both cash balances and margin [22:26] balances as a percentage versus the [22:28] natural drift of [22:29] >> retail got blown out of spaxs and all [22:31] that [ __ ] just like everybody else did. [22:33] >> Nvidia felt 75% and so did Meta in 2022. [22:36] I mean, that was terrible. It was yeah [22:37] it was it was a bad time for retail and [22:40] and retail genuinely lost money but they [22:43] didn't they did sell but they didn't it [22:45] wasn't [22:46] >> they didn't they didn't completely get [22:48] out of dodge. Yeah. And then ever since [22:50] then 23 24 25 26 regardless of whether [22:53] we've had those economic wobbles or not [22:55] they've just kind of been pretty sick. [22:57] >> So so how is the data from the '9s [22:59] relevant to today? [23:01] >> So the the in terms of the comparable [23:04] >> Yeah. [23:04] >> Yeah. So, we've blown through everything [23:05] we saw in the '90s. The ownership of of [23:09] the US household ownership of stocks [23:11] today, you've got approximately 2/3 of [23:14] all American homes that own stocks. [23:17] That's way higher than [23:19] the other stat which I I sort of love to [23:22] throw around is that the the next [23:24] highest the 35% [23:27] Yeah. 35% of US household net worth is [23:30] now in equities, the highest ever. The [23:32] next highest wealth channel is [23:34] unsurprisingly um home equity. But the [23:38] gap between [23:40] stocks and home equity in terms of your [23:42] wealth percentage for closing [23:44] >> is the widest it's ever been in history [23:46] going back to the 1960s. [23:49] We've only [23:50] >> But isn't it obvious why like homes have [23:52] appreciated 0% over three years thanks [23:54] to interest rates? [23:56] >> The stock market has doubled. Yep. [23:58] >> Of like of course of course [24:00] >> it's not causal. It's it's the it's the [24:02] reverse. [24:03] >> It it's there's a whole bunch of other [24:05] stuff. I mean, think about over the past [24:06] really since 2022, mortgages have been [24:08] completely unaffordable. You know, new [24:10] homes and existing home sales have [24:11] plummeted. The housing market's [24:13] basically been dead or, you know, [24:15] zombie. [24:15] >> It's a housing market recession. [24:17] >> It's for all intents and purposes with [24:18] housing really not doing much, right, in [24:20] terms of prices. [24:22] >> Um, so you've got a bunch of, you know, [24:24] Gen Y, Gen Zers with money that they [24:28] savings that they've accumulated. Where [24:29] are they going to put it? They're not [24:30] putting it into a new home anymore. [24:32] They're putting it into the stock [24:32] market. [24:33] >> Yeah. [24:33] >> Right. So, you've had that constant [24:35] inflow that we estimate if you think [24:37] about where like typical new mortgage [24:39] origination from home equity down [24:41] payments $2 to $300 billion a year. [24:43] Plus, you got the 401k money that comes [24:46] in every year. That's between8 and $900 [24:48] billion on our estimation. Right? So, [24:50] between that, you've got over a [24:51] trillion. Then you got a trillion [24:53] dollars of buybacks. Even if you haircut [24:55] that a little bit for the executions [24:57] being low, you're still talking about [24:58] 750 billion. guaranteed. This is guar [25:01] almost guaranteed buying like 401k [25:04] money. [25:04] >> 401k is guaranteed [25:05] >> like the weather like it's coming in. [25:07] >> It's coming in as long as people have [25:08] jobs [25:09] >> and it's valuation insensitive and it's [25:11] just it's it's automated. [25:13] >> Yeah. Because it's effectively like, you [25:14] know, life plan, you know, retire [25:17] target, life target, whatever it is, you [25:19] know, 65. So it sold. [25:20] >> You had that at 900 billion a year. [25:22] >> Yeah. [25:23] >> Okay. So the way I worked it out was uh [25:26] number of uh homes in America um and [25:30] then we sort of haircut that number to [25:32] assume that not every home can [25:33] contribute to a 401k whether you for [25:35] whatever reason then we assume 12% of [25:37] your cont 12% contribution uh of your [25:40] paycheck. Um and then and your and your [25:42] median household income is $88,000. [25:44] >> Yeah. [25:45] >> So we kind of you sort of solve for that [25:47] maths and you heck out a little bit [25:48] more. You get to between 8 and 900 [25:49] billion. [25:50] >> Yeah. So, I wrote I wrote something [25:52] about that 11 or 12 years ago on my [25:55] original blog. And I called it the [25:57] relentless bid and I basically said [25:59] there's like three things driving this [26:02] cuz the question back then in 2014 the [26:05] market had just made an all-time high in [26:07] 13 and there was still a lot of bad news [26:09] and euro zone stuff and people I don't [26:12] understand why does this market ever [26:13] sell off, right? Like it it just crashed [26:17] spectacularly. doesn't every time [26:19] there's bad news it wobbles and then it [26:21] goes right back to highs and people were [26:23] struggling with that and I basically [26:25] said it's the combination of buybacks [26:27] shrinking the float wasn't even at a [26:29] trillion a year yet was on the way um [26:32] this 401k money which I felt had been [26:34] underappreciated and then a third factor [26:37] that I don't think anybody was talking [26:38] about the business model transition from [26:41] stock brokers to financial planners and [26:45] wealth management transitioning from my [26:47] broker trades stocks for me to my broker [26:50] allocates the ETFs and uh that was [26:53] trillions of dollars. Like you look at [26:55] Morgan Stanley going from predominantly [26:57] trading commissions to predominantly [26:59] feebased. What does that do to the the [27:01] way the market behaves? I argued it's a [27:04] volatility dampener. There's more [27:06] guaranteed money into the same 500 [27:09] stocks every time people allocate to [27:11] their wealth accounts. So, we've been [27:13] living in that world now for over a [27:15] decade. [27:16] >> Oh, yeah. [27:16] >> Yeah. [clears throat] And I think it's [27:18] like [27:19] >> I think it's a big reason for why the [27:22] VIX so quickly evaporates every time it [27:24] gets going. Like it just you've got all [27:27] right, two weeks go by, the 401k money [27:30] comes in. Um the financial adviserss who [27:33] used to trade stocks, now they're just [27:34] rebalancing ETFs. You can almost set [27:36] your watch by it. [27:37] >> Yeah. And then there now on top of that [27:39] you've got the lever you've got the the [27:40] the cool overwriting ETFs. you've got [27:42] kind of very deliberate velling products [27:44] that are further suppressing or or [27:46] supplying gamma and Vega not to use [27:49] these Greek terms but like uh you know [27:51] supplying volatility to the market as [27:53] well on top of that but you know to [27:55] answer the original question you know [27:56] retail are kind of allin on stocks part [28:01] of it involuntary because they've kind [28:03] of lost faith in the home the home [28:06] ownership slice and also just the [28:09] fractional nature of stocks is very [28:10] appealing to younger individuals, right? [28:12] I don't need to cough up, you know, if [28:14] your median home price in the US, I [28:16] don't know the number off top of my [28:17] head, but I'm guessing it's between [28:18] three and $400,000. You know, you're not [28:20] have to cough up $60,000, $50,000 for a [28:23] deposit. I can go and buy a share of [28:26] Nvidia or a share of the SM of SPY or [28:28] whatever and just forget about it. It [28:30] unfortunately does create [28:31] vulnerabilities in our economy, you [28:33] know, and people ask me all the time, [28:35] >> and I know we're getting a bit [28:36] tangential here, but like what keeps you [28:37] awake at night? Is it like led ETFs or [28:40] is it Iran or is it AI? I'm like, no, [28:42] it's retail. It's retail is the scary [28:45] one in the sense that because we've got [28:47] such a financialized economy, right? [28:50] Like you even last year during the [28:53] tariffs when the S&P fell 20%. There was [28:55] a visible drop in consumption over the [28:58] subsequent two months. [28:59] >> Negative wealth effect. [29:00] >> Huge. And because it's real time, [29:02] unlike.com where [29:03] >> that also scared people. There's massive [29:04] uncertainty in terms of what the economy [29:06] was going to do. I don't think it was [29:07] just the stock market falling. [29:08] >> You're right. It wasn't just the stock [29:10] market falling, but the average [29:11] individual really had no concept about [29:13] what the implication of tariffs was [29:15] going to be in that moment. Right. [29:16] >> They were scared. [29:17] >> I think they were scared. And but more [29:19] importantly, in 2000 when we had even a [29:22] it wasn't comparable today, but it was [29:24] reasonably comparable in terms of the [29:26] amount of equity ownership in [29:28] households. You had to wait either for a [29:30] monthly or quarterly brokerage report, [29:31] right? Most people did, you know, they [29:32] were looking [29:33] >> getting alerts into your hand sometimes [29:35] a day. you are seeing your real time [29:37] wealth adjustment, right? And that's [29:39] very powerful on both the way up and the [29:41] way down. You know, I kind of had this [29:43] running joke [29:44] >> that the stock market is the economy [29:45] now. The old adage was stock market is [29:47] not the economy. No, it is the economy [29:48] now. [29:48] >> I agree with that. [29:49] >> I say it all the time and people get [29:51] pissed off about it, but I really think [29:52] it's more than ever. [29:53] >> I agree. [29:54] >> The centricity [29:55] >> of of the stock market. It used to be [29:59] like an avatar, loosely speaking, for [30:01] how companies are doing. And then people [30:04] would say, "All right, but what does IBM [30:07] and Coca-Cola have to do with Main [30:08] Street?" Like, now I think it's [30:11] basically um we've built retirement [30:14] around the stock market. Everybody's all [30:17] in. Every household with a high income [30:19] also has a large 401k. Now we've tied [30:22] compensation to the stock market. You [30:24] think about all the young upand [30:26] cominging millionaire households. They [30:29] all work for companies that are about to [30:31] go public, just went public, have stock [30:33] prices, like um this is it. If you can't [30:37] have a prolonged 2-year bare market [30:40] without a recession, it's impossible [30:43] because everybody's wealth effect and [30:45] everyone's mentality is based on how [30:47] much money their accounts are worth. And [30:49] so, I totally agree with you. [30:51] >> Yeah. And the messed up thing is is that [30:52] I I would challenge any economist on the [30:54] planet, no matter how good they are, and [30:56] there are some phenomenal economists I [30:57] work with at Barclays, but I don't think [30:59] any of them actually have a cell in [31:00] their econometric model that measures [31:02] S&P. [31:03] >> Yeah. [31:04] >> Right. So, they're they're monitoring [31:05] everything. [31:06] >> Who does? The Fed does. [31:07] >> Well, the Fed have their FCI process. [31:09] >> The Fed is looking at the the 10-month [31:11] moving average, looking at the stock [31:14] market. I'm I think they know when [31:16] they've overstepped. I watched that play [31:18] out in 2018. It was the most obvious [31:20] thing in the world. It come out It come [31:22] out early December. We're nowhere near [31:25] uh neutral. [31:26] >> And then on Christmas Eve of all times [31:29] of year, we're in a 21% [31:32] statistical bare market. Like just past [31:34] that 20% threshold. That's doesn't [31:37] really mean anything. But in the media, [31:39] bare market, they reversed themselves on [31:42] Christmas Eve. [31:43] >> Yeah, I remember. And a month later, not [31:46] only are we perfectly near neutral, [31:48] actually we're going to start cutting [31:49] rates again. That's literally what [31:52] happened. How could it not be the the [31:54] 10-month moving average or the or the [31:56] 200 day or whatever. [31:57] >> You know what's amazing about that [31:58] period in 2018? [31:59] >> Yeah, [31:59] >> it was also a midterm year [32:01] >> and it was and it was also a new Fed [32:02] share [32:02] >> and Minutuchin was very chatty. [32:04] >> Remember that he came out out of nowhere [32:06] and said, "Don't worry, every there's [32:08] ample liquidity." And everybody was [32:09] like, "What the?" In the great in the [32:11] great financial crisis um 10 years [32:14] before there was one Bloomberg terminal [32:17] at the Fed in Washington. [32:18] >> Yeah. [32:19] >> There's one person in there looking at [32:20] the markets. You want to bet there's a [32:22] thousand now. [32:23] >> Oh no, but I I know that the both the [32:25] Treasury Department have Bloomberg [32:27] terminals because they they ping me and [32:29] then so so does the New York Fed as [32:31] well. [32:31] >> Well, there's a hedge fund manager [32:32] running it. So I'm just saying like [32:34] everything is the stock market. [32:35] >> Yeah. [32:36] >> Can I show you a stock market chart? I [32:37] want to get your thoughts on this. So [32:38] this has been an interesting year. um [32:40] friendly and different reasons. And even [32:41] though there is the relentless bid um AI [32:45] has caused a lot of onoff days, same [32:48] with software, same with energy, there [32:51] are a lot of days where the stock market [32:52] does X and the average stock goes the [32:55] other way. So this is weird negative [32:56] beta effect. Um, Sockgen has a great [32:59] chart, chart 15, Daniel, that shows [33:03] the they say there are 124,000 750 [33:07] possible pairs of companies with the [33:09] within the S&P and at present 40% are [33:12] negatively correlated. So they show the [33:16] average stock correlation within the S&P [33:18] 500 and they compare that today. So the [33:21] the red line is the average and the [33:23] black line is today. Mhm. [33:24] >> 40% of days we've seen stocks going [33:28] different directions in the overall [33:29] index. It is a very funky time within [33:32] the stock market. [33:33] >> Yeah. So, a lot less like uh 2017 every [33:37] stock went up together. There was no [33:39] volatility. Zero. [33:40] >> Matter of fact, today's a perfect [33:41] example. So, the last couple of weeks [33:43] we've been talking about the same thing. [33:44] Breath sucks. Breath sucks. Breath [33:45] sucks. And we've been saying, "All [33:47] right, but that's a feature of a bull [33:48] market. You want the biggest stocks [33:50] doing the heavy lifting usually when [33:53] breath is leading. Not usually. Um the [33:55] average stock does better in a bare [33:57] market. Just think about 2000 is a [33:58] perfect example of this right after the [33:59] do bubble bust. So today you have the [34:02] equal weight up 98 basis points. The S&P [34:04] is down 53. [34:06] >> And it seems to be that this is [34:07] happening more with more and more and [34:09] more frequency. There's a bad uh MAG7 [34:11] report Google today for example and the [34:13] rest of the market catches a bid. [34:15] >> Yep. And you're you're asking [34:18] >> just just we're just talking. [34:20] >> We're just talking. Okay. So, I can [34:21] actually give you there's a there's a [34:23] pretty good answer for why this is [34:25] happening. [34:26] >> Oh, yeah. I don't think it's a mystery. [34:27] >> Uh oh. I I'm curious. [34:28] >> I don't know the answer. What's What's [34:30] the What's the reason? [34:31] >> Well, I think earlier in the year it was [34:33] energy. It was a lot of energy stocks [34:36] going up when the market went down and [34:37] that was this weird negative beta [34:39] effect. Yeah. And then it was interest [34:40] rates, right? interest rates went up, [34:42] consumer staples got killed, re got [34:43] killed, utilities got killed, Max 7 was, [34:45] especially Apple completely undeterred, [34:47] and Apple's biggest stock in the world. [34:49] And then it was the software uh AI days [34:52] on, off, on, off, on, off. I think it's [34:54] it's been those three things. What else [34:55] am I missing? [34:56] >> So there there's there's like a unified [34:58] uh narrative that that basically puts [35:00] all those together. Um and it goes back [35:02] to our friends at uh at RV Strategies, [35:05] relative value, right? So if you think [35:08] about um multi-manager AUM in 2018 2019 [35:13] uh before co was around about $300 [35:15] billion and today it's like a trillion. [35:18] >> Yeah. [35:19] >> Right. So it's basically tripled since [35:20] co [35:21] >> and [35:21] >> so who are you talking about like [35:22] Citadel and and Millennium and those [35:24] type of companies? won't name names [35:26] right but but uh you know multi-managers [35:28] pods you know the basically guys where [35:30] effectively their job is to isolate [35:32] idiosyncratic alpha as opposed to be [35:34] taking large directional bets right so [35:36] they'll encourage their managers to run [35:38] high gross low net exposure and it makes [35:41] sense right why why has there been [35:43] explosion at aum since co well because [35:45] well really since 2022 when interest [35:47] rates went up and it's because in a high [35:48] interest rate environment RV strategies [35:50] are absolute no-brainers right because I [35:53] deploy a lot of growth on my long side [35:54] load of gross on my short side. I [35:56] actually haven't really deployed a [35:57] dollar unit, so to speak. I've got [35:59] plenty of risk, but I take my actual [36:01] capital haircut for collateral for my PB [36:04] requirements. [36:05] >> And then you're earning a SP uh you're [36:06] earning an interest rate on the rest. [36:07] >> Correct. Right. So, whatever your [36:09] collateral commitments are, let's say [36:10] that I 15% aside or whatever, but you're [36:12] taking the 70% and you're earning that [36:14] you're earning that 5%. You probably [36:16] even get more than that in relatively [36:17] safe. [36:18] >> That's the low net part. you're [36:19] offsetting with your longs and your [36:20] shorts and then you're collecting uh an [36:23] interest rate on whatever you don't have [36:25] to allocate. [36:25] >> Correct. And and because these [36:27] businesses are run so they're run so [36:29] efficiently, that's why you're given [36:31] very strict mandates on what your net [36:33] exposures can be, right? Because they [36:35] it's not you breaching some horrible [36:37] limit. It's actually just because as a [36:39] as an organization, they want to sort of [36:42] >> manage effectively the there's a [36:44] treasury function essentially. Yeah. [36:45] >> And it's a very good business model. [36:47] It's, you know, for for and all you're [36:49] doing is you're trying to isolate that [36:50] idiosyncratic alpha. Now, here's the [36:52] catch. If everyone's doing it, right? [36:56] It's not just a handful of funds, [36:58] >> right? [36:59] >> Everyone's doing it. Everyone has access [37:00] to the same risk parameter metrics. So, [37:02] they're neutralizing their value and [37:03] their quality and their size and their [37:05] growth and all the rest of it. They're [37:07] also obviously neutralizing their sector [37:08] exposures and they're neutralizing their [37:10] market exposure. But the number of [37:12] stocks you can trade has shrunk. We [37:14] talked about this deequitization [37:15] already, right? So um [37:16] >> have like 3,000 stocks basically [37:18] >> right and that number has gone down over [37:20] time not up right it's [37:21] >> started in the business with like 10,000 [37:23] >> and it's net negative issuance in the [37:25] sense that we're having few we've had [37:26] fewer IPOs and more buybacks so by [37:28] definition there's less available stock [37:29] to trade right so with those combination [37:32] of factors plus the ballooning of aum [37:34] plus leverage from PBS on top you end up [37:36] with this crowding out effect and [37:38] everyone will say oh okay hang on a [37:40] second it doesn't matter because I've [37:41] neutralized everything no all you've [37:42] done is push the crowding into other [37:44] aspects of your portfolio that you are [37:47] either aware of or unaware of, right? [37:49] And this really matters because when [37:52] you've ballooned the size of the number [37:54] of strategies, so to speak, you're [37:56] creating more of these vicious or the [37:59] propensity for these vicious unwinds. [38:01] >> So every everyone unwinds one aspect of [38:03] a portfolio at once, it hits the stocks [38:05] three times harder than it otherwise [38:07] would have. [38:08] >> Correct. which is why you're seeing very [38:10] little index vol and an absolute ton of [38:13] factor V single stock V negative [38:16] correlation [38:17] >> and you see the stocks that are like [38:19] obviously Spotify and Netflix are [38:21] relatedish [38:22] but for a year it was the exact same [38:25] chart like identical and there's a [38:28] million different stocks like that that [38:29] trade that just trade they're just in [38:31] the same basket of stocks that people [38:32] are buying and selling [38:33] >> correct and and by the way a lot of [38:34] those people are actually just computers [38:36] right so you got obviously very sific [38:38] sophisticated quant funds that are [38:39] running sort of real-time stat [38:40] strategies that are looking for the one [38:42] day or even intraday you know price [38:43] differential they can play mean [38:45] reversion so there's those playing the [38:46] super short-term stuff you've got the [38:48] discretionary alpha guys which are [38:50] looking for time horizons IM which is [38:52] called let's call it 3 to six week and [38:54] that's generally on single stock pairs [38:55] as well for me inversion around earnings [38:57] that's why on earnings day the majority [38:59] of of a single stock flow is coming from [39:01] discretionary and then on non-earnings [39:03] day the vast majority is coming from [39:05] quant and retail [39:06] >> and then in the media people are looking [39:07] at the preponderance of all of this data [39:10] and they're coming up with the story [39:12] that explains it to non-market [39:14] participants who don't appreciate the [39:17] degree to which this is computers [39:18] trading with computers. Yeah. [39:20] >> And it sort of I taught I taught my my [39:24] daughter my daughter had a a unit I [39:26] remember in elementary school. They were [39:28] learning about Greek mythology. So she's [39:30] like sitting reading this book. She's [39:32] like she can't make head heads. Why are [39:34] they giving us fairy tales to read? I [39:36] said you don't understand. This was [39:37] science. Like why does the sun rise in [39:40] the east and set in the west? Oh. Uh, [39:44] it's because Apollo on his chariot is [39:46] being pulled across the sky and he's [39:49] glowing and so that's the sun. Uh, why [39:52] is there an earthquake? The gods are [39:54] mad. You did this. Why is there a [39:55] volcano? Like this is what humans do. [39:59] They need even if it's absurd. [40:01] >> They have to explain it. So in the [40:03] media, [40:04] >> they're looking at the net result of all [40:06] this statarb and everything that's [40:08] happening. That's not a satisfying [40:10] explanation. Why did the market change [40:13] its mind about AI today? Is yes and [40:16] because and then they find a headline [40:18] that explains why it changed its mind. [40:20] >> I mean, this is the shameless plug to to [40:22] to like the team, you know, like the the [40:25] whole point of why we exist is to kind [40:28] of we the Yeah, the bets team at [40:30] Barlays. Like [40:31] >> how many of you are there [40:32] >> globally now? We're 11. [40:34] >> Okay. Uh, so we've got three in New [40:37] York, we've got three in um in London, [40:41] and we've just hired additionally in [40:42] Asia. So we've now got five. [40:43] >> There's 11 people on the bets team. [40:45] >> So for now, yeah. [40:46] >> Now, is it true you guys are allowed to [40:48] kill? [40:49] >> We we are we are kill only. Yeah, [40:52] capture. Yeah, kill. Yeah, we're not [40:54] we're not in the capture business. [40:55] >> Okay. What do you do about earning [40:56] season? So like for example, today [40:58] Service Now is down 27%. Is there like a [41:00] blackout period where you're like, "All [41:02] right, we're just what we do on the [41:04] tactical side does not work when there [41:06] are these major announcements." [41:09] >> It's a really good question. So, um, [41:11] this is why maybe one day bets will be [41:13] 12 or 13 because we want to get more [41:14] involved in a single stock business. I [41:17] mean, Barclays has phenomenal spec sales [41:20] that cover specific, you know, sectors [41:21] and so on and so forth. That's that's [41:22] their business and we partner with them. [41:24] But from a kind of macro thematic [41:27] perspective, we tend to stay out of the [41:29] way. So for for me, earning season is an [41:31] environment where [41:33] >> it's landmines [41:34] >> there. That was the word I was trying. [41:36] >> There's almost no [41:36] >> tactical. [41:37] >> There's no structure. There's no [41:38] structure to it until you put one on it. [41:41] Like you can't the traditional way of [41:42] thinking about how stocks are trading. [41:45] Then a company comes out and a random [41:47] com comment in a conference call could [41:51] like take a stock from being up 10% to [41:53] down 10%. [41:54] >> Totally. Look, you can model that. the [41:55] the only the only time where we really [41:57] get involved in any season is the fact [41:59] that uh equities I mean to the benefit [42:02] of my career equities have become more [42:04] macro right AI is a macro theme right [42:07] hyperscalers are no longer just [42:09] institutional securities for the you [42:11] know for moms and pops they are they are [42:13] they have eyeballs on them from every [42:15] single corner of the investment [42:17] community whether you're a credit guy [42:18] macro guy retail whatever right so the [42:22] those businesses during earning season I [42:25] I will scrutinize just as much as as the [42:27] next man. I'll listen to the conference [42:29] calls. You know, we will we will, you [42:31] know, put them through natural language [42:32] processing. We'll talk to obviously our [42:33] spec sales and we'll talk to research. [42:35] We'll talk to all the great teams that [42:36] we can within Barclays to sort of be as [42:38] informed as an equity investor. We are [42:40] equity investors, but be informed at a [42:42] micro level. Luckily for me, I started [42:45] my career as a single stock analyst in [42:48] 2004 at JP Morgan. So like my DNA is [42:51] kind of there already and I pivoted to [42:53] macro in 2012. So I've got kind of like [42:57] a bit of that legacy um muscle memory to [43:00] be able to go sort of go into that [43:01] during early season if we need to. And [43:04] we kind of need to now because you know [43:05] the Google Google results were all about [43:08] that. Microsoft results were all about [43:09] that. We're obviously going to get I [43:10] think we're getting Amazon tonight, [43:12] right? So you know these do matter and [43:14] and whilst AI is a theme and you know we [43:17] can talk about that whether it's going [43:18] to be a theme for the future. driving [43:20] like twothirds of the economic growth [43:21] are coming from some capex related [43:25] spending that has to do with AI. [43:27] >> You know, you want to know wild stat? [43:29] >> Yeah. [43:29] >> Yeah. So, next year, next year 2027, [43:32] consensus S&P earnings, [43:35] 75% of that is expected to come from [43:38] semiconductors and hyperscalers. [43:40] >> Great. Love it. [43:41] >> Right. [43:41] >> God, let's hope there's not a financing [43:43] problem uh uh problem that extends into [43:46] next year. [43:46] >> Like, it's it's wild. I mean I to I mean [43:49] we kind of joke around it. It's like the [43:51] S&P's just become the Cosby, right? [43:53] >> Yeah. [43:54] >> Yeah. Almost. [43:54] >> Meta is spending 70% of their top line [43:57] on cap X of their topline of theiring [44:01] revenue just on cap X. It's [44:03] unbelievable. Dan, you could throw up [44:05] chart 8. So I'm I'm curious to get your [44:07] take, Alex, on how much the equity [44:09] issuance is impacting the stock market. [44:12] So we're looking at Microsoft cloud [44:14] revenue growth. [44:16] I think he needs like 10 minutes to [44:18] digest everything. So, here's what's [44:20] going on. So, the chart on the right, [44:22] the chart on the right is year-over-year [44:24] revenue growth. And as you can see, it's [44:26] accelerating. Okay? [44:28] >> Obviously, because of the AI trade, [44:29] cloud is is the beneficiary. So, it's [44:32] growing at 30%. It hasn't grown at this [44:34] rapid of a rate since the third quarter [44:37] of 2022. [44:38] >> Mhm. [clears throat] [44:38] >> At that time, it was doing $23 billion [44:42] in revenue. The fact that it is [44:44] accelerating when it is literally almost [44:46] three times the size is unmodelable, [44:50] unfathomable, unpredictable, [44:52] unpredictable, [44:52] >> unprecedented. And yet today, the $25 [44:56] billion debt offering and the stock is [44:58] down. [44:59] >> I mean, they they could not have done [45:01] they could not be doing any better. [45:02] Could not have had a better quarter. [45:04] Stock is down. [45:04] >> Yeah. [45:06] So, shouldn't we just be asking the [45:08] question [45:10] of if you're spending that much money [45:13] and right now, yeah, you you've got [45:14] phenomenal revenue growth, what's it [45:17] going to come at the cost of, right? [45:20] Last time I checked, I think Microsoft's [45:22] still still operating cash flow positive [45:23] right now. [45:24] >> It is. I think it's the only one. [45:25] >> Also, the stock had a monster run. All [45:27] right. So, like it it's near it's at an [45:28] all-time high basically. Um, but there [45:32] the the bar is high. Well, ju just the [45:34] shock that we were in a 15-year period [45:37] of [45:38] um massive buybacks and now that's gone [45:41] into reverse. Now we're looking at stock [45:44] issuance from not every company but from [45:46] these gigantic companies that are super [45:48] meaningful to the index. [45:50] >> Yeah. Yeah. It's a different [45:51] environment. [45:52] >> We we talked about so we yeah we did a [45:53] lot of work around this and it was it's [45:55] okay. So you you can't just look at [45:58] equities in isolation. That's the first [46:00] thing right. So you're right. So 2026 on [46:02] our modeling numbers will be a positive [46:04] net issuance year. [46:05] >> Yeah. [46:05] >> Which is basically going to be from what [46:08] we can tell essentially the first time [46:10] >> barely. [46:10] >> Okay. [46:11] >> But goes back to that discussion earlier [46:13] when if you're if you typically were [46:15] retiring net negative issuance of four [46:17] five $600 billion a year and now you've [46:20] had a multiund billion swing. [46:24] >> It's different. [46:24] >> It's different. Right. [46:26] >> And then the second thing is the [46:28] competition for capital. Right? because [46:31] 2026 net IG and high yield issuance is [46:35] going to be $1.2 trillion, right? And [46:39] then obviously the government needs to [46:40] go cap in hand as well, [46:41] >> but it's all being absorbed still. [46:43] >> It is for the most part, [46:44] >> but but risk premier has widened as a [46:46] function of it, right? [46:47] >> Let's let's put some of his charts up [46:49] and uh let's do chart 11, Daniel. [46:54] >> So So tell us what's going on here. [46:56] Okay, so this is just simply looking at [46:59] the uh it's just an equal weighted index [47:03] of the spread of a bunch of data center [47:06] bonds [47:07] against uh the equivalent [47:10] uh benchmark which I I think the average [47:13] average duration is about 17 years. So, [47:15] we just took the 20-y year 20-y year um [47:17] uh Treasury and you know, you can see [47:20] that that's sort of steadily widened and [47:23] now it's sort of not steadily widening [47:26] and like so yes, you're right. It is [47:28] being absorbed but at a price, right? [47:31] And so the prevailing market narrative, [47:33] right, is at the moment we talk about [47:35] narratives is that there is no price [47:38] that this AI trade doesn't make sense. [47:41] Nvidia will tell you that. Meta to a [47:43] degree will tell you that. Microsoft [47:45] probably tell you that based upon their [47:46] revenue growth numbers, right? But the [47:48] reality is a little bit different. There [47:50] is a price, right? And I think what the [47:52] market is doing right now is trying to [47:54] get to the point where that where that [47:57] price is. It's trying to work out is [47:59] there a price where you guys won't want [48:01] to pay. [48:02] >> Put up a chart 12. [48:05] >> So you are saying [48:06] >> this is my favorite chart. the first [48:08] time there has been real competition for [48:10] capital in a higher interest rate [48:12] environment. What are we looking at? [48:13] >> Right? So, this goes back to the point [48:15] that you're not you're not just issuing [48:16] in a vacuum in equities. You're issuing [48:19] at a time when the government's trying [48:22] to get money and they're running a [48:24] significant deficit. The debt market [48:26] whether it's hyperscala issuance or [48:28] otherwise just the corporate financing [48:30] needs are issuing as I said $1.2 [48:32] trillion of net issuance. [48:34] >> That's the competition. [48:34] >> Yes. Right. [48:35] >> And then equities is now you've [48:37] completely offset or mitigated the [48:38] buyback effect, right? And so you take [48:41] each market in any given year and you [48:44] work out how much issuance was as a [48:46] percentage of that market size and then [48:48] you get a percent obviously you add up [48:50] those three percentage numbers and then [48:52] you basically put it on that as a dot [48:54] and you compare it to where two-year [48:55] yields were in that year. [48:57] >> So why is this such an anomaly this [48:59] particular year? [49:00] >> Good question. So effectively this chart [49:02] is just telling you that you know [49:04] corporate finance and econ 101 holds [49:06] true over a very long period of time and [49:09] this goes back to I think 1995 if I'm [49:11] not mistaken [49:12] >> meaning higher cost capital less [49:13] issuance [49:14] >> correct kind of makes sense but [49:15] apparently [clears throat] [49:16] >> not today [49:17] >> we've just we've decided and this is [49:18] like the AI narrative AI does AI can [49:20] ignore that [49:21] >> right and the the chart before the the [49:25] you know the IG uh chart of of data [49:28] centers debt is telling you you can't [49:30] ignore because spreads are widening and [49:32] they're going to continue widening to [49:34] the point where it's effectively going [49:35] to either test or break the market and [49:39] we're we're we're seeing like cracks [49:41] already. That's what coreweave is about. [49:43] >> Yeah. [49:44] >> Right. That's what Neocloud's more [49:45] holistically about. It's about the fact [49:47] that actually AI financing does have [49:50] limitations. It may not apply today to [49:52] hyperscalers who still are cash [49:54] volcanoes, but it is certainly applying [49:56] to you know levered [49:58] sort of plays on forward compute. [50:00] >> Well, so the market efficiently is uh I [50:04] guess like the way to think the way I [50:06] would think about this is it's [50:08] rationality in the market. [50:10] >> Yes. [50:10] >> It got to coreweave before it got to [50:12] Alphabet which is the way it should [50:15] work. I mean okay give me uh 13 guys. [50:18] What um you're saying equities are also [50:22] finally catching up to the moves in [50:24] credit by way of a bid to quality and [50:28] this is your balance sheet efficiency. [50:31] What like I guess explain to the viewers [50:34] what they're looking at and why this is [50:35] meaningful to you. [50:36] >> Yeah, sure. So look, obviously what what [50:38] we try to do as a team is try to take [50:42] relatively complex concepts in the [50:45] market that may or be produced by other [50:47] parts of the bank or by us and present [50:50] them in ways that kind of make sense. [50:53] Right. [50:53] >> Right. [snorts] [50:54] >> And so equity markets are amazingly [50:58] efficient below the surface. Right. A [51:00] lot of people just look at the S&P and [51:01] go it just goes up or you know whatever. [51:04] And yeah, and as we know recently [51:05] there's these tons of rotations and [51:07] those rotations, some of them make [51:08] sense, some of them may not. But this is [51:10] a classic example of like one of the [51:12] visualizations of that rotation, right? [51:14] And it's hard because this is a slightly [51:16] longer time series, but essentially what [51:19] it's telling you the orange line is the [51:20] behavior of [51:22] >> those are stocks. [51:23] >> Those are this is a basket of stocks on [51:25] the long side versus a basket of stocks [51:27] on the short side. So it's an RV trade [51:29] just like the triple C versus double P [51:31] spread is also. So this is intuitive. If [51:33] you need to borrow money, money is [51:35] expensive. You're in much bigger trouble [51:36] than the companies that have a stronger [51:38] balance sheet. [51:38] >> Exactly. It's it's it's actually a very [51:40] straightforward concept, which is the [51:41] market is is efficiently starting to [51:44] basically tell you that if you're a [51:46] lower quality balance sheet business, [51:48] yeah, the equity is going to start [51:50] getting hurt because the chances are you [51:52] might need to tap that equity channel to [51:54] start raising money because the debt [51:55] channel is not available to you. Is [51:56] there a world where the neo clouds that [51:58] seem to be uh challenged to continue [52:01] raising money at the rate that they [52:02] have, is there a world where they flame [52:04] out and this never ends up touching the [52:07] the traditional hyperscaler stocks and [52:09] they sort of just sail past it? Is there [52:12] a world in which this is actually a new [52:15] uh comparative advantage for the [52:17] traditional hyperscalers and it actually [52:19] helps their stocks for there to be chaos [52:22] amongst uh the nebuses of the world? [52:25] Like what? How do you think about that? [52:27] >> So, you know what I call that? I call [52:28] that the trade after the trade. [52:30] >> Yeah. So, [52:31] >> yeah, I call it that too. But let's [52:32] explain to Michael what what [laughter] [52:33] we mean [52:35] >> that in in the sense that [52:38] >> trade after the trade like just like I I [52:40] mean I could explain it, but I want to [52:42] hear [52:42] >> Cory's only $33 billion market cap. When [52:44] is like could Nvidia if it goes low [52:46] enough would Nvidia just buy it and just [52:47] put this to bed? So if I listen I don't [52:50] know exactly what Jensen's thinking in [52:51] his head other obviously last week he [52:53] told basically in in corporate [52:54] parliament you know told uh corweave to [52:58] kind of go pound sand right he basically [53:00] said I have no interest in helping you [53:02] guys out at all why would he is trading [53:04] above the IPO price so so that that move [53:07] lower needs to be a lot lower before I [53:09] think he'd even consider uh from an [53:11] equity perspective but why would you [53:12] even do that I mean [53:13] >> who doesn't want the equity he wants the [53:14] assets so you buy the bonds [53:16] >> but is there a Jensen [clears throat] [53:17] put like if things get bad enough he [53:18] could just say one thing and and chill [53:20] everybody out. Maybe it's much lower, [53:22] but [53:22] >> my suspicion is given how [53:26] central in N invidia have become to the [53:29] entire ecosystem and we've all seen some [53:30] version of that chart with the [53:33] >> sun. Yeah. [53:34] >> Yeah. It's like your Greek mythology [53:36] analogy [53:37] >> and um so I I don't want to sort of [53:40] speak and say yes there would definitely [53:42] be a point where where Jensen would do [53:44] something like that but it would be [53:45] logical that he would if there was a [53:48] proper shakeout that someone like him [53:50] who's already got vested interest would [53:51] would look to sort of hoover up assets [53:54] but also you got to remember that could [53:56] come at the expense of his own share [53:57] price too right so everyone assumes that [54:00] the hyperscalers are completely immune [54:02] from this and the way I think about it [54:04] is and it's not the right analogy but [54:07] it's the only one I can think of is [54:08] subprime and gibs right right the just [54:12] because you assume that these businesses [54:15] have got fantastic operating cash flow [54:17] that they're going to be completely [54:18] immune from any AI unwind and I think [54:21] that's a very unrealistic proposition [54:23] >> they went they almost they basically [54:24] went to zero [54:26] >> you talking about like Fanny Freddy [54:28] >> right but again I don't want to draw too [54:29] too strong a parallel to the GFC I just [54:31] mean in terms of you know there were [54:33] guys who like who really saw the [54:34] subprime thing coming and they're like [54:35] I'm going to short the equity tanch I'm [54:37] going to go long the AAA's and they both [54:39] went to zero right I'm not saying that [54:40] hypers scale is going to zero what I am [54:42] saying is is that to that there is the [54:45] AI is touching everything like we've [54:47] talked about factors in this in this in [54:49] this conversation there is no way to [54:51] correctly map the AI factor from what I [54:53] am aware I've tried my team has tried [54:56] and we can't do it it keeps moving too [54:58] quickly the beaters keep changing right [55:00] so you're either long AI or you're short [55:01] AI goes with that saying the hypers [55:03] scale is a long AI. So if there is a [55:05] prolonged or protracted unwind of the AI [55:08] trade or the AI factor, [55:09] >> nothing is immune. [55:10] >> Nothing's immune. [55:11] >> But until then of but of course but [55:14] immune is very strong of a word but they [55:16] have shown to be [55:17] >> resistant resilient whatever you want to [55:19] call it Daniel chart 14. So similar [55:21] concept as what you were just showing [55:23] the spread between uh this is from the [55:26] daily chart book. The spread between B [55:28] and triple C credit has reached nearly [55:30] 8%. something seen only four other times [55:33] since 1996. And on every other occasion, [55:35] the S&P 500 was already at least down at [55:38] least uh 12% below its 52- week high. [55:40] And of course, we're we're right near [55:42] all-time highs right now. It's just an [55:44] an unusual environment. And they have [55:46] proven to be resistant. [55:48] >> Yeah, they really have because they've [55:49] got this phenomenal earnings delivery [55:51] and you showed that Microsoft chart as [55:52] an illustration of that. [55:54] >> Helps it. It helps enormously. But what [55:56] we have seen is the market derated [55:57] spectacularly including those [55:59] hyperscalers themselves, right? And and [56:01] a lot of that is driven by memory [56:03] stocks. [56:03] >> When you say derated, do you mean on the [56:05] forward multiple? Do you mean on the [56:06] >> Yeah, on the forward multiple. So [56:07] >> Nvidia is 15 times. Nobody wants it. [56:10] >> I mean, I think there's also holding [56:12] constraints as it represents it's 8% or [56:14] 9% of the S&P now. So yeah, like a lot [56:17] of these mutual funds, they only have [56:18] they can only buy up to 5% of their [56:20] portfolio in it, [56:21] >> right? And [56:22] >> that's a size problem. [56:23] >> I think part of it is a size problem. [56:24] Part of it is also that sort of co [56:26] codependency in terms of the whole AI [56:28] ecosystem. There's that problem. Thank [56:30] god they're not trading at 40 times [56:31] forward earnings. It'd be $13 trillion. [56:33] It [laughter] [56:34] >> Well, the S&P would be at probably [56:36] 12,000. [56:37] So that that that would be a high [56:39] quality problem certainly from a [56:40] American wealth effect perspective. [56:42] >> I want I want to double I want to double [56:44] double click on the uh the net issuance [56:46] because I really think that that is [56:48] gamechanging just in terms of changing [56:50] the mentality of the market. I really [56:52] think that people just felt like no [56:54] matter how bad a company's results, S&P [56:57] 100 company, you know, they're buying [56:59] back stock and I can just ride it out. [57:01] And now that's no longer the case. Um, [57:05] chart 16, guys. You're showing us net US [57:08] equity issuance. Um, I I I guess you've [57:13] got uh you've got four colors here. So [57:16] give us like the story of the markets [57:18] because this this a lot of years of this [57:21] is 30 years of data and this is a really [57:23] big change now. Yeah. Versus the [57:25] environment that most of us have grown [57:27] up in. So tell us what you think is [57:30] what's going on here. What's the [57:31] implications? So we we did we produced [57:34] this chart a few [clears throat] months [57:36] ago. Um and in I'm I'm quite happy to [57:41] say that it's a lot of it was based upon [57:43] predicted or estimated for 2026 and for [57:45] the most part we've been pretty accurate [57:47] on this. So uh and the easy bit is IG. [57:50] We know really what gross issuance uh [57:53] sorry this is just in equity. So we know [57:55] broadly what the M&A slice is going to [57:56] be because a lot of most announced M&A [57:58] is sort of completing in the year. So [57:59] that that sort of is easy. [58:00] >> So the deals that were announced in 20 [58:03] the deals that are announced in the [58:04] first half of 26 [58:06] >> Yeah. [58:06] >> you you have a reasonable amount of [58:08] certainty like most of them will go [58:10] through and they'll happen in the second [58:11] half of 26. [58:12] >> Exactly. So how much stock is getting [58:14] issued? How much stock is being retired [58:15] from the M&A. Um and then and then [58:18] obviously from the actual the issuance [58:21] perspective we also have at the money [58:23] offerings from hyperscalers. We have [58:25] converts. we have, you know, any other [58:27] um additional equity raises that need to [58:29] be done. Um and then obviously we have [58:31] IPOs, right? So IPOs is probably the [58:34] bigger moving target that and the at the [58:36] money offering and obviously we had one [58:38] large IPO this year. [58:40] >> We'll have the two biggest of all time [58:41] inside of six months. [58:42] >> Uh well, I mean, who knows? [58:44] >> SpaceX and then anthropic if it if it [58:46] works. [58:46] >> SpaceX and Anthropic and obviously [58:48] OpenAI got punted into into 2027, we [58:51] think. So, but obviously this is just a [58:52] 26 forecast. So the idea being and the [58:55] one thing that we didn't solve for we we [58:57] solved for we guessed is that the [58:58] buybacks you'll see that the the the red [59:00] line or the red bar is basically flat [59:03] yearonear. So what we did was we [59:05] executions in buybacks are always going [59:07] to be less than announcements. Everyone [59:08] like loves to highlight oh $1.2 trillion [59:11] of announced buybacks. I'm like yeah but [59:12] if you [59:13] >> authorized but not completed [59:15] >> correct and and and the execution rate [59:17] is typically somewhere between 60 and [59:20] 70% of the authorization. they never [59:22] they never complete all of it. And so we [59:24] we basically said, "All right, we'll [59:26] keep it static year on year, which I [59:28] actually think is quite a generous [59:29] number because of, you know, let's face [59:31] it, all these businesses that are doing [59:33] buybacks historically are now doing [59:35] massive capex program. So I don't even [59:37] know if we're going to reach that kind [59:38] of buyback number, but give the benefit [59:41] of the doubt to say yes, they'll do [59:42] approximately the same number as last [59:44] year." You can clearly see that the [59:45] gross issuance has exploded. A lot of [59:47] that's because of IPOs. And [59:50] consequently, you've you've h you [59:51] haven't increased your buyback. [59:52] >> Blue bar is along the top. Exactly. It [59:55] is now [59:56] >> double the size of last yearish. Like [60:00] just eyeballing the issuance the new [60:02] stock. [60:02] >> Yep. [60:03] >> Is this keep I mean can it is there a [60:06] theoretical limit where it has to stop? [60:08] Well, there there isn't a theoretical [60:10] limit, but you if you sort of hawk back [60:12] to the the.com period, you know, you can [60:15] obviously see in 99200, you were dealing [60:17] with sort of similar explosion in the [60:19] numbers. I know they look small by [60:20] today's comparison, but at the time they [60:22] were massive. And ultimately, if you [60:24] remember what broke the equity market [60:25] back in, we had a prolonged period of of [60:28] negative equity risk premium, but that [60:31] had that persisted for years. It was the [60:33] fact that we were just bringing too much [60:34] paper to the market and the economics of [60:37] those deals just didn't make any sense. [60:39] >> Yeah. Right. [60:40] >> We got uh some interesting news in the [60:41] after hours. [60:42] >> Oh yeah. Amazon. [60:43] >> Intel down 25%. [60:46] >> Is that bullish? [60:47] >> Amazon, guys. Amazon down 12. [60:50] >> Is that is Intel the one the government [60:52] owns? [60:53] >> Sandis. [60:54] >> Don't worry. Sandis is down nine. [60:56] >> Oh, [60:57] >> Amazon. Amazon's down 12 [60:59] >> right right off the bat. Um semis were [61:01] down 4%. So, all right. Tomorrow's going [61:03] to be a day. [61:04] >> There we go. Did I Did I mention I'm [61:06] bearish [laughter] [61:07] >> positioning? Um, what what is your [61:10] outlook between now and the end of the [61:11] year? That that seems to be your time [61:12] frame. [61:13] >> Couldn't we have led with that because [61:14] now now I look like, oh, I'm bearish. [61:16] >> No, no, no. I read all your stuff. I [61:18] know that you've been I I wouldn't say [61:20] getting like more risk off, but most of [61:22] your notes are cautioning. [61:24] >> No, we know you've been wrong all year [61:25] until now. [61:28] What do what do you tell what do you [61:30] what is the message from the bets [61:32] research just about the environment [61:35] let's say between now and and year end [61:37] >> okay so um backdrop is I mentioned Betty [61:40] our sort of quantitive framework so that [61:41] basically had been signaling various [61:43] degrees of risk off ever since the end [61:44] of May so that we we were like mega [61:47] bullish through the Iran thing and then [61:49] end of May we're like okay this is this [61:51] is our model was just telling us to get [61:52] out like really to reduce risk and then [61:55] we had that big momentum unwind in [61:56] summer, right? At the end of August, we [61:58] we changed our fundamental assessment. [62:00] We were still cautious on the market, [62:02] but it was like, okay, we have no idea [62:04] where these AI revenues are going to [62:05] come from. You know, we need a trillion [62:07] dollars of revenues basically by the end [62:09] of 2028 to pay for all of this capex. [62:11] >> All the commitments in capex. [62:13] >> Yeah. And right now, our run rate for [62:15] then is going to be somewhere around [62:17] about 400, right? [62:19] >> Okay. [62:19] >> It's just we're we're not there. We need [62:21] to plug that gap. And so, you know, we [62:23] went out to Asia. We really just went to [62:25] the lion's den, spoke to a bunch of [62:26] investors out there, and we're like, [62:27] "Hey, this is our whole road map. This [62:29] is basically where the capex is." The [62:31] capex number can't go any higher. 1.2 [62:33] trillion for next year, 1.4 for the year [62:35] after. Like maybe it moves by 100 [62:37] billion here and there, but we're kind [62:38] of at terminal point now. Um I just [62:40] don't see where the revenue is going to [62:42] come from. S&P topline 17 trillion. It's [62:45] about $5.5 trillion of profits. and [62:47] we'll say all of a sudden we're going to [62:49] magic a trillion dollars. The cumulative [62:51] revenue of Apple and Microsoft, two of [62:53] the most successful tech companies in [62:55] history since 1990 to today, right, has [62:59] been $9.8 trillion inflation adjusted, [63:02] right? Not even 10 trillion. Meanwhile, [63:03] SpaceX is telling you that their TAM's [63:05] 30, right? So, of the the TAM of the [63:08] universe [63:08] >> in fairness, their TAM is every dollar [63:10] ever spent on human capital [63:12] >> out in the universe, right? It's not [63:14] just not just on Earth. So it's the GDP [63:17] of America. [63:18] >> They have to adjust for Pluto, Neptune. [63:20] >> That's right. [63:20] >> Yeah. [63:20] >> Uranus. [63:21] >> Okay. So, you know that joke, [63:23] >> so is the mark So is the So is the [63:25] market now going through the throws of [63:27] this realization that there's just not [63:30] enough revenue to make any of this make [63:33] sense until all the prices adjust? Is [63:35] that what that what we're living [63:37] through? I think we're living through a [63:39] a reassessment about on the ROI debate [63:41] which by the way has not it's not the [63:44] first time we had we had that debate in [63:45] 2024 too and it came with a capex hickey [63:48] in the summer at the same time with the [63:50] BJ and a week NFP and the S&P took like [63:52] a 12 14% [63:54] >> uh dent then as well and we had one in [63:57] 2025 although it was comingled with [63:59] everything that was going on with [64:00] tariffs so to have these AI summers and [64:02] winters is not uncommon right and so the [64:05] end of August was sort of our [64:06] demarcation to say, "All right, I'm [64:07] ready to lean into a much more bearish [64:10] view on AI because of that revenue [64:12] aspect." And then we also had sort of [64:15] separately but related talked about this [64:17] sort of competition for capital [64:18] narrative back in the summer as well. [64:20] And then we sort of just merged those [64:22] two stories together and that's kind of [64:24] where we sit today. So our quant model [64:25] was tell still telling us be very [64:27] bearish tactically, right? Camouflage [64:30] and then uh and then the AI bit and the [64:33] financing bit. So people say like, "Oh, [64:34] what's the catalyst for the market to go [64:36] down?" I'm like, "At this point, I don't [64:38] really need a catalyst, right? I'm kind [64:40] of like, you've got all the ingredients. [64:41] They're already there. The financing [64:43] bits kind of already catalyzing. [64:46] The revenue bit, I just don't think [64:48] there is a specific catalyst that we can [64:50] really lean into. The capex commitments [64:52] are kind of already there. So those [64:55] numbers aren't changing at the very [64:56] earliest until probably the four-year [64:58] results in January, right? So I'm like [65:01] and then our quant model is kind of [65:03] telling you that real rates are high [65:04] spot v correlation is high that CTAs v [65:08] control more less CTAs more vault [65:10] control and QIS strategies are extremely [65:12] long equity risk right now retail's [65:13] extremely long equity [65:14] >> so the B so the bulls might be trapped [65:16] at least for the time being until [65:18] expectations change or prices change. [65:21] >> Yeah. So people say like what what would [65:22] get you more bullish? Well, you know, [65:23] again, our quant framework would be [65:25] helpful for that. But really, just for [65:27] me, the clearing price of risk being [65:29] lower just would be quite frankly very [65:31] healthy for this market, even if it [65:33] comes with some near-term damage. [65:34] >> We could use a little bit of a slap on [65:36] the wrist. It's been a minute. [65:37] >> It has been a minute. You know, [65:38] >> it's been below 20 forever and ever. [65:39] >> And the largest pullback we had so far [65:41] in 2026 around Iran, I think, was 8 and [65:43] a half%. [65:44] >> Not enough. [65:45] >> It's It wasn't a lot. So, in my head, [65:48] I've got between 71 and 7200, which is [65:50] around the 200 day moving average, [65:52] right? That gets us down to about 17 and [65:54] a half times forward NTMP. [65:56] >> Take a lot of risk out along the way to [65:59] get like at that level. You're just [66:01] repricing everything basically. [66:02] >> You reperric everything and then I think [66:04] we can go again. You know, I think [66:05] there's probably another round of chip [66:06] financing. There's probably another [66:07] interesting [66:08] >> innovations that people get excited [66:10] about. [66:10] >> Yeah. Like I mean do you remember when [66:12] the iPhone came out the most exciting [66:13] bit of technology at the time was when [66:15] you had that app uh that you could like [66:17] drink a beer, you know, it had the [66:19] gyroscope in it. Yeah. Like clearly that [66:22] was not peak iPhone, right? So I don't [66:25] think we've yet even discovered or even [66:27] imagined what peak AI is going to look [66:29] like years from now. So I'm being a very [66:32] I'm trying to be very open-minded about [66:33] the bullish aspect of AI too. Everyone's [66:35] like, "Oh, you're just a bear." I'm [66:36] like, "First of all, I haven't been a [66:37] bear all year." And secondly, like, [66:39] >> and it's tact what you're saying is [66:40] tactical. [66:41] >> We've spoken to bears. This is not This [66:43] is [66:43] >> No. Then you get into Q1 and some uh [66:46] biotech company says, "We just got [66:48] through a phase three trial. 30% faster [66:51] than we would have without uh AI or the [66:54] robots start showing up in workplaces. [66:57] It's just like there'll be more to it. [66:59] But I do agree with you like all right [67:02] now we're getting these tech earnings [67:03] and the reaction is negative. [67:05] >> Yeah. [67:05] >> Where do we go from here? If not a [67:07] repricing of [67:08] >> You're right. The capex spend as we're [67:09] seeing as we're learning with these [67:10] earnings announcements. It's exhausted. [67:12] People are just not buying up buying it [67:13] anymore. The stock price [67:14] >> people don't want to hear it anymore. [67:16] And you know who the first were to to [67:17] really sort of cry wolf about that? It [67:20] was it no it was the Asian invest [67:22] investment community [67:24] >> because even at this at the when was [67:26] this right at the beginning of September [67:27] is when I was out there in Hong Kong and [67:29] Singapore and I came out there with like [67:31] my big oh I don't know where the AI [67:33] revenue is going to come from I'm being [67:35] really smart about this and I went out [67:37] there and every single meeting every [67:39] single meeting I had push back there [67:40] they would be like what do you mean like [67:42] but semis are trading so cheap and this [67:43] and that and I would give them my whole [67:45] rationale as to why I I I I was sort of [67:47] cautious and at the end of the [67:49] conversation [67:50] 90% of them will go, "Yeah, we agree [67:51] with you." [67:52] >> Oh my god. And [67:53] >> look at you. Look at you. [laughter] [67:57] >> Love that. [67:58] >> And then and but so that's so the [67:59] realization for you is okay, there's [68:02] this facade of everybody wants to seem [68:04] positive, but everybody has that same [68:06] reservation or that same concern deep [68:08] down. [68:09] >> Yeah. I mean, look, look at the poll [68:10] poll I sent around today. Uh [68:13] the vast majority of investors still [68:15] want to be bullish, right? They want to [68:17] buy semis because they're cheap, right? [68:20] They want to be in the AI trade because [68:21] they use it and they [68:22] >> I think they want to be seen as they get [68:24] it is another big component to this. [68:27] >> Mass very much on the West Coast that [68:29] >> Yeah. people don't want to be the anti- [68:31] AI guy. You look like a crazy person or [68:33] a loser. [68:34] >> Well, also chances are if you've been [68:36] anti-A any period for the past 3 years, [68:38] you [68:38] >> you look like an idiot [68:39] >> or you've been you've lost your job. [68:40] >> You've lost your Okay, I agree with [68:42] that. [68:42] >> Surprise for the urine, but we're not [68:43] super bearish. Just a little bit [68:44] bearish. No, I'm I'm like, as I said, I [68:47] mean, [68:47] >> tactically bearish due to positioning. [68:49] >> Yeah, there we go. Perfect. [68:51] >> Is that fair? [68:51] >> I think that's pretty fair. [68:53] >> Why do they call you Aldi? [applause] [68:55] >> Why do they call me Alti? Uh, so when I [68:58] when I started in high school, [69:00] >> Cuz I have a better nickname for you, [69:01] but I'm not going to say it until you [69:02] you say why. [69:03] >> Oh, I thought they say it off off the [69:05] air. [69:05] >> No, I want to hear like where did the [69:06] nickname come from? [69:07] >> Okay, so in my when I went got to high [69:09] school, there were five Alex's in my [69:11] class. Five Alex's, two Alisters, one [69:14] Alec, and one Alli. [69:16] >> Right. So, everyone got last name. [69:17] Everyone got I went to this sort of like [69:19] fancy school in London. [69:21] >> I can't believe it. [69:22] >> That's where the the accent's real and [69:24] uh and and so everyone got called by the [69:27] by their last name, right? Okay. And it [69:29] kind of annoyed me. I So, it was it's [69:31] very like teacher pupil thing in the UK. [69:34] And so, when I got into the workplace, I [69:36] just didn't want to be called by my last [69:37] name. And there were there were [69:39] obviously Alex is a pretty common name. [69:40] So I was just like, "All right, let can [69:42] we let's just find something else that [69:44] you can call me that isn't going to [69:45] offend me." Like I feel like I'm being [69:47] talked to by my like teacher. [69:48] >> Oh, so when you hear Alman, it sounds [69:50] like the headmaster has called you down [69:51] to his office [69:52] >> a little bit. That not anymore. Now I [69:54] don't care. But like it kind of So that [69:55] started in like 07 and then it kind of [69:57] stuck with my team and then it kind of [69:59] spread out from there. [70:00] >> So you said, "Call me alty." And then [70:01] that that stuck. [70:02] >> Yeah. [70:02] >> Okay. I would call you liquid alt. [70:05] [laughter] [70:06] >> That's I like it. [70:07] >> The other one's the other one's control [70:09] alt and delete. control. Oh, I I like [70:12] that, too. Did you have fun on the show [70:13] today? [70:14] >> I had a great time. [70:15] >> All right, so we're going to take like a [70:16] 2-hour recess for supper. You can go to [70:18] the gym and then we're going to come [70:19] back and do another two hours. Great. [70:21] You ready? [70:21] >> Sounds good, guys. Alex Alman, I want [70:23] So, where do people You guys are talking [70:26] to institutions, hedge funds primarily. [70:28] Um, so if they're customers of Barclays [70:30] Bank, they could subscribe to your uh [70:34] desks research. You guys doing anything [70:36] public facing or beyond or? [70:39] >> We're doing this. [70:39] >> Okay, so that's so that's good. [70:41] >> Hey, send this green. What do you say [70:43] now? [70:44] >> Send this green. [70:44] >> It's great. [70:45] >> There you go. [70:46] >> All right. Delete everything we just [70:47] said. Start again. [70:49] >> All right, Alex, this has been so much [70:50] fun for us. We really appreciate all the [70:52] insights and we hope you'll come back [70:54] sometime. [70:54] >> I'd love to. Thanks so much. [70:55] >> Are you busy tomorrow? [70:56] >> Oh, yeah. Come back. Let's do it. Let's [70:58] make it weekly. [70:58] >> Guys, Alex Alman, shout out to Barclays [71:01] and the Bets team. Much appreciated. [71:03] Thanks for listening. Thanks for [71:05] watching and we'll see you soon. [71:10] [applause]