Tech Outperformance Hits 6-Sigma Peak as SpaceX IPO Looms

科技股超额表现触及6西格玛峰值，SpaceX IPO临近 DataTrek研究显示，近期科技股涨幅达到六个标准差的极端水平，预示着资金可能轮动至软件板块；与此同时，SpaceX直接面临1.8万亿美元估值的检验，将揭示市场是否面临1999年式的泡沫风险。 专家小组分析了科技股大幅跑赢大盘这一罕见的统计异常现象：尽管软件板块走弱，但半导体和AI基础设施需求仍推动科技股大幅走高。他们探讨了算法流动性如何加剧抛售，以及2年期美债收益率上升如何改变宏观经济前景。此外，他们剖析了即将到来的SpaceX IPO结构，利用历史首日涨幅评估市场情绪，并阐明了为何当前市场状况与1999年互联网泡沫时期的真正狂热存在显著差异。 标普 500 科技板块 ETF（XLK）在 50 天滚动窗口内录得 +6 个标准差（+6σ）的历史极端跑赢表现，超额收益达 29.3 个百分点，触发量化算法拉大买卖价差并放大了周五的技术性抛售。 (2:00) 半导体板块与软件板块走势分化达到 +4σ 的极端水平，在强劲的 AI 基础设施支出与盈利预期上修推动下，半导体跑赢软件板块达 44 个百分点。 (6:25) 受强劲就业数据推动，2 年期美债收益率升至周期新高，由于历史上收益率见顶后 1 至 2 年往往伴随经济衰退，过快上升的利率正重新引发对高估值科技股的担忧。 (17:27) SpaceX 计划以每股 135 美元固定价格（估值 1.7 万亿至 1.8 万亿美元）直接挂牌上市并绕过传统机构询价；其首日涨幅将被视作核心情绪指标，若飙升超 70% 则标志着市场重回 1999 年式的极端泡沫。 (19:45)

DataTrek Research co-founders Jessica Rabe and Nick Colas argue that while recent US tech outperformance reached an unprecedented six standard deviation anomaly, broader market breadth and structural checks suggest the sector is experiencing extreme positioning and potential rotation rather than a 1999-style bubble collapse. Statistical Extremes and Algorithmic Selling (2:00) Through June 2nd, the S&P 500 Tech Sector ETF (XLK) outperformed the broader S&P 500 by 29.3 percentage points over a rolling 50-day window. Compared to the historical 2015-to-present average outperformance of 1.6 percentage points and a standard deviation of 4.0 percentage points, this surge exceeded the 6-sigma threshold of 25.6 points, marking the widest divergence in a decade. Rabe notes this move surpassed prior extremes, including the 2020 pandemic rebound and the 2022 rate shock. Colas and Rabe explain that algorithmic market-making models, which track multi-year volatility thresholds, registered tech as severely overextended heading into day 47 of the 50-day window. When tech began to pull back, these quantitative systems widened bid-ask spreads and pulled liquidity, accelerating the decline into Friday's sell-off. Colas warns that such statistical anomalies prevent reliable short-term forecasting and warrant tactical risk management rather than reflexive dip-buying. Semiconductor Dominance and Software Dislocation (6:25) Within technology, performance diverged sharply between hardware and software. On June 3rd, semiconductor stocks (SMH) outperformed software (IGV) by 44 percentage points over a rolling 50-day window, compared to a historical norm of 2.7 percentage points—a 4-sigma divergence. Rabe attributes this spread to persistent AI infrastructure demand and positive earnings revisions for hardware names. In contrast, software only began to recover immediately prior to the pullback, led by cybersecurity firms like CrowdStrike and Palo Alto Networks alongside Oracle, while Microsoft lagged. Given the wide statistical divergence, Rabe suggests tactical investors consider rebalancing capital by trimming semiconductor positions into software. For long-term portfolios, she recommends maintaining index-weight tech allocations to balance AI infrastructure exposure against single-industry concentration risk. Broader Market Trajectory and Macro Shifts (9:40) In contrast to tech's 6-sigma move, the S&P 500 registered a more moderate 2-sigma advance, gaining 15.3% over 50 days against its 10-year mean return of 2.4%. Historical precedents since 2015 show that after crossing the 2-sigma threshold, the S&P 500 was higher 50 trading days later in 100% of cases, averaging a forward gain of 7.3% with a minimum return of 2.6%. At the same time, shifting macroeconomic signals injected friction into risk assets. A robust US jobs report with upward revisions pushed two-year Treasury yields to new 12-month highs, prompting Goldman Sachs to remove year-end rate cut projections from its baseline outlook. Colas cautions that economic recessions historically arrive one to two years after two-year yields reach a cycle peak, meaning rising yields could threaten tech multiples if elevated capital costs begin to impair corporate balance sheets. SpaceX IPO Structure and Bubble Scorecards (19:45) The panel views the anticipated SpaceX offering—expected at a fixed $135 per share, or a $1.7 trillion to $1.8 trillion valuation—as an operational test of market sentiment. Underwriters bypassed traditional institutional price-discovery book-building, leaving valuation entirely to day-one secondary trading. Colas frames day-one performance using Jay Ritter's historical IPO data: an opening-day close between $154 and $166 (a 14% to 22% gain) aligns with normal tech listings, while a close at $231 (a 71% gain) would match peak 1999 euphoric conditions. The offering also introduces technical market structure risks. SpaceX is slated for rapid inclusion into the Nasdaq-100 within roughly 15 days, which Colas estimates will drive $15 billion to $20 billion of passive buying against a $75 billion float. Furthermore, non-traditional lockup expiries allow 20% of non-Musk insider shares to trade after Q2 earnings, followed by additional tiers tied to share price gains and Q3 earnings. Colas and host Josh Brown recall Facebook's 2012 debut, noting heavy opening-day retail volume can overwhelm exchange infrastructure. Current Market Structure Versus the 1999 Dot-Com Bubble (29:46) Addressing concerns of a structural 1999 rerun, Colas reviews historical price action to show that true market bubbles require violent, two-way volatility. Between late 1999 and March 2000, the Nasdaq surged 84% in under six months (rising 37% in Q4 1999 and another 24% in early 2000) before collapsing 34% across five weeks. Colas emphasizes that genuine asset bubbles reflect a complete breakdown of fundamental pricing anchors through rapid double-digit booms and busts. Brown and Rabe observe that today's market capitalization makes such compounded momentum nearly impossible. Given the multi-trillion-dollar scale of modern mega-cap tech, the broader market lacks the liquidity necessary to double index valuations over a single quarter. With the Nasdaq up approximately 34% over the prior year, current gains remain far below the threshold of 1999 euphoria. While tech's 6-sigma outperformance and semiconductor concentration flash statistical caution signs, broader equity indicators and the sheer size of mega-cap balance sheets differentiate the current backdrop from late-1990s bubble conditions. Whether Friday's sharp tech decline proves to be a momentary reset or the trigger for broader sector rotation remains contingent on yield trajectories and market liquidity.

DataTrek Research 联合创始人 Jessica Rabe 与 Nick Colas 指出，美股大型科技股近期跑赢大盘的幅度突破了6个标准差的统计极限，虽然量化交易回撤与美债收益率反弹引发了短期剧震，但目前的市场环境与1999年互联网泡沫破裂期存在结构性差异。 科技股极端跑赢与算法交易放大回撤 (2:00) 根据 Jessica Rabe 追踪的2015年至今数据，以科技精选行业指数ETF（XLK）相对标普500指数的50日滚动收益率衡量，科技股截至6月2日的超额收益达到29.3个百分点。历史均值仅为1.6个百分点，标准差为4.0个百分点，6个标准差门槛为25.6个百分点，这一读数创下了十年来的极值纪录，超越了2020年疫情反弹及2022年加息周期等历史极端区间。 Nick Colas 补充说明，依赖波动率与标准差参数运行的量化做市与算法模型在科技股处于过度延伸状态时表现高度敏感。当周五抛售发生时，算法系统迅速拓宽买卖价差并减少流动性供给，从而加剧了市场的短线下挫。韩国等海外市场同样出现极端动量，甚至连续三天触发熔断机制。Rabe 认为，在进入前所未有的统计偏离区间后，战术交易者理应提高警惕，而非单纯将下跌视作普通的买入时机。 半导体与软件的4西格玛内部裂痕 (6:25) 在科技板块内部，半导体行业与软件板块的分化同样刷新了纪录。Jessica Rabe 引用半导体ETF（SMH）与软件ETF（IGV）的对比指出，虽然硬件端长期受技术升级驱动平均跑赢软件2.7个百分点，但在6月3日，半导体50日超额收益率高达44个百分点，触发了4个标准差的极端事件。这主要由基本面超预期、盈利预测上调以及对AI基础设施建设的持续信心所推动。 Josh Brown 观察到，在周五抛售前，CrowdStrike、Palo Alto Networks 以及 Oracle 等网络安全和企业软件龙头已呈现回暖迹象。Rabe 就此提出战术性操作建议，认为战术投资者可考虑减持部分涨幅过大的半导体仓位并轮动至软件板块，而长期投资者则宜维持基准科技权重，以平衡AI配置与行业集中度风险。 标普500胜率历史与宏观加息隐忧 (9:40) 相较于科技股的极端波动，标普500指数整体走势展现出历史统计上的积极信号。截至6月1日，标普500指数50日收益率达到15.3%，属于高于历史均值（2.4%）2个标准差的走势。自2015年以来该指数共出现过5次类似情况，在随后的50个交易日内上涨概率达100%，平均远期收益率为7.3%，最低一次涨幅亦有2.6%。 然而，强劲的就业报告与前值上调打乱了宏观利率预期。2年期美债收益率创下12个月新高，高盛移除了年内降息的基准假设，市场关注度全面转向由 Kevin Warsh 主持的6月16日至17日美联储议息会议。Colas 警告称，从历史周期来看，经济衰退往往发生在2年期美债收益率触及周期顶点后的1到2年内，若无风险利率飙升过快，估值承压与再融资风险将成为现实威胁。 SpaceX超常规IPO对市场情绪的检验 (19:45) 拟议中的 SpaceX IPO 绕过了常规的投行机构询价建档流程，由 Elon Musk 直接将发行价固定在每股135美元，对应1.7万亿至1.8万亿美元的极高估值。Colas 表示，机构投资者普遍更倾向在80至100美元（1.2万亿至1.5万亿美元估值）建仓，这一固定定价将定价权直接推给了首日二级市场的散户交易者。 根据佛罗里达大学 Jay Ritter 教授的历史数据，1997至1998年科技新股首日平均涨幅为14%至22%，近年均值为17%，而1999年泡沫峰值期首日平均涨幅达71%。Colas 认为，若 SpaceX 首日收盘冲至231美元（即涨幅达到71%），将成为市场步入全面泡沫的明确信号。此外，该股预计在挂牌约15天内快速纳入纳斯达克100指数，有望在750亿美元流通盘中吸引150亿至200亿美元被动配置资金，但二季度财报后非马斯克持股解禁等条款亦可能带来剧烈波动与交易系统承压风险。 当前环境与1999年互联网泡沫的区别 (29:46) 针对市场频发的泡沫质疑，Nick Colas 对比了1999年末至2000年初的纳斯达克综合指数表现。当时纳斯达克在1999年第四季度飙升37%，2000年一季度再涨24%，在不到6个月内累计复合涨幅达84%，随后在短短5周内崩跌34%。Colas 强调，真正的资产泡沫必须伴随双向的极致波动，即80%以上的非理性狂飙紧随30%以上的断崖式瓦解，这代表底层定价锚彻底失效。 Josh Brown 与 Jessica Rabe 补充指出，由于当前万亿美元级别科技巨头的体量极其庞大，整个市场缺乏促成纳斯达克指数在数月内翻倍的流动性基础。过去一年纳斯达克约34%的温和涨幅与1999年的全面亢奋不可同日而语，当前更多表现为头部集中后的风格轮动，而非全面性的估值幻灭。 本期讨论表明，美股大型科技股与半导体板块已进入统计学意义上的罕见超买阶段，短线面临量化去杠杆及美债收益率反弹引发的轮动压力。然而，在缺乏持续翻倍式投机狂热与宏观衰退实据的前提下，将当前调整等同于1999年泡沫破裂仍缺乏证据支持，核心关注点依然在于利率变动对资本开支预期的实际冲击。

## Background 本期视频由财经播客频道 The Compound 于 2026 年 6 月 13 日在 YouTube 上发布。根据页面简介，主持人 Josh Brown 携手 DataTrek Research 的两位联合创始人 Nick Colas 与 Jessica Rabe，深入探讨了科技股近期跑出 6 个标准差历史极值超额收益的影响、半导体与软件板块的分化、利率逆风，以及 SpaceX 备受瞩目的重磅 IPO 与 1999 年互联网泡沫的深刻教训。 ## TL;DR 节目围绕美股科技板块在 50 天内超越标普 500 指数达 6 个标准差的历史级暴涨展开，剖析了半导体与软件之间拉大至 4 个标准差的内部割裂，以及宏观利率飙升和做市量化算法如何在周五触发级联式抛售。随后，嘉宾深度拆解了 SpaceX 以 135 美元固定价格上市背后的反常架构、首日涨幅作为泡沫试金石的意义，以及纳斯达克快速纳指规则带来的供求冲击。最后，通过复盘 1999 年纳斯达克半年暴涨 84% 紧随 5 周重挫 34% 的真实历史，团队强调眼下市场虽需戒备但基本面与体量远未达到当年价格发现彻底崩溃的泡沫程度。 ## Atmosphere & style 整场对谈兼具华尔街老将的沉稳专业与老友重聚的轻松风趣，在严谨的数据量化图表与生动的市井隐喻间自如切换。从亲身经历 1999 年 SAC 交易台风暴的老派视角，到用“马路街头曲棍球看车避让”来形容当下的风险心理，对话节奏紧凑且充满对市场的敬畏感。聆听体验既有在统计极值面前的警醒与理性解构，也有面对历史泡沫和重磅 IPO 时带来的强烈智性激荡。 ## Bulletin - [02:00](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=120s) 美股科技板块（XLK）在截至 6 月 2 日的 50 个交易日内跑赢标普 500 指数达 29.3 个百分点，突破 6 个标准差（6 Sigma）统计极值，刷新过去十年最高纪录。 - [05:18](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=318s) 半导体板块（SMH）50 天内跑赢软件（IGV）整整 44 个百分点，偏离历史均值 4 个标准差，引发做市算法在极端位点撤出流动性及板块内部轮动预期。 - [13:58](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=838s) 非农数据上修叠加两年期美债收益率创年内新高，彻底消除降息预期并放大量化级联抛售，科技板块单日暴跌 6.66%。 - [19:45](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1185s) SpaceX 拟以 135 美元固定价格发行上市，对应 1.7 至 1.8 万亿美元估值；因缺乏机构建簿询价，其首日溢价幅度将成为对照 1999 年狂热与否的核心量尺。 - [30:07](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1807s) 对比 1999 年末纳斯达克不到半年暴涨 84%、随后 5 周崩跌 34% 的价格发现崩溃，当前纳指年涨幅仅约 34%，底层基本面与资产体量与当年不可同日而语。 ## Key takeaways - [02:00](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=120s) 科技板块相对大盘出现 6 个标准差的史诗级拉升，超越了历次危机反弹，进入近代统计学上的未可知之境。 - [05:18](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=318s) 半导体与软件板块的相对回报差拉开至 4 个标准差的极值，战术配置上适宜减仓芯片并寻找软件补涨，或切换为均衡配置的行业指数。 - [08:37](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=517s) 标普 500 自身 50 天上涨 15.3% 仅触及 2 个标准差，历史规律显示冲刺突破后大盘在未来 50 天全部收涨且未显现系统性非理性泡沫。 - [10:57](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=657s) 做市算法依赖于 1500 天日收益率波动率模型，面对统计极端拉升并在市场反转时主动拉宽买卖价差并撤单，放大了级联抛售。 - [10:57](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=657s) 韩国股市此前因半导体狂热暴涨 100%，近期因极端偏离逆转连续触发暂停交易的“边车机制”，反映出跨市场量化清仓的同步性。 - [13:58](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=838s) 宏观利率预期的剧烈逆转直接击垮了高杠杆散户看涨期权头寸，两年期美债收益率飙升与 FOMC 政策隐忧放大了科技股调整烈度。 - [17:48](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1068s) 面对 6 个标准差异动，投资者不能抱持“街头曲棍球”式恢复如常的侥幸心理，而应在维持看多的同时时刻防备下一轮冲击。 - [20:37](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1237s) SpaceX 打破传统投行路演询价流程，直接敲定每股 135 美元固定价格发行，将实际的价格发现责任全部甩给上市首日的散户交易。 - [21:51](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1311s) SpaceX 首日收盘表现将成为泡沫参照系：若收在 154 至 166 美元属于 1997-1998 年温和常态，若冲至 231 美元（+71%）则意味着迈入 1999 年级别的极度狂热。 - [23:49](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1429s) 特斯拉是马斯克旗下 2 至 3 年内依赖机器人变现的近期期权，而 SpaceX 则是承担庞大资本支出且跨越 10 年以上的远期期权，两者短期并无合并基础。 - [26:01](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1561s) SpaceX 上市 15 天内将因纳指快速纳入规则迎来 150 亿至 200 亿美元被动买盘，但罕见的前期快速分批解禁机制将加剧上市初期的筹码博弈与波动。 - [27:15](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1635s) 2012 年 Facebook 上市时因天量挂单导致纳斯达克交易系统故障的前车之鉴表明，超大体量散户涌入需警惕微观交易系统承压的流动性风险。 - [31:46](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1906s) 真正界定网络泡沫的是“价格发现机制的彻底失灵”：1999 年 Q4 至 2000 年 3 月指数暴涨 84%，紧接着 5 周内雪崩 34%。 - [31:46](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1906s) 依照“纳指翻倍才算泡沫”的经验法则，当前由数万亿美元巨头构成的美股市场缺乏足够的边际资金推动整体指数短期翻倍，基本面稳固度远胜 1999 年。 ## Detailed walkthrough ### [00:13](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=13s) 节目赞助与开场引言 节目开场由 WisdomTree 赞助播出。WisdomTree 强调日本正迈入一个全新的时代，公司治理改革与更强有力的股东回报政策正在彻底改变游戏规则，连沃伦·巴菲特（Warren Buffett）等知名投资人也早已关注并布局。WisdomTree 推出的日本机遇基金（WisdomTree Japan Opportunities Fund，代码：OPPJ）旨在投资那些有望受益于宏观经济政策、产业创新以及贸易和消费者行为转变的日本企业。 主持人 Josh Brown 热情开场，欢迎观众收看最新一期的《What Did We Learn?》。Josh 抛出了眼下整个股市面临的核心疑问：刚刚在周五发生的那场史诗级的科技股小型崩盘（mini crash），究竟是一场更大级别风暴的开端，抑或只是多年后回望时的一个小小波澜？为了解答这个关键问题，Josh 邀请了 DataTrek Research 的两位联合创始人——Nick Colas 和 Jessica Rabe。他们每天撰写的《DataTrek 早间简报》（Morning Briefing）发送给超过 1,500 家机构和零售客户。Josh 在向两位老朋友致以夏日问候后，迅速切入核心数据讨论。 --- ### [02:00](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=120s) 6 个标准差事件：科技股相对 S&P 500 的历史性极值 ```mermaid flowchart TD accTitle: 50-day relative return comparison between Tech and S&P 500 accDescr: Charting the 50-day rolling outperformance of the XLK ETF over the S&P 500, highlighting the extreme 6 standard deviation surge in June. subgraph Baseline ["历史基准 (2015 至今)"] Mean["50 日平均超额收益: +1.6%"] SD["标准差 (1 Sigma): 4.0%"] Sigma6["6 Sigma 阈值线: +25.6%"] end subgraph Event ["2026 年 6 月 2 日极值"] Outperform["XLK 超额收益达: +29.3%"] Extreme["> 6 个标准差事件，十年未见"] end Mean --> Sigma6 SD --> Sigma6 Sigma6 -.突破.-> Outperform Outperform --> Extreme ``` Jessica Rabe 指出，美国大盘科技股刚刚跑出了过去十年在统计学意义上最极端的超额收益表现，甚至超越了历次危机后的反弹。截至 6 月 2 日的过去 50 个交易日内，以科技精选行业 SPDR 基金（XLK）为代表的科技板块跑赢标普 500 指数达 29.3 个百分点。 从 2015 年至今的数据来看，科技板块在任意给定的 50 日持有期内，相对标普 500 的平均跑赢幅度为 1.6 个百分点，其标准差为 4 个百分点。因此，超越平均水平 6 个标准差（6 Sigma）的红虚线阈值位于 25.6 个百分点。而 6 月 2 日记录到的 29 个百分点（准确为 29.3%）不仅超过了 6 个标准差，更是数据集里遥遥领先的最极端数值。更令人震惊的是，这 50 个交易日的起点甚至还处于 3 月底市场触底之前的 6 天，这意味着该涨幅已经完全包容并消化了年内最大的一轮暴跌行情。 针对这一极端形态，Jessica 梳理了三大核心观点： - **缺乏历史经验对照**：没有任何近期的历史脚本可以解释科技股眼下的走势。这场涨幅直接碾压了以往每一次承压复苏阶段——包括 2020 年的疫情危机、2022 年的加息冲击、2025 年的贸易政策冲击，以及反复上演的 AI 情绪恐慌。虽然 2025 年 3 月的低点解释了行情的起点，但完全无法解释其膨胀的量级。 - **做市量化算法的戒备**：支撑美股交易的底层量化算法完全清楚这些数学概率。在上周五遭遇抛售时，已经是自 3 月 30 日低点以来的第 47 个交易日，科技股早已处于极度过度延伸（overextended）的状态。做市算法接下来可能会保持高度谨慎，并放大科技股的任何后续抛售。 - **风险管理的审慎信号**：任何具备成熟风险管理流程的交易员都会将该数学极值视作警惕科技股的决定性理由。至于周五的下跌究竟只是一次短暂喘息，还是一场大规模行业板块轮动（rotation）的序幕，眼下"在统计学上纯属未可知之境"（genuinely unknowable at this point because we are in truly uncharted statistical territory）。 --- ### [05:18](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=318s) 历史类比：对比 1999 年与半导体对阵软件的割裂 面对这种超出统计范畴的极端涨幅，Josh 直言在市场中历练多年的投资者凭借直觉都会联想到过去：极端科技暴涨往往不会迎来美妙的收局，多数人脑海里立刻浮现的就是 25 年前互联网泡沫（Dot-com bubble）的破裂。Josh 追问，眼下这 50 天的暴涨强度，是否能与 1998 年或 1999 年的巅峰相提并论？Nick 与 Jessica 明确回应，眼下的走势虽然极度夸张，但若与 1999 年相比依然完全不可同日而语，节目后半程将用具体数据详述当年的真实情况。 Jessica 顺势引入了第二张深度图表，剖析科技板块内部的两大子支柱——半导体（Semiconductors）与软件（Software）。该图表以 VanEck 半导体 ETF（SMH）和 iShares 扩展科技软件 ETF（IGV）作为代理变量，追踪两者 50 日滚动相对回报率： - **芯片压制软件的长期逻辑**：历史上半导体表现向来优于软件。尽管摩尔定律（Moore's law）近年有所放缓，但芯片每美元算力的提升翻倍依然迅速，推动产品不断拓展至更多应用场景。半导体在任意 50 天内平均跑赢软件 2.7 个百分点，且这种超额收益优势还在扩大。 - **4 个标准差的内部割裂**：截至 6 月 3 日，半导体在 50 天内跑赢软件整整 44 个百分点，高出长期均值整整 4 个标准差（4 Sigma），刷新过去十年以来的最高分化纪录。虽然半导体基本面表现极其强劲、盈利预期不断上修、且资金对 AI 基础设施建设满怀信心，但市场走势从来不是单向线性的。 - **战术与配置建议**：当板块领涨力量过度集中、相对表现达到此类极值时，下一步往往是板块内部的轮动，而非单纯单边加速。战术型投资者理应削减半导体敞口并审视软件板块的补涨机会；而对于看好 AI 长期叙事的长线投资者，采取跟随大盘权重的科技行业指数配置方案（如 XLK）更为合理，既保留了行业 Beta，又能规避半导体一旦均值回归可能带来的过重惩罚。 --- ### [08:37](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=517s) 软件股突围与 S&P 500 的后续空间 Josh 观察到，在过去几周里，科技股加速跑赢大盘的一个隐秘催化剂，在于软件股本身出现了大幅报复性反弹。资金并没有从半导体中抽逃出来买软件，而是两大阵营并驾齐驱地同时狂飙。虽然微软（Microsoft）走势相对平淡，但作为 IGV 核心成分股的甲骨文（Oracle）王者归来，网络安全股更是陷入疯狂——无论是 Josh 自己持有的 CrowdStrike 还是 Palo Alto Networks 都在暴涨，科技板块内部原本的拖累项一夜之间反转为助推器，形成了科技全线齐涨的盛况。 Jessica 对此表示认同，并调出第三张图表进行宏观视角（Zoom out）的审视：标普 500 指数自身的 50 日滚动回报率。 | 评估指标 | 标普 500 指数表现（S&P 500） | 科技板块表现（XLK / 半导体） | |:--- |:--- |:--- | | **近十年 50 日平均回报率** | 2.4% | 超额跑赢大盘 1.6% | | **当前 50 日涨幅水平** | 截至 6 月 1 日上涨 15.3% | 科技跑赢 29.3% / 芯片跑赢软件 44% | | **标准差偏移幅度** | 达到 2 个标准差（2 Sigma） | 达到 6 个标准差（XLK）与 4 个标准差（SMH/IGV） | | **过往历史发生次数（2015 至今）** | 曾出现过 5 次突破 2 Sigma | 本次为 10 年来最极端的首例孤本 | | **突破后的历史走势（未来 50 天）** | 5 次全部走高，平均未来回报达 7.3%（最差为 +2.6%） | 缺乏历史前例，做市算法高位戒备 | Jessica 总结道，价格动量（price momentum）在资本市场中是一股极度强大的力量。即便科技股估值和涨幅延伸过长，半导体也亟需休整，但大盘本身并没有展现出等同于科技股的非理性泡沫。历史规律清晰地表明，大盘在经历 2 个标准差的强势冲刺后，标普 500 整体依然留有继续上行的充裕空间。 ### [10:57](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=657s) 市场结构的算法连锁反应与跨市场抛售 主持人 Josh Brown 与嘉宾们继续深入探讨 Jessica 拿出的科技股统计图表。Nick Colas 赞叹道，这组图表之所以被放在开头，是因为它极其精准地勾勒出了眼下的市场全貌：科技板块经历了一轮过去十年未曾见过的超级大涨。Nick 回忆起自己在 1999 年于知名对冲基金 SAC 交易台上的岁月，直言任何试图将当下直接与 1999 年类比的人，若当年没亲自坐在交易台前操作过那些股票，就不应妄下定论，因为两者完全是“两码事”（apples and oranges），1999 年的疯狂与暴涨幅度远比今天剧烈得多。然而，即便如此，眼下科技股的回报率也已经到了不可思议的程度。 Nick 将核心症结指向 Jessica 提到的“市场结构”：掌控并调控全美资本市场流动性的各类量化与做市算法，本身就是建立在数学模型之上的。这些算法底层逻辑完全依赖于 1500 天日收益率标准差等统计数据。上周五之所以出现瀑布般的级联式抛售，正是因为做市算法瞬间识别出当前处于统计上极度异常的偏离位，在市场发生反转（rollover）时直接选择撤退并大幅拉开买卖价差（widening out bid-ask spread），拒绝在下跌动量面前接飞刀承担风险。 在被问及抛售源头究竟是韩国股市波及美股存储芯片板块（如 Micron、SanDisk 等），还是由美股传导至亚洲时，Nick 倾向于认为是美股率先走弱，随后迅速蔓延至韩国。但韩国市场的表现比美股更夸张——韩国股市此前暴涨 100%，相对于新兴市场基准的超额表现，甚至比美股科技股相对于标普 500 指数还要极端。这种过度拉升触发了同样的量化逆转机制，导致高度波动的韩国市场在过去几个交易日内接连触发所谓“边车机制”（Sidecars）——即在市场波动剧烈、达到特定跌幅阈值时，交易所规则强制放缓或暂停交易 10 至 20 分钟以重置买卖盘的熔断机制。 ```mermaid flowchart TD accTitle: 极端统计异动下的算法抛售级联机制 accDescr: 展示从科技股收益率偏离到做市算法撤单、宏观利率共振最终导致暴跌的完整逻辑传导过程。 subgraph StatAnomaly["统计极端偏离"] A["科技股相对标普创下 6 个标准差暴涨"] B["韩国股市暴涨 100% 相对新兴市场过度拉伸"] end subgraph MarketStructure["市场微观结构与量化反应"] C["量化做市算法基于 1500 天波动率模型预警"] D["做市商拒绝接飞刀，大幅拉宽买卖价差"] E["触发交易所熔断与边车机制（Sidecars）"] end subgraph MacroShock["宏观数据共振"] F["非农就业上修，2年期美债收益率创年内新高"] G["高杠杆与零售期权多头仓位踩踏"] end StatAnomaly --> MarketStructure MarketStructure --> MacroShock MacroShock --> H["周五科技板块暴跌 6.66% 级联抛售"] ``` - **SAC 历史对比**：Nick 强调 1999 年 dot-com 繁荣的狂热度与杠杆性质与今日截然不同，但当下的科技股回报数字在近代统计分布中同样极其罕见。 - **算法模型触发**：现代做市算法紧密追踪基于 1500 天滚动窗口的日收益率标准差，一旦触及异常高位回撤，流动性会在瞬间抽离。 - **韩国边车机制（Sidecars）**：韩国股市近期在极端波动中连续多日触发相当于熔断的交易减速机制，反映了跨市场量化清仓的同步性。 --- ### [13:58](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=838s) 宏观逆风：非农修正、两年期美债新高与加息预期阴霾 上周五的极端抛售并不仅仅是半导体行业内部的获利了结，宏观层面的重磅炸弹直接将火势推向高潮。Nick 分析指出，当天公布的美国非农就业报告以及对前两个月数据的向上修正，彻底重塑了市场对美国劳动力市场的认知——从去年被认为的大幅放缓，逆转为依然十分强劲。利率市场对此反应极其剧烈，两年期美国国债收益率瞬间飙升并创下 12 个月以来的新高。市场对美联储不仅无法降息、甚至可能因政策失误而再度加息（Fed rate hike）并推高资金成本的担忧全面爆发。 Josh Brown 赞同这一观点，认为宏观利率叙事直接将一场原本由账面浮盈获利回吐引发的正常调整，演化成了剧烈踩踏。当前科技股中充斥着大量散户资金、杠杆头寸以及激进的期权看涨押注，浮盈的快速蒸发叠加宏观抛压，最终让盘面呈现出惊人的杀伤力。Jessica 指出，当天科技板块单日跌幅达到了充满神秘玄学色彩的 6.66%：“纯粹因为半导体板块走弱，科技股大概也就跌个 3%，但宏观因素的共振彻底放大了这场灾难。” 讨论随即转向科技股是否需要承受更多估值挤泡沫的阵痛。Jessica 提出，利率上升本身不一定会摧毁科技股，只有当利率上升的速度过快、快到市场无法适应并引发合理的经济衰退恐慌时，问题才会致命；正如 2022 年所发生的那样，超大规模云服务商（Hyperscalers）本质上也具备周期性现金流属性。Josh 则补充了一个令人不安的转变：仅仅在六个月前，全市场还在热火朝天地讨论防范性“保险式降息”（insurance cuts），而如今高盛已经彻底在其年终预测中剔除了降息选项，市场即将迎来 Kevin Warsh 主导下的首次 FOMC 会议。虽然市场对 AI 资本支出（CapEx）依然信心充足——无论是财报呈现的指引，还是私募市场对 Anthropic 高估值的追捧，抑或是 SpaceX 对生成式 AI 潜在市场空间（TAM）的乐观预期——但利率预期的彻底逆转，无疑给市场投下了全新的不确定性阴霾。 - **两年期国债的警示**：两年期美债收益率紧跟联邦基金利率预期，历史周期表明，两年期美债收益率见顶后的一到两年内，美国经济通常会步入衰退。 - **科技板块跌幅**：科技板块上周五单日暴跌 6.66%，其中半导体自身调整贡献约 3%，其余主要来自宏观利率冲击与杠杆解除。 - **降息预期归零**：高盛等华尔街大行正式撤销降息预测，FOMC 会议成为科技股多头不得不警惕的新催化剂。 --- ### [17:48](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1068s) 六个标准差事件之后：“街头曲棍球”隐喻与风险戒备 面对这一轮凶猛调整，Josh 向嘉宾们抛出了一个尖锐的核心问题：周五的暴跌究竟只是一场类似一月份 DeepSeek 引发的“茶壶里的风暴”（tempest in a teacup），可以很快被多头修复，还是意味着整个行情性质发生了深刻转变？ Nick 给出了旗帜鲜明的谨慎回应：“面对一个偏离度高达 6 个标准差（6 standard deviation move）的历史级事件，我绝不会拍拍屁股假装无事发生，然后说一切还会照旧。”他强调，虽然长期保持看多科技股没有错，但面对这样极端的图表，盲目乐观是危险的，投资者必须提高警惕并保留一份敬畏心。 为了形象地讨论眼下的投资者心态，Josh 分享了一个生动的童年回忆：小时候他们没有附近的真冰场，便戴着护具穿上旱冰鞋在马路上打“街头曲棍球”（street hockey）。大家把球门摆在马路两旁激战，每当有汽车开过来时，所有人就大喊一声把球网搬到人行道上避让；等汽车呼啸而过，大家立刻大喊一声“比赛继续！”（Game on!），把球门搬回马路正中央，仿佛刚才什么都没发生过一样，立刻无缝投入激战。Josh 追问 Nick：现在市场里的人是不是不应该抱着这种“车走了，球门搬回来接着狂欢”的心态？ Nick 借用这个隐喻幽默且严肃地警告道：“继续玩下去当然可以，但是，你得时刻紧盯着路口，随时防备下一辆汽车撞过来。”Jessica 同样重申，虽然他们团队一直以来都是科技股的长期看多者，但 6 个标准差的偏离度已经彻底破表（off the charts），绝对不能等闲视之。 --- ### [19:45](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1185s) SpaceX 即将重磅上市：异常的 135 美元固定价格发行 在告别科技股波动的话题后，Josh 将焦点转向了当前资本市场最为瞩目的焦点事件——SpaceX 的首次公开募股（IPO）。据市场最新传闻，这家航天巨头极有可能在本周末正式挂牌上市。Josh 转向参与过无数 IPO 项目的老将 Nick Colas，询问他面对如此庞大的募资规模、前所未有的投资者狂热以及铺天盖地的媒体轰炸，究竟作何感想。 Nick 透露自己专门为此准备了一份演示文稿，梳理出了投资者参与该交易必须搞清楚的五个核心要点。他开门见山地点出了第一个极其反常的细节：SpaceX 的上市交易架构非常怪异（a very odd deal），与传统 IPO 普遍采用的浮动价格区间询价不同，它竟然采用的是每股 135 美元的“固定价格发行”（fixed price issuance）。这一特立独行的定价方式，立即为这场注定载入史册的重磅 IPO 奠定了不同寻常的基调。 ### [20:37](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1237s) 非典型的 SpaceX IPO 与散户驱动的价格发现 在传统的 IPO 流程中，承销投行通常会带着公司拜访机构投资者、摸底市场意向，并进行建簿（book building）。投行随后会根据订单簿给出一个建议发行价，公司确认后完成上市定价。然而，SpaceX 的 IPO 完全打破了这一惯例：公司直接告知投行，发行定价为每股 135 美元，对应估值高达 1.7 万亿至 1.8 万亿美元。投行在此交易中毫无议价空间，面对埃隆·马斯克（Elon Musk）只能全盘接受，争相挂名于招股书封面。 这导致本次 IPO 省略了传统的价格发现环节，直接把定价抛给市场。机构投资者在 80 美元、90 美元或 100 美元（对应 1.2 万亿至 1.5 万亿美元估值）时会是非常积极的买家；而 1.8 万亿美元的估值公认属于过度拉伸。因此，真正的价格发现不会发生在发行阶段，而是完全由散户在上市首日的首次撮合成交及后续交易中敲定。 - **传统流程**：投行路演 -> 机构建簿 -> 协商确定发行价 - **SpaceX 模式**：公司直接敲定 135 美元 / 股（1.7~1.8 万亿美元估值）-> 投行照单全收 - **机构心理价位**：80~100 美元（1.2~1.5 万亿美元）视为舒适买入区间 - **核心特征**：发行环节无价格发现，首日交易完全由散户情绪主导 --- ### [21:51](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1311s) 首日涨幅风向标：对照 1997-1998 与 1999 年的泡沫量尺 佛罗里达大学的杰伊·里特（Jay Ritter）教授对历史 IPO 首日回报率有着深入研究。历史数据显示，在泡沫酝酿期的 1997 年和 1998 年，科技股 IPO 的平均首日涨幅分别为 14% 和 22%。若按此常态区间计算，SpaceX 在 135 美元发行价基础上的首日收盘价应在 154 至 166 美元之间，属于正常、温和的科技股首日表现。如果仅达到去年的平均 IPO 表现，股价大概落在 174 美元左右。 然而，在狂热的 1999 年网络泡沫顶峰时期，IPO 平均首日涨幅飙升到了 71%。乔什·布朗（Josh Brown）调侃自己当年买遍了所有 IPO，并且全是二级市场上追高充当“接盘流动性（exit liquidity）”。如果 SpaceX 在首日收盘冲上 231 美元（即 71% 的涨幅），鉴于其 1.8 万亿美元的庞大体量，这就明确标志着市场已经完全陷入 1999 年式的疯狂泡沫。周五上市首日的溢价幅度将成为衡量当前 AI 繁荣究竟处于 1997/1998 年还是 1999 年的最关键试金石。 ```mermaid flowchart TD accTitle: SpaceX IPO首日收盘价对应的历史时期参照 accDescr: 展示SpaceX基于135美元发行价的不同收盘价所对应的市场狂热程度及历史泡沫时期 subgraph Pricing["发行基础定价"] Base["SpaceX 发行价: $135"] end subgraph Normal["理性与常态区间"] R97_98["1997-1998年平均涨幅 (14%~22%)"] Price_Normal["首日收盘: $154 - $166<br/>温和中枢科技股表现"] LastYear["去年平均表现 -> 首日收盘: $174"] end subgraph Bubble["极端泡沫区间"] R99["1999年网络泡沫平均涨幅 (+71%)"] Price_99["首日收盘: $231<br/>确凿无疑的 1999 级别泡沫"] end Base --> R97_98 --> Price_Normal Base --> LastYear Base --> R99 --> Price_99 ``` --- ### [23:49](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1429s) 特斯拉与 SpaceX：马斯克资产的近期与远期期权 SpaceX 的上市终于为特斯拉（Tesla）提供了一个直接的估值锚点。长期以来，作为马斯克旗下的核心上市公司，特斯拉因独特的溢价而缺乏可比对象。如今特斯拉市值在 1.5 万亿美元左右，而 SpaceX 以 1.8 万亿美元估值登场，两者规模旗鼓相当。关于两家公司未来是否合并的猜测不断，但从证券合规角度来看，由于承销投行对诉讼风险极度敏感，投行在尽职调查时会彻底审查董事会会议记录；若存在任何实质性合并探讨，必须依法披露在 S-1 招股说明书中。 从公司基本面来看，短期内不太可能出现合并。历史上马斯克通常只在旗下某家公司陷入危机时才会推动合并，但未来潜藏着另一种逻辑协同：特斯拉在 24 至 36 个月内能否凭借机器人业务创造海量自由现金流将见分晓；而 SpaceX 目前最欠缺的就是现金流，上市融资正是为了解决巨额资本开支。本质上，特斯拉好比马斯克愿景的“近期期权”（2~3 年兑现），而 SpaceX 则属于“远期期权”（跨越 5 年、10 年甚至 15 年）。对投资者而言，对待这两只股票的逻辑通常非黑即白：要么全部配置以买入马斯克的长远愿景，要么因过高估值和特殊治理结构而全盘回避。 --- ### [26:01](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1561s) 纳斯达克快速纳指规则与非典型解禁节奏 在指数被动买盘与流通盘供应方面，SpaceX 面临着极为独特的供求对冲格局： 一方面，纳斯达克针对超大型 IPO 调整了快速纳入规则（fast entry rule），SpaceX 预计在上市约 15 天后便被快速纳入纳斯达克 100 指数（QQQ 等对应指数）。以 750 亿美元的流通市值计算，指数被动配置需求预计将带来 150 亿至 200 亿美元的强劲买盘，占比高达约 20%。 另一方面，该交易打破了传统 IPO 惯用的 180 天内部人禁售期（前 6 个月市场零新增解禁），设计了罕见的分批快速解禁机制： - 扣除马斯克个人持股后，20% 的股票将在第二季度财报公布后立刻解禁； - 若届时股价较发行价涨幅超过 30%，将额外追加解禁 10%； - 另有 70% 的股票将在少于 180 天的若干个时间节点陆续解禁； - 在第三季度财报公布后还将有 28% 解禁。 这意味着解禁抛压将在上市前几个月加速涌现，从而在一定程度上对冲指数被动买盘。但由于供需释放存在时间错配，上市最初几周内股价将出现剧烈波动。 --- ### [27:15](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1635s) 市场微观结构风险与 2012 年 Facebook IPO 的前车之鉴 上市首日还需警惕市场微观结构层面被天量交易击垮的风险。2012 年 5 月 Facebook（Meta）在纳斯达克上市时，开盘价从 40 美元仅冲高到 42 美元维持了 10 秒便迅速回落至 40 美元，由于挂单成交量远超负荷，系统故障导致投资者数小时内无法确认订单是否成交，甚至连实时价格都难以计算。叠加其后承销团分析师选择性披露公司在移动端变现缺乏进展的丑闻，严重摧毁了市场信任，致使 Facebook 股价一路深跌至 12 到 18 美元区间。 类似的大规模系统承压现象在 2015 年 8 月的 ETF 闪崩中也曾上演。尽管纳斯达克当前大概率已进行了严苛的压力测试，但面对散户极具爆发力的涌入，系统发生突发故障的风险依然不可忽视。在进入下一个讨论前，主持人打趣询问嘉宾杰西卡·拉贝（Jessica Rabe）是否想从 SpaceX 猎鹰 9 号火箭上跳伞；杰西卡笑称自己有一位同样热爱跳伞的闺蜜正是 SpaceX 的工程师，内部人脉让大家充满期待。最后，讨论转向核心复盘：回顾疯狂的 1999 年到底发生了什么。 ### [30:07](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1807s) 拿当下对比1999：人类的直觉反应与图表真相 谈到科技股的剧烈走势，投资者往往身不由己地产生历史联想。人类面对未知与剧烈波动时，第一反应往往是追问：“这像1999年吗？哪一部分有1999年的影子？又在哪些方面完全不同？”现场打了个生动的比方：如果一个人住在休眠火山脚下，火山突然苏醒冒烟，居民的第一念头必定是“庞贝古城要重演了吗？”这就是人类本能的思维模式。为了理清情绪与现实的边界，Nick Colas 调出了一张展示1999年至2000年底纳斯达克指数走势的图表，用最原始的数据客观复盘那段真正的狂热期。 ```mermaid flowchart LR accTitle: 1999年至2000年纳斯达克极端波动走势 accDescr: 展示1999年Q4至2000年4月纳斯达克指数经历暴涨84%与崩跌34%的过程。 subgraph Rally ["史诗级暴涨（不到半年）"] Q4["1999年Q4<br/>上涨 37%"] Q1["2000年年初至3月10日<br/>上涨 24%"] Peak["累计复利涨幅：84%<br/>（接近翻倍）"] Q4 --> Q1 --> Peak end subgraph Crash ["价格发现崩溃（5周断崖）"] Peak --> Fall["2000年3月10日 - 4月14日<br/>重挫 34%"] end ``` 复盘数据揭示出令人难以置信的极端行情：1999年第四季度，纳斯达克在短短一个季度内就飙升了37%；进入2000年，从1月初到3月10日触顶，又继续上涨了24%。由于收益是复利计算的，从1999年第四季度初到2000年3月的历史高点，纳斯达克在不到半年的时间里累计暴涨了84%，规模几乎翻了一倍。Nick 直言，如今的市场“完全不存在这种景象，绝无可能（period, full stop）”。 - 1999年第四季度纳斯达克单季涨幅：37% - 2000年1月至3月10日峰值涨幅：24% - 不到6个月内指数累计复合涨幅：84% - 时代背景背景：自1995年起纳斯达克年年收涨，叠加千禧年（Y2K）末日与狂欢交织的氛围，垃圾公司扎堆上市首日即暴涨 ### [31:46](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1906s) 泡沫的真正本质：暴涨之后的价格发现崩溃 真正界定泡沫的，不仅是上涨的斜率，还有随之而来的崩塌。图表的下半段记录了更惊心动魄的一幕：从2000年3月10日见顶到4月14日，纳斯达克在短短5周内暴跌了34%。尽管当时有投资者自我安慰“相比1999年四季度初依然是赚的”，但在如此短的时间内出现向上84%、向下34%的剧烈震荡，清晰地宣告了市场“价格发现机制（Price Discovery）已经彻底崩溃”。 当一个资产类别能在6个月内暴涨80%以上，紧接着在1个多月内抹去三成，说明所有人都在追逐纯粹的动量，市场上已经没有任何基本面锚点。真正的泡沫并非科技股上涨20%或30%，而是“先暴涨50%以上，然后骤跌30%”。Josh Brown 指出，如今在个股或特定极端概念（比如微小市值的量子计算概念股）里或许能看到这种走势，但在如今由庞大万亿美元巨头构成的纳斯达克100指数中，由于体量过于庞大，市场上根本没有足够的边际美元能驱动整个巨头指数在6个月内暴涨84%。Jessica Rabe 补充提出一条简洁直观的经验法则：“在纳斯达克综合指数中，翻倍才算泡沫（A double is a bubble）”；而眼下纳指过去一年的涨幅约为34%，距离当年的狂热程度相去甚远。 - 2000年崩盘斜率：3月10日至4月14日仅用5周时间重挫34% - 泡沫判定法则：价格发现机制失灵，由纯粹动量主导而失去基本面锚定，呈现“急速暴涨紧接着断崖式暴跌” - 经验法则：“翻倍才算泡沫”（A double is a bubble） - 当下环境对比：过去一年纳指上涨约34%，不仅基本面支撑强劲得多，且绝对体量与波动率的极端程度均无法与1999年同日而语 ### [33:43](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=2023s) 结语与嘉宾信息推荐 Josh 总结指出，当前市场虽然在情绪上偶尔让人隐约觉得有那个时代的“回声”，但底层的基本面差异巨大，市场表现出的极端程度也远远未达到当年的级别。在结束本次讨论前，Josh 向听众推荐了 Nick Colas 与 Jessica Rabe 创办的宏观与市场研究机构 DataTrek Research，鼓励大家通过节目简介中的链接关注他们的 YouTube 频道并订阅其官网每日研报。三人互道致谢并相约近期再次连线交流。 ## Notable quotes > [02:00](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=120s) 在统计学上纯属未可知之境（genuinely unknowable at this point because we are in truly uncharted statistical territory） — Jessica Rabe > [10:57](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=657s) 两者完全是两码事（apples and oranges），1999 年的疯狂与暴涨幅度远比今天剧烈得多。 — Nick Colas > [13:58](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=838s) 纯粹因为半导体板块走弱，科技股大概也就跌个 3%，但宏观因素的共振彻底放大了这场灾难。 — Jessica Rabe > [17:48](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1068s) 面对一个偏离度高达 6 个标准差（6 standard deviation move）的历史级事件，我绝不会拍拍屁股假装无事发生，然后说一切还会照旧。 — Nick Colas > [17:48](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1068s) 继续玩下去当然可以，但是，你得时刻紧盯着路口，随时防备下一辆汽车撞过来。 — Nick Colas > [19:45](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1185s) SpaceX 的上市交易架构非常怪异（a very odd deal），与传统 IPO 普遍采用的浮动价格区间询价不同，它竟然采用的是每股 135 美元的固定价格发行。 — Nick Colas > [30:07](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1807s) 如今的市场完全不存在这种景象，绝无可能（period, full stop）。 — Nick Colas > [31:46](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1906s) 在纳斯达克综合指数中，翻倍才算泡沫（A double is a bubble）。 — Jessica Rabe > [31:46](https://www.youtube.com/watch?v=ymc4q-oMEPo&t=1906s) 真正的泡沫并非科技股上涨20%或30%，而是先暴涨50%以上，然后骤跌30%。 — Nick Colas ## Mentioned WisdomTree, WisdomTree Japan Opportunities Fund (OPPJ), 沃伦·巴菲特 (Warren Buffett), Josh Brown, Nick Colas, Jessica Rabe, DataTrek Research, 标普 500 指数 (S&P 500), 科技精选行业 SPDR 基金 (XLK), VanEck 半导体 ETF (SMH), iShares 扩展科技软件 ETF (IGV), 微软 (Microsoft), 甲骨文 (Oracle), CrowdStrike, Palo Alto Networks, SAC Capital, 美光 (Micron), 闪迪 (SanDisk), 美联储 (Federal Reserve), 高盛 (Goldman Sachs), Kevin Warsh, Anthropic, SpaceX, 埃隆·马斯克 (Elon Musk), 特斯拉 (Tesla), 杰伊·里特 (Jay Ritter), 纳斯达克 (Nasdaq), Invesco QQQ Trust (QQQ), Meta / Facebook, DeepSeek ## Video description On this episode of What Did We Learn, ⁠Josh Brown⁠, ⁠Nick Colas and Jessica Rabe⁠ examine one of the most extreme stretches of tech outperformance in recent market history and discuss what it could mean for investors going forward. They break down the growing gap between Semiconductors and Software, the role of rising interest rates, and whether leadership is becoming too concentrated. They also take a closer look at SpaceX and why its highly anticipated IPO could be one of the most important market events in years. The crew also revisits the lessons of the dot-com boom, exploring what really happened in 1999, how investors knew the market had gone too far, and what today's environment can teach us about spotting excess before the tide turns. This episode is sponsored by WisdomTree. To learn more about learn more about OPPJ and the broader suite of geopolitical opportunity ETFs WisdomTree offers, visit: ⁠https://www.wisdomtree.com/us/strategies/geopolitical-opportunities @NickColasandJessicaRabe ►00:00 - Intro ►02:00 - A Major Tech Event Just Happened ►19:45 - SpaceX IPO ►29:46 - What Really Happened in 1999   👕 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 #spacex [Source](https://www.youtube.com/watch?v=ymc4q-oMEPo)

[2:00] Jessica Rabe: 在统计学上纯属未可知之境（genuinely unknowable at this point because we are in truly uncharted statistical territory） [10:57] Nick Colas: 两者完全是两码事（apples and oranges），1999 年的疯狂与暴涨幅度远比今天剧烈得多。 [13:58] Jessica Rabe: 纯粹因为半导体板块走弱，科技股大概也就跌个 3%，但宏观因素的共振彻底放大了这场灾难。 [17:48] Nick Colas: 面对一个偏离度高达 6 个标准差（6 standard deviation move）的历史级事件，我绝不会拍拍屁股假装无事发生，然后说一切还会照旧。 [17:48] Nick Colas: 继续玩下去当然可以，但是，你得时刻紧盯着路口，随时防备下一辆汽车撞过来。 [19:45] Nick Colas: SpaceX 的上市交易架构非常怪异（a very odd deal），与传统 IPO 普遍采用的浮动价格区间询价不同，它竟然采用的是每股 135 美元的固定价格发行。 [30:07] Nick Colas: 如今的市场完全不存在这种景象，绝无可能（period, full stop）。 [31:46] Jessica Rabe: 在纳斯达克综合指数中，翻倍才算泡沫（A double is a bubble）。 [31:46] Nick Colas: 真正的泡沫并非科技股上涨20%或30%，而是先暴涨50%以上，然后骤跌30%。

[14:00] When leadership becomes this concentrated and relative performance reaches these kind of extremes, the next move is often rotation rather than continued acceleration. [21:51] Nick Colas: If SpaceX closes day one [at] 231, we are legit in a '99 bubble. Particularly with a deal of this size. [29:46] Nick Colas: Tell me that tech goes up 50% and then down 30% and then I'll start thinking about a bubble. Because a bubble means prices have to go up and then down.

[0:13] Today's episode is [music] brought to [0:14] you by WisdomTree. [0:16] WisdomTree believes Japan is entering a [0:18] new era. Corporate reforms and stronger [0:21] shareholder policies are [music] [0:23] changing the game and investors like [0:26] Warren Buffett have taken notice. [0:28] The WisdomTree Japan Opportunities Fund, [0:30] ticker is designed to invest in Japanese [0:36] companies positioned [music] to benefit [0:38] from macroeconomic policies, industrial [0:40] innovation, and shifts in trade [music] [0:43] and consumer behavior. See the link in [0:45] the show notes to learn more about OPPJ [0:49] and the broader [music] suite of [0:50] geopolitical opportunity ETFs. [0:54] Ladies and gentlemen, [0:56] is that a good start? I know I've done [0:58] that before. All right. [1:00] Ladies and gentlemen, welcome to an all [1:01] new edition of What Did We Learn? On [1:03] today's show, we're going [music] to [1:05] answer one of the biggest questions [1:07] facing the stock market today. [1:09] And that is [1:12] was the epic technology mini crash of [1:15] Friday [1:17] the start of something bigger? [1:20] Or maybe something [music] we'll just [1:21] look back on as a blip. [1:25] I got two really smart people here to [1:26] help me [1:27] >> [music] [1:27] >> with uh the [singing] answer to that [1:28] question. Joining me once again, Nick [1:30] Colas and Jessica Rabe, co-founders of [1:32] DataTrek Research [1:34] and the authors of DataTrek's Morning [1:37] Briefing newsletter, [music] which goes [1:38] out daily to over 1,500 institutional [1:43] and retail clients. Nick and Jessica [1:45] also have their own YouTube channel, [1:47] which you can find a link to in the [1:49] description below. Welcome back, guys. [1:51] So good to see you. Happy summer. How's [1:54] everything? [1:55] >> Happy to be back. Thank you for having [1:56] us. [1:57] >> All right. Uh [1:59] let's get right into it. Jessica, you [2:01] pointed out that US large cap tech [2:03] stocks just outperformed [2:06] to the most statistically extreme degree [2:10] in the last decade. You note tech beat [2:12] the S&P by 29.3 percentage points over [2:16] the prior 50 days as of June 2nd. [2:21] That is a six standard deviation event. [2:24] Let's put your chart up and tell us [2:26] what's going on here. [2:27] >> Yeah, that is wild. Like you just said, [2:30] US large cap tech stocks just [2:31] outperformed by the most statistically [2:33] extreme degree in the last decade, [2:35] including after every crisis. So, we [2:38] thought we'd start by just discussing [2:40] what happened and see what's next for [2:42] this market-leading sector. For So, this [2:44] first chart shows the relative 50-day [2:47] price returns between the S&P 500 tech [2:49] sector using the XLK ETF as our proxy [2:53] and the S&P from 2015 to the present. [2:56] So, when the blue line's above or below [2:58] the x-axis, tech has out or [3:00] underperformed the S&P by the number of [3:02] percentage points shown on the y-axis. [3:05] Now, since 2015, uh [3:07] the tech sector has outperformed the S&P [3:09] by an average of 1.6 percentage points [3:12] over any given 50-day holding period. [3:15] The standard deviation around that mean [3:16] is four points. So, we noted the six [3:19] devi- the six standard deviation upside [3:21] level you just mentioned um of 25.6 [3:24] points with the dotted red line. You'll [3:27] see on the right side of the chart that [3:28] tech just beat the S&P by 29 percentage [3:32] points over the prior 50 days on June [3:35] 2nd. That's over a six standard [3:37] deviation event and the most extreme [3:40] reading in our data set by a wide [3:41] margin. And what's even more amazing [3:43] about that is that 50 days before that [3:46] gain was still six days before the late [3:49] March lows. So, as for what this means, [3:52] I have three three quick points. [3:54] >> Well, I'm sorry. Let's Let's Let's [3:56] double-click on that. [3:58] So, that is inclusive of the biggest [4:00] sell-off of the year so far. [4:02] >> Yes. And so, and that which actually [4:05] leads into my first point really well [4:07] because the first is that there's no hit [4:09] recent historical playbook for what tech [4:11] has just done. This move dwarfs every [4:14] prior stress and recovery period [4:16] recovery episode, including the 2020 [4:19] pandemic crisis, 2022's rate shock, [4:22] 2025's trade policy shock, and repeated [4:25] AI sentiment scares. None produced a [4:27] relief rally of this magnitude. The [4:29] March 2025 lows help explain the start [4:32] of the move, but not at scale. [4:35] So, [4:36] second, the algorithms that underpin the [4:39] US stock markets already knew this math, [4:42] that tech was very overextended heading [4:44] into Friday's sell-off. It was a 47th of [4:47] those 50th trading day off the March [4:49] 30th lows. So, market-making algos will [4:51] likely remain cautious and amplify any [4:54] further selling in tech names. And [4:57] third, any any trader with a good risk [5:00] management process will likely see this [5:02] math as [clears throat] a compelling [5:04] reason to be cautious on tech. Whether [5:06] Friday was a pause or the start of a [5:09] major rotation out of the sector is [5:11] genuinely unknowable at this point [5:13] because we are in truly uncharted [5:16] statistical territory. [5:18] >> I think intuitively, though, [5:21] people who have been around for a long [5:22] time will probably ask you guys, or [5:25] maybe already have, okay, [5:28] this is so far off the charts that it is [5:30] unprecedented, and therefore the [5:33] reaction to this to the downside is [5:35] unknowable. [5:36] But I think people would say, like, all [5:38] right, [5:39] but mentally I can picture previous [5:41] extreme rallies in tech, [5:43] and I know that those [5:46] didn't necessarily have a nice [5:48] aftermath. [5:49] And like everybody will revert back to [5:51] 25 years ago, the bursting of the dot [5:54] com bubble. We're like we're saying that [5:56] this 50-day move, not the whole tech [6:00] rally, but this particular moment in [6:02] time is extreme even [6:05] versus that period of time. Any 50-day [6:07] period, let's say in 1998 or 1999. Like [6:11] we think it's up there with those? [6:12] >> No, it's still not comparable. [6:14] >> So we we we [6:16] we have that math in our final section [6:18] of today's video. So we'll get there. [6:20] But it's exactly our point. [6:21] >> That's a great point. Yeah, we'll get [6:23] there. I think I think what it what is [6:25] useful and I'll I'll this is a good what [6:28] seg- segway to my second point is let's [6:30] look at the two major subcomponents of [6:32] tech, namely semis and software. [6:35] And we have another chart for that as [6:37] well. So this compares the rolling [6:38] 50-day relative returns of US large cap [6:41] semiconductor and software stocks. We [6:43] use the VanEck semiconductor ETF SMH and [6:46] the iShares expanded tech software ETF [6:49] IGB as our proxies. When the blue line's [6:52] above zero, semis are outperforming [6:54] software. Now [snorts] historically, [6:56] semis do typically outperform software [6:58] and there's actually a really good [7:00] reason for that. Moore's law may may [7:02] have slowed down somewhat, but semi [7:04] compute per dollar still doubles fairly [7:06] quickly and that allows these companies [7:08] to leverage their products forever for [7:11] an ever-increasing range of use cases. [7:14] The long-run average is 2.7 percentage [7:17] points of outperformance in favor of [7:19] semis over any given 50-day stretch and [7:22] that edge has been increasing. But like [7:25] the tech chart that we just looked at, [7:27] as of June 3rd, semis have beaten [7:29] software by 44 points. That's four [7:32] standard deviations above the long run [7:34] Yeah, that's four standard deviations [7:37] above the long-run mean and once again, [7:39] the most extreme reading we've seen in [7:41] the last decade. Now of course, yes, [7:43] this move has not come out of nowhere. [7:45] Some semi fundamentals have been [7:47] exceptionally strong. Earning revisions [7:49] con- continue to move higher. Momentum [7:51] has been historic and investors remain [7:53] highly confident in the AI [7:55] infrastructure buildout. But, and you're [7:58] getting to this, Josh, markets are are [8:00] rarely linear. Historically, when [8:02] leadership becomes this concentrated and [8:04] relative performance reaches these kind [8:06] of extremes, the next move is often [8:08] rotation rather than continued [8:10] acceleration. So, for tactical [8:13] investors, that argues for reducing semi [8:15] exposure and looking more closely at [8:18] software here. And for longer-term [8:20] investors who still believe in the AI AI [8:22] story, we think an index weight approach [8:25] to the tech sector as a whole makes [8:27] sense because you maintain exposure to [8:29] the theme while reducing the [8:31] concentration risk that a semi reversion [8:34] would disproportionately punish. [8:37] >> I like that idea. Um I wanted to ask [8:39] you, it seems to me that in the last [8:41] couple of weeks, the accelerant [8:44] for tech outperformance versus the S&P [8:47] was actually because the software names [8:49] bounced. [8:51] And [8:51] >> Yes. [8:52] >> the money did not rotate out of semis to [8:54] buy them. [8:55] The software and semi names started to [8:57] rally together. [8:58] Not all software, like notably Microsoft [9:01] sort of sat things out, but Oracle came [9:03] back with a vengeance. That's a big [9:05] component in the IGV. [9:07] We also saw the cybersecurity stocks go [9:10] absolutely crazy. So, CrowdStrike, which [9:12] I own, Palo Alto, or like all In other [9:15] words, all of a sudden, this drag on [9:17] tech sort of reversed itself. [9:20] And then it was like, "Oh, wait a [9:22] minute. This is turning into a tech [9:23] rally for the ages. They're all going up [9:26] together." And that was sort of like [9:29] what went on right before Friday. [9:32] >> Yeah, and that and [9:34] again, that [9:35] that does lead us to our third point [9:37] where the pictures actually more [9:40] positive for the S&P [9:42] and we should see that as you do see it [9:44] starts to see these other groups [9:45] working. So, [snorts] [9:47] I thought we'd just zoom out and also do [9:49] one last third chart on the S&P. It [9:52] shows the S&P's rolling 50-day price [9:54] return since 2015. Unlike tech, the [9:57] broader market isn't exhibiting the same [9:59] level of excess. The S&P's average [10:01] 50-day return over the last decade is [10:04] 2.4%. [10:05] And as of June 1st, index was up 15.3% [10:08] over the prior 50 days. That's a two [10:10] standard deviation move. So, still [10:12] strong, but not nearly as extreme. And [10:14] what's interesting is what happened next [10:16] in the in the in prior instances. So, [10:18] since 2015, the S&P has first crossed [10:21] that two sigma threshold five other [10:24] times. And in every case, the market was [10:26] higher 50 trading days later. And the [10:28] average forward return was 7.3%. [10:31] Even the weakest outcome still gained [10:33] 2.6%. So, ultimately, price momentum is [10:37] a powerful factor in capital markets. [10:39] Tech may be stretched and semis may be [10:41] due for a pause, but history [10:43] history says the S&P itself still has [10:46] room to run from here. [10:48] >> Okay, so we're not against [10:50] two standard deviation S&P rallies. [10:54] Uh they don't it's not extreme enough [10:57] that the risks get out of control. And [11:01] to your to your point, most of the time [11:03] you sort of get a a happy outcome by [11:06] that level of confidence in the market [11:08] until the crosses over obviously into [11:10] something a little bit more exaggerated. [11:13] >> Correct. [11:14] >> Okay. Nick, what's your read on on [11:17] on those charts from Jessica? [11:19] >> I mean, the first one is just we started [11:21] it with it for a reason. It is an [11:22] amazing chart and it really does put [11:24] everything in perspective because you've [11:27] got this massive rally in tech like [11:29] literally you haven't seen in the last [11:30] decade. And we'll talk about it but I'm [11:32] always wary about 1999 comparisons. I [11:34] was in and it was I was at SAC in 1999 [11:37] and I I was thinking back in my head [11:39] like if you weren't on a desk somewhere [11:41] trading those stocks, don't try to [11:43] compare today to then cuz it's apples [11:45] and oranges. That was a much bigger [11:46] rally, a much crazier time. [11:49] But still that tech number, that tech [11:51] return number is out of this world and I [11:53] keep coming back to Jessica's market [11:54] structure point, uh which is that algos [11:58] that do all the sinking of the capital [12:00] markets in this country, they know that [12:02] math. They are built on that math. They [12:04] literally are built on 1500-day standard [12:06] deviation of daily returns. So, that's [12:08] one reason why I think you saw that [12:10] cascading sell-off on Friday because the [12:12] market making algos said, "Oh, I know [12:14] what these things have been. I'm not [12:15] stepping in front of them. I'm widening [12:17] out my bid-ask spread. I'm not going to [12:19] take any risk cuz I see this is a [12:20] rollover after a statistically very [12:23] anomalous period." [12:25] >> Right. So, uh was the origin [12:28] was the origin of that sell-off in [12:30] Korean stocks and then it carried over [12:33] into the memory trade here in the US or [12:37] was it the other way around where the [12:40] Sandisks and the Microns got hit first [12:42] and then that spilled over into Asia? I [12:45] sort of can't tell. [12:46] >> I see it yeah, I mean it's a great [12:48] point. I see it as [12:51] our market first and then South Korea, [12:53] but South Korea's up 100% and it looks [12:56] even crazier versus EM than tech does [12:59] versus the S&P. And so, you have the [13:01] same dynamic at work. All of a sudden [13:04] every statistically oriented trader [13:05] says, "I get what's going on. There's a [13:07] pullback." And South Korea's a [13:09] notoriously volatile market. I mean, [13:10] they're literally doing what they call [13:12] sidecars, which we'd call like trading [13:13] slowdowns every day the last couple of [13:15] trading days. [13:17] >> What does that What does that mean? Can [13:18] you explain Can you explain what that [13:19] is? [13:19] >> Yes. In every market there are circuit [13:22] breakers of various kinds where when [13:24] things get too volatile and there's a [13:25] statistical measure of what percentage [13:27] drawdown the market slows down or stops [13:29] for 10, 15, 20 minutes to reset and [13:31] allow buyers to come back in. [13:33] >> That's the exchange that's the exchange [13:35] rule. [13:35] >> Yes, exactly. [13:36] >> Okay. [13:36] >> So, it's the same kind of thing we have [13:38] here with circuit breakers. Those have [13:40] been getting hit like every day the last [13:41] 3 days in South Korea. [13:42] >> They call what do they call them? [13:44] >> Side cars. [13:45] >> Sa- I wonder why they call it that. [13:47] >> I don't know. [13:48] >> Okay. All right, me either. Okay. So, [13:51] there have been a a bunch of those [13:53] because the selling has gotten extreme [13:55] enough to [13:57] uh [13:57] require that. [13:58] >> and look in in fairness, you know, not [14:00] to not to drag too much into this, but [14:03] the jobs report on Friday and the [14:04] revisions to the last 2 months really [14:06] reshape the picture of the US labor [14:08] market from one that had been slowing a [14:10] lot last year to one that actually looks [14:11] pretty pretty healthy. [14:13] Um rate market wasn't was not happy with [14:16] that. 2 years hit a new 12-month high. [14:18] And you know, the the worries about a [14:19] Fed rate hike and potential Fed policy [14:22] mistakes and higher cost of capital, [14:25] that didn't help either. So, it was it [14:27] was more than just the semis. [14:29] >> Okay. Right. I think the rates story [14:33] sort of exacerbated what would have been [14:35] a sell-off either way. [14:37] This had like this you had people [14:39] sitting on huge paper profits that all [14:42] of a sudden started to go away. [14:44] >> Yeah. [14:45] >> Think you have a lot of retail in these [14:47] stocks now [14:48] and a lot of leverage and a lot of [14:50] options. So, I sort of think it would [14:52] have been an extreme day, [14:54] but then having the broader macro [14:57] related selling on top of that is [15:00] probably what made it look as extreme as [15:02] it ended up looking. What do you What do [15:04] you guys think of that idea? [15:05] >> That's fair. I mean, tech was down 6.66% [15:09] which kind of was a eye-popping number [15:11] for, you know, numerological reasons. [15:13] Probably would have been down 3% just on [15:15] the semis rollover, but the macro did [15:17] not help. [15:18] >> Do we have to take more pain in tech if [15:21] uh [15:22] if the projection for the next interest [15:24] rate move increasingly [15:26] looks like it'll be a hike rather than a [15:29] cut. Do does does more air have to come [15:32] out of that balloon or not necessarily [15:35] because the earnings growth continues to [15:38] to manifest? Like what what's your [15:41] what's your view on that? [15:42] >> Uh that's a great question. Yeah, higher [15:44] rates we think higher rates are only an [15:46] issue if they go up so quickly that [15:48] there's a rational fear of recession. So [15:50] if inflation continues to increase and [15:52] the Fed gets behind the curve, that's a [15:54] legitimate worry for tech as we saw in [15:56] 2022, especially as hyperscalers [15:59] ultimately have cyclical cash flows. [16:02] >> We were talking about insurance cuts as [16:04] recently as 6 months ago. [16:07] No one is talking about insurance cuts [16:08] anymore. [16:10] Goldman Sachs just removed any interest [16:12] rate cut from its year-end forecast. Um [16:17] we're going to hear from the FOMC on [16:18] June 16th, June 17th. [16:21] And uh that'll be the first Kevin Warsh [16:23] Fed-led meeting. [16:25] I think tech tech stock investors are [16:27] paying attention to that. [16:29] Um I don't think people are worried [16:30] about the AI CapEx theme [16:33] uh right this moment. It seems like we [16:35] just got all the affirmation we needed [16:38] not only from the earnings reports, but [16:40] just in the confidence of the private [16:42] markets to invest in um Anthropic [16:46] at the valuation they're investing in. [16:49] And [16:50] like the size of the TAM that SpaceX is [16:54] projecting for generative AI stuff. Like [16:57] I think everyone feels good that the [16:59] earnings will show up for the AI theme, [17:02] but the rates thing now all of a sudden [17:04] is like a new dimension that we weren't [17:07] really worried about a quarter or two [17:08] ago. [17:10] >> No, that's very true. And as Jessica [17:11] said, I mean rates only matter when they [17:13] go up so fast that the market can't keep [17:15] up. [17:16] >> rate [17:16] >> change. [17:17] >> Yeah, and look I mean the Friday again [17:19] was a two it was a one year high in [17:21] twos. Uh twos track fed funds futures [17:24] and sentiment about fed rate increases. [17:25] So when you get a new high in twos, you [17:27] think you know, the fed's going to have [17:29] to be raising rates a couple of times [17:30] and then I've got to worry about what [17:32] effect that has on the economy and we're [17:34] already very late in the cycle and if [17:36] you look at every if you look at a long [17:37] term cycle chart of two year yields, you [17:39] get a recession about a year or two [17:41] after a cycle high in twos. So when you [17:43] break out to a new high, you start [17:45] thinking oh, now I've got to worry about [17:46] a recession again. [17:48] >> I don't think what we saw on Friday is [17:50] enough to knock the investor enthusiasm [17:53] for AI completely off course. [17:56] And I know it's like unknowable, but [17:58] like [17:59] gun to your head based on all the data [18:01] that you guys have seen, do you think [18:03] that we'll look at what happened on [18:06] Friday as sort of like a a a tempest in [18:09] a teacup like the deep seek sell-off [18:12] uh in January or [18:14] do you like do you think like this [18:16] really bears watching the recovery from [18:19] it because it could end up being way [18:21] more meaningful. Like what what would [18:22] you what would you guess if you had to [18:24] take a guess right now? [18:27] >> I guess I'll go first. I will not walk [18:30] away from a six standard deviation move [18:32] and say oh well, I'm just going to say [18:34] things continue as they were. [18:36] Uh that's that is a something to say [18:38] look, pay attention. It's okay to be [18:40] bullish. I think it's okay to be [18:41] bullish, but let's not walk away from [18:43] that first chart because it [18:45] it's it's incredibly important. Let's [18:48] let's be a little cautious here. [18:49] >> So it's not a game on moment to you. [18:52] Like uh I don't know if you do you play [18:54] street hockey as a kid? We used to play [18:56] street hockey. [18:57] Um we didn't have an ice skating rink [18:59] nearby, so we would just put the pads [19:01] on, roller skate in the streets and we [19:03] put the goals [19:05] put the goals in the streets and then a [19:06] car would come. [19:08] So you had to take the the goal the net [19:11] >> All right. [19:11] >> and move it to the sidewalk, let the car [19:13] pass. The car would pass. All right, [19:15] game on. [19:16] Um and it was like nothing ever [19:18] happened. Like right back into it. And [19:20] you don't you don't necessarily think [19:21] that people should have that attitude [19:22] now. Just put the goals back in the [19:24] street. Let's keep playing. [19:26] >> I think it's okay to keep playing, but [19:28] you know, keep looking for the next car. [19:31] >> Yeah. Okay. [19:32] >> the six sigma thing is do not walk away [19:35] from that. It's important. [19:37] >> we're long-term we've always been [19:38] long-term bulls on on tech, but six [19:41] standard deviations is just absolutely [19:43] off the charts. [19:45] >> Yeah. All right. I want let's I [19:46] mentioned SpaceX. [19:48] Um [19:49] we think this comes public at the end of [19:51] this week possibly. That's the latest [19:53] talk that I've heard. [19:54] >> Yeah. [19:54] >> Uh okay. Uh [19:58] Nick, let's go to you on this. You've [20:00] been involved in a lot of IPOs. You've [20:02] certainly seen your share of companies [20:04] come public. Uh tell us what your [20:06] reaction is to not just the size and and [20:09] how much they're raising, but just I [20:11] guess the amount of clamor amongst [20:14] investors for the deal, the amount of [20:17] media attention. Like what What's your [20:19] What's your reaction? [20:21] >> Yeah, I pulled together a little [20:22] PowerPoint page, so let's just walk [20:23] through that cuz there's five points [20:25] that I think investors are going to need [20:26] to know. [20:27] The first one is alludes to what you [20:29] discussed, which is this is a very odd [20:31] deal. And it's a fixed price issuance. [20:33] It's $135 a share. It's not a [20:37] Typically speaking, when you market an [20:38] IPO and I marketed a ton of IPOs in the [20:40] '90s, you go around, you visit [20:41] institutional investors, you gauge their [20:43] interest, you build a book, the bankers [20:45] look at the book and say, "Okay, we [20:47] think the price should be X." The [20:48] company says, "Okay." And off you go. [20:51] Here, the company has said, "It is $135 [20:54] a share at 1.8 trillion, 1.7 trillion." [20:56] >> They told Sorry, they told the [20:58] underwriters, "This is the amount you're [21:00] This is the amount you're valuing the [21:02] company at." And the underwriters said, [21:04] "Yes, sir. Just put us on the on the [21:05] cover of the book." [21:07] >> Yeah. Well, you have no choice with Elon [21:09] and you have no choice on this deal, [21:10] period. Now, that means that you're not [21:13] going through the traditional book [21:14] building process. You are basically [21:16] saying, here it is, come one, come all, [21:18] this is the price. I can guarantee you [21:20] that institutional investors will be [21:21] very happy buyers of this stock at 80, [21:23] 90, 100, you know, a 1.2, 1.3, 1.5 [21:27] trillion dollar valuation. 1.8 is a [21:29] stretch and there's 50,000 ways to [21:31] assess that, but it's a stretch. I think [21:33] anybody would agree with that. But it [21:35] leaves retail investors basically [21:36] creating the day one valuation for [21:38] better or worse. Um and that's an [21:40] important thing to know. This is not [21:41] your standard IPO. Just in so many ways, [21:44] but that's an important one. Price [21:45] discovery is not happening in this IPO. [21:47] Price discovery is happening on the [21:49] first trade and thereafter. [21:51] >> What? [21:51] >> point to The second point to understand [21:53] is there are classic ranges for how much [21:55] an IPO should go up on the first day. [21:57] And I pulled together a couple of [21:58] numbers from Jay Ritter who's got all [22:00] this data at the University of Florida, [22:01] does fantastic work on IPO returns. The [22:04] average day one IPO gain in '97 and '98, [22:08] okay, going into the bubble, was 14 and [22:11] 22%. [22:12] That translates into a first day close [22:16] for SpaceX of 154 to 166. So, if you see [22:19] the thing close 154 to 150 166 on day [22:22] one, that's a pretty average, nice, [22:25] middle of the road tech stock first day [22:27] return. However, in '99 when things got [22:30] bananas, the average IPO first day [22:32] return was 71%. [22:35] >> I know cuz I bought them all. [22:38] And not at the IPO price. [22:40] >> Yeah, so you [22:41] >> there. I was there. [22:42] >> I hear you. So, if if [22:44] >> exit liquidity. [22:46] >> If if if SpaceX closes day one 231, we [22:50] are legit in a '99 bubble. Particularly [22:53] with a deal of this size. [22:54] >> That's such an interesting way of [22:56] looking at this. [22:57] Um but before before we get to that, [22:59] like [23:01] is it '98 or is it is it 97 for AI or is [23:05] it 99? [23:07] >> Yep. [23:07] >> Well, here are the average IPO returns. [23:09] You tell me. I love it. [23:11] >> Yeah, that's how you know. I mean, this [23:12] is from for anybody, regardless whether [23:15] you're investing in the thing or not, [23:16] this is the most important tell for [23:18] Friday's price action, full stop. [23:20] >> Okay. [23:21] >> If we only get last year's average IPO [23:23] return, it's 174. So, let's look and see [23:26] we'll all see on Friday where we're at. [23:28] We literally going to set the calendar [23:30] for where we are versus 1999 or 1997. [23:33] >> So, the premium to that 135. Okay. I [23:36] love I love it. That gives uh [23:39] that gives my compounders uh a [23:41] scorecard, something to something to uh [23:43] monitor for. [23:45] >> Yeah, just put that on a sticky uh [23:46] sticky next to your trading screen and [23:48] that's all you need to know. [23:49] Um a couple of other final points. [23:51] Tesla's got a Tesla finally has a [23:53] valuation comp for as long as Tesla's [23:55] been public, we've had no real comp to [23:57] it cuz it's a Musk company and who knows [23:59] how to value that. We now finally have [24:01] one and actually SpaceX is coming public [24:03] at that valuation, slightly more than [24:05] where Tesla was trading on Friday. So, [24:06] Tesla was 1.5 trillion, SpaceX is coming [24:09] 1.8. There's going to be questions about [24:11] merger. I can tell you from having [24:13] written a lot of S-1s, if there were [24:15] ever merger discussions, they would have [24:17] had to have been in the S-1 that was [24:19] filed for SpaceX. [24:20] Otherwise, you're going to get sued. [24:22] >> Who would [24:24] I don't think he cares about being sued. [24:26] I think they discuss it every single [24:28] day. [24:29] And it almost doesn't matter if anyone [24:31] discusses it. If he's discussing it in [24:33] his own brain, [24:34] that's as powerful as a board meeting uh [24:37] in my in my estimation. [24:39] >> you're right about Elon, but you're not [24:40] right about the bankers. Bankers care [24:42] about getting sued a lot. [24:44] >> Okay. [24:45] >> And they're they're on the cover of the [24:46] S-1. [24:47] >> discussions. [24:48] >> They would have gone through and read [24:50] the [24:51] um they would have read the board [24:53] meeting notes [24:54] to to their due diligence to make sure [24:56] this wasn't a thing. [24:58] >> I don't really think that he has an [25:00] interest in merging them until one of [25:02] them gets into trouble or both of them. [25:05] And then that's sort of like, well, I [25:08] always wanted to do this all along. [25:09] That's what prompted all of the mergers [25:11] that he's done with his own companies. [25:13] So I I agree with you. I don't think [25:15] that's going to happen near term. [25:16] >> It's historically true, but I would put [25:17] a little different spin on it. He might [25:19] want to merge Tesla with SpaceX when [25:21] Tesla's making a lot of free cash flow [25:23] from robots. [25:24] Because the one thing SpaceX is short is [25:26] cash flow. [25:28] To uh to to finish and that's literally [25:30] why they're going public to raise the [25:31] capital. So Tesla to my mind is like a [25:33] near dated Musk option. If the robot [25:35] thing works out, we'll know in 24 to 36 [25:37] months. [25:38] And then Tesla will trade on that. [25:40] SpaceX is a very long dated version of [25:42] Musk. It's 5, 10, 15 years. And so [25:46] there're going to be people who think [25:46] should I own one or the other? I think [25:48] you're either it's either both or none. [25:51] You either believe in Musk's vision and [25:52] you want a near dated version and a long [25:54] dated version or you say no, this is too [25:57] too highly valued and the governance is [25:58] too weird. I don't want any [26:01] Um in index inclusion, we should just [26:03] cover that briefly. [26:04] Um it's going to go into the queues 15 [26:07] days or so after um the IPO on a fast [26:10] entry rule which the Nasdaq just changed [26:12] for these big deals. [26:14] That's going to add roughly, let's call [26:17] it 15 to 20 billion dollars of demand. [26:20] On a 75 billion dollar float, so roughly [26:22] 20%. So a lot of demand. The thing [26:24] people aren't talking about is that the [26:26] lockups are shorter for this deal than [26:27] usual. Typically speaking with an IPO, [26:30] if you are an insider, you can't sell [26:32] for 180 days. So the first 6 months [26:34] there's no stock available. [26:36] That's different here. 20% of the stock [26:39] ex-Musk stake comes available right [26:41] after Q2 earnings. Another 10% becomes [26:44] available if the stock is more than 30% [26:46] higher than the IPO price going into the [26:48] earnings. Then another 70% becomes [26:50] available um at various days short of [26:54] 180 days and 28% becomes available after [26:57] Q3. So, there's a lot more stock coming [27:01] than just the usual IPO lockup, a lot [27:03] more. That's going to balance out the [27:05] index demand [27:07] but not right away. So, there'll be a [27:09] lot of volatility over the first couple [27:10] of weeks as the market tries to figure [27:12] out the index price value. [27:14] >> Okay. [27:15] >> Um, final point, which I just want to [27:17] touch on briefly. I was around when [27:19] Facebook went public and it was very [27:21] difficult because the volume kind of [27:23] overwhelmed the NASDAQ and the market [27:26] makers. I'm sure the NASDAQ has stress [27:28] tested their system, but this is going [27:30] to be a very big deal with a lot of [27:31] retail demand. So, there is always some [27:34] risk of a market structure hiccup on [27:36] Friday just because of what's going on [27:39] with this one stock. So, that's [27:40] something to be watchful for as well. [27:42] >> I remember watching the Facebook IPO in [27:45] 2012. It was May of 2012. [27:48] And it was NASDAQ, not New York Stock [27:50] Exchange. Although it had two letters in [27:53] the ticker symbol, which confused some [27:54] people. [27:55] Um, [27:56] came public at 40, got to 42 for 10 [27:59] seconds. [28:00] >> Yep. [28:02] >> It It sold off to 40. [28:04] A lot of people were like, "Where Where [28:06] is the pop? Like, why isn't it doing [28:08] anything? Why isn't it" The volume was [28:10] off the charts. I remember there was [28:12] some issue at the NASDAQ. You probably [28:14] know better than I do. They had some [28:16] issue like identifying the current price [28:20] in the moment because they were [28:22] unaccustomed to having a new issue with [28:24] that many shares trading all at once. [28:26] >> Yep. [28:26] >> And then, uh, [28:28] ultimately, I think it fell to 12 or 18 [28:31] before [28:33] before it really found its footing and [28:35] its actual shareholder base. [28:37] >> Yeah, there were two problems. The first [28:38] was market structure on day one. People [28:40] didn't know if they got filled on their [28:41] orders for hours. So, you can imagine [28:43] what that was like. [28:45] >> Right. second [28:45] >> was that they gave selective disclosure [28:47] to some of the analysts on the deal [28:49] about um their lack of mobile traction. [28:52] And when that came out, it really eroded [28:53] trust in the company. That's why the [28:55] stock got to 12. So, there's a market [28:57] structure on day one. Again, I don't [28:59] think this is going to happen to SpaceX, [29:00] but it's worth just bringing up. Let's [29:02] Let's watch the trading action on Friday [29:04] cuz it's going to be a deluge of action. [29:06] And that historically has been sometimes [29:08] difficult. We saw it also with the ETF [29:10] flash crash in August 2015. So, huge [29:12] volume sometimes overwhelms the system [29:14] and let's just be on the lookout for [29:16] that. [29:16] >> Okay. Before we move to our last topic, [29:18] Jessica, uh anything on SpaceX IPO? [29:22] Would you like to jump out of a SpaceX [29:24] Falcon 9 rocket? [29:26] Uh what do what do you [29:27] What do we What are we thinking? [29:29] >> I have a My one of my best friends, [29:32] she's also skydiver, is an engineer at [29:34] SpaceX. [29:35] >> Okay. [29:36] >> So, yeah. [29:37] >> So, they they must be pretty excited. [29:39] >> I maybe [29:40] Maybe I have a little inside connection [29:42] there. We'll see what we can make [29:43] happen. [29:44] >> Yeah, yeah. All right. [29:46] Um we're going to we're going to end [29:47] with what really happened in 1999. [29:50] Um this is [29:52] this is I think a really important [29:54] um [29:55] sort of comment because [29:58] anytime tech stocks are rallying as [30:00] furiously as they've rallied [30:03] and then have like a massive sell-off, [30:05] it's the first thing like we almost [30:07] can't help it as human beings. [30:09] It's just the first thing that we go [30:11] back to. [30:12] Like, is it '99? [30:14] What aspects of this feel like 1999? [30:17] How is it not 1999? It's just uh it's [30:20] it's hard. Because Listen, if you live [30:23] in a town and there's a dormant volcano [30:26] that all of a sudden springs back to [30:28] life, [30:29] is this Pompeii? Like, this is the way [30:31] we think. We're humans. [30:34] Yes. But But talk to us. Show us the [30:36] chart. [30:37] >> Okay. This is a chart of the Nasdaq, [30:39] very simple. The Nasdaq from 1999 [30:41] through the end of 2000. [30:43] And I've noted a couple of numbers and [30:45] I'll just read them out. In Q4 of 1999, [30:49] the NASDAQ was up 37% in a quarter. [30:52] In from January through the March 10th [30:55] highs, it was up another 24%. [30:58] Those numbers don't add, they actually [31:00] compound. And so from the beginning of [31:02] Q4 '99 to the top in March 2000, the [31:05] NASDAQ was up 84%, almost a double in [31:09] less than half a year. [31:11] We don't have anything like that today, [31:13] period, full stop. Because let's not [31:15] forget the NASDAQ had been up every year [31:16] since '95. It was this [31:19] amazing rally. And I remember it [31:21] vividly. You just couldn't under And [31:23] this all came around Y2K as well. So [31:26] there was just this weird kind of [31:29] millennium kind of vibe combined with [31:31] this amazing rally in tech. The [31:33] crappiest names were going public and [31:35] you saw the IPO pops the first day. It [31:37] was absolutely insane. But an 84% gain [31:41] over less than 6 months is [31:44] not to be believed, literally. And that [31:46] came through in the second part of the [31:48] chart, which honestly is just as [31:49] important, is from March 10th to April [31:52] 14th, the NASDAQ fell 34%. [31:55] Now, it was still up versus the [31:56] beginning of Q4, that's how people [31:58] rationalized it. But when you see this [32:00] kind of volatility both to the upside [32:02] but just as importantly to the downside, [32:04] a 30% correction in a matter of 5 weeks, [32:07] you know price discovery is broken. You [32:09] know the market no longer has any [32:11] fundamental underpinnings because [32:12] there's no reason why any asset class is [32:15] up 80% over the course of 6 months and [32:18] then falls 30%. That means people are [32:21] just chasing momentum, nobody has a [32:23] fundamental anchor, and that's when you [32:24] start getting concerned. But the [32:26] numbers, the volatility numbers are much [32:28] higher than what we're seeing even like [32:30] on Friday. So to me the message to this [32:32] is, [32:33] you know, don't talk to me about a 20% [32:35] gain in tech or 30% gain in tech. Tell [32:37] me that tech goes up 50% and then down [32:40] 30% and then I'll start thinking about a [32:42] bubble. Because a bubble means prices [32:45] have to go up and then down. [32:46] >> Right, you might have that in individual [32:47] stocks [32:48] or little corners of tech, um, tiny [32:52] market cap things like quantum stocks, [32:54] but like to get that in the modern [32:57] Nasdaq 100, an 84% 6-month rally, I [33:02] almost don't think you can because the [33:03] dollar amounts of the largest market cap [33:06] companies are too large. Like I don't [33:08] think the dollars exist for that. [33:10] >> Mhm. [33:10] >> Well, you you could even just use our [33:12] simple rule of thumb in comparison to [33:14] 99, a double is a bubble in the Nasdaq [33:17] composite that happened at the height of [33:19] the dot com bubble. We're up like 34% [33:22] over the last year, we're nowhere close. [33:25] >> Yeah, it's not it's not even right. It [33:27] feel there's some [33:29] there are some aspects of the of this [33:31] where it feels like it's an echo of that [33:34] era, but the fundamentals are very [33:37] different and the extremities are not [33:40] extreme enough [33:41] to to the the current right. All right, [33:43] I it's a great place to leave it. I want [33:45] to tell people where they can find out [33:47] more from from you guys. Guys, if you [33:49] are into Nick and Jessica's research and [33:51] I know you are, make sure you click the [33:54] link in the show notes, um, so you can [33:56] find their YouTube channel and of course [33:59] you can become a subscriber yourself to [34:00] DataTrek Research and it's really easy [34:02] to find. datatrekresearch.com [34:06] uh, and you can uh, get a delivery of [34:09] Nick and Jessica [music] every day if [34:12] uh, if if you so choose. Guys, thank you [34:14] so much for joining me. [34:15] >> Thank you. [34:16] >> summer. I know we're going to check in [34:19] uh, soon. So, we will be back with what [34:22] did we learn? Thank you guys for [34:24] watching. Thanks for listening. [34:27] >> [music] [34:30] >> I'm a I'm