The index looks calm. Underneath it, sector correlation just hit its lowest level since September 2000. We ran 27 years of data to ask the question everyone is asking: does this mean top?
Something strange is happening in this market. The S&P 500 is up about 10% over the past quarter, sits barely 1% off its all-time high, and day to day it feels almost sleepy. Realized volatility is running under 15%, close to its long-run average. If you only looked at the index, you would call this a quiet, healthy bull market.
Underneath, it is anything but quiet. The same ninety days that produced that placid index contained a semiconductor up 112% (AMD) and an IT-services giant down 25% (Accenture). In June, the gap between winners and losers among the largest US stocks reached the 92nd percentile of the last 28 years. And the violence has not been one story. It has come in acts.
The year opened with a hard rotation out of mega-cap tech and expensive software. By late February, Intuit had lost 38%, Salesforce, Oracle and Adobe were each down about 25%, and Microsoft had shed 19%, while the money showed up in the old economy: Lockheed, Deere, FedEx and Caterpillar all gained 30% or more. Then the war in Iran sent energy vertical, the sector peaking in late March up 41% on the year as inflation fears worked their way back into the conversation, and the index finally cracked, an 8% air pocket into early April. The ceasefire and a strong first-quarter earnings season flipped the script entirely: semiconductors rose 40% in April alone, and tech put in its best forty sessions of the year, up 45% from early April into June. That leg is where the quarter’s eye-popping numbers come from, AMD’s 112% and Intel’s 63% among them.
Which brings us to the last twenty sessions, where the rotation has turned inward, on tech itself. Semiconductors have gone flat since mid-June while the index added 3%, and the splits inside the group are violent: AMD up 11%, Qualcomm down 10%. Enterprise software is being sold outright, with Oracle down 34%, IBM down 24% and Accenture down 18% in a month. The mega-cap AI platforms are still holding their bid, Meta up 15% and Nvidia up 6%, while the tape’s new leaders are financials, utilities, industrials and healthcare. Money is not leaving this market. It is sprinting from one part of it to another, and lately it has been sprinting away from the very leadership that carried the spring rally.
How does the index stay quiet through all of that? Correlation. We measure how much the nine major sector ETFs move together over a rolling month. The average pairwise correlation right now is 0.105. The median since 1999 is 0.55. Today’s reading sits in the 1st percentile of everything we have measured, and in 27 years of data the only stretch lower was September 2000.
That is the mechanical answer to why the market feels quiet while portfolios feel whipsawed. Enormous moves are happening every day and they are cancelling each other out at the index level. This is a rotation regime, and 2026 has been defined by it: 84 of 130 trading sessions this year have met our rotation criteria, and the current run is the longest in our sample.
Before asking what this means, one distinction matters. High dispersion by itself is ambiguous, because it happens in two opposite worlds.
When dispersion is high and correlation is also high, everything is moving violently and moving together. That is crisis tape: October 2008, March 2020. When dispersion is high and correlation is low, big moves are netting out. Money is leaving one part of the market and entering another at unusual speed, but it is staying in the market. That is rotation, and it is where we are now, at close to the most extreme reading on record.
We identified every episode since 1999 in which sector correlation fell into its bottom 5% versus the trailing three years. There are twenty of them. The typical episode lasted about a month. Here is what followed.
Most resolved uneventfully. The median next quarter was +3.4%, almost exactly the market’s normal drift, and the prior trend was still intact three months later 79% of the time. Episodes in 2006, 2014, 2016, 2017, 2019, 2024 and 2025 all passed without much drama. Getting defensive purely because correlation was low would have been a losing habit.
The tail is a different story. The chance of a quarter worse than minus 8% was 21% after these episodes, against 5.4% on ordinary uptrend days. That is roughly four times the normal tail risk. And the two generational tops in the sample both launched from exactly this configuration: August 2000, which was followed by an 11.6% down quarter and the start of the dot-com bear, and November 2021, which was followed by an 8% down quarter, a 15% down half-year, and the January 2022 top.
So the honest summary reads like this: the base case is trend continuation, and the exception is not a small one. It is the big one.
It is too early to tell, and we say that as a measurement, not a hedge. In the data, big tops and aggressive-but-healthy rotations start from the same place. They only separate later, and the separation shows up in identifiable ways. Three things distinguish the episodes that became tops from the ones that resolved higher:
One caution on the other side: not every top announces itself this way. The 2007 top arrived through a different door, choppy consolidation with high dispersion and high correlation, everything falling together in waves. And that version came with a loud accompaniment: credit stress. Through the back half of 2007, while equities chopped within a few percent of their highs, Treasuries returned nearly 10% and high-yield bonds returned nothing at all, an enormous flight-to-quality gap opening in plain sight as the banking system strained. The bond market was pricing trouble months before the stock market accepted it.
That ingredient is absent today, and it is one of the first things we check every morning in our regime tracker. High-yield spreads sit at 2.72%, in the tightest decile of the past three years and about a dozen basis points from the lowest print of that stretch. Investment-grade spreads are similarly quiet at 0.79%. More telling than the level is the behavior: through the entire inward rotation of the past month, with software names repricing 20 and 30% and sector leadership changing hands, high-yield spreads moved one basis point. Credit is confirming none of the equity market’s internal violence. Pristine credit is not immunity, but it removes the specific accelerant that turned 2007’s choppy top into 2008, and it is a tell we would expect to fire early if this rotation were something worse. Which is why we track credit alongside both equity configurations rather than falling in love with one signal.
And one thing that is not a warning sign, even though it feels like one: leadership change. Through these regimes, the stocks and sectors that led going in usually do not lead coming out. That churn is a property of the regime, not evidence of breakage. It is already well underway: semiconductors have stalled since mid-June after their vertical spring, software is being repriced name by name, and financials, utilities and industrials have taken over the last month’s leadership.
Here is the number we think matters most for anyone running real money through this. The roughly one hundred large-cap US stocks we track are currently averaging about 35% annualized volatility, which is genuinely hot. The index prints under 15% only because correlation of 0.13 is netting those moves against each other. Hold every stock’s volatility exactly where it is, let correlation simply return to its historical median of 0.28, and the same portfolio produces about 1.4 times the volatility. No new shock required. No news event needed. Just the netting going away.
This is what makes air pockets in these regimes feel sudden. The catalyst does not need to be large when the netting is this stretched, because correlation mean-reverts, and when it does, index volatility jumps mechanically. Anyone sizing positions or leverage off the trailing volatility of a decorrelated market is carrying more risk than the number on the screen suggests.
The correlation statistics will not decide which way this resolves. Fundamentals and macro will. Q2 earnings are the nearest catalyst, and in a market this dispersed they matter twice over, because rotation regimes reprice single names violently on results. Iran already moved this market once this year, and ceasefire or not it remains a live variable for energy prices, inflation expectations, and the risk premium investors demand. And Fed policy sits over all of it, since the path of rates will determine whether the leadership rotating into favor can hold its bid.
Any of those can be the thing that either re-correlates this market to the downside or confirms the rotation as a durable change in leadership within an ongoing trend. We are not going to pretend to know which. What we can say is what the tape has told us so far: this is one of the most aggressive rotations of the last three decades, the index trend is intact, the historical base case from here is continuation, and the tail is fat enough to respect.
Blackworks Capital runs systematic strategies, generally trend-following, so the systems do not trade this view; the systems trade the tape. For now the tape says the trend is up: we are near all-time highs, no technical, macro, or fundamental signal has broken, and the systems are positioned accordingly.
What an environment this extreme changes is not the positioning but the level of scrutiny. First-percentile conditions are where a strategy’s assumptions get tested, so we are watching our systems closely, and we are watching the two tells history points to: a three-month trend rolling over while correlation stays floored, or correlation snapping higher while dispersion stays extreme. Today the readings say trend intact, credit quiet, rotation among the most extreme on record. The data will show what this is before any opinion does.
DISCLAIMER
Any description or information involving investment processes or allocations is provided for illustrative purposes only and does not constitute investment advice nor an offer or solicitation to subscribe for any security or interest. Any statements regarding correlations or other similar behaviors constitute only subjective views, are based upon reasonable expectations or beliefs, and should not be relied on. All statements herein are subject to change due to a variety of factors including fluctuating market conditions and involve inherent risks and uncertainties both generic and specific, many of which cannot be predicted or quantified and are beyond Blackworks Capital's control. Future evidence and actual results or performance could differ materially from the information set forth in, contemplated by or underlying the statements herein. Blackworks Capital accepts no liability for any inaccurate, incomplete or omitted information of any kind or any losses caused by using this information. Blackworks Capital does not give any representation or warranty as to the reliability or accuracy of the information contained in this document.