Over one month this summer the ten largest stocks in the S&P 500 realized 45% volatility. The index that holds them realized about 40 points less, because the stocks were barely moving together at all.
The standard reading of that is diversification working. Dean Curnutt's reading is that the market has priced a correlation regime with no precedent as though it will last, and that the people selling correlation are doing it with no cushion left.
"Why are those stocks so uncorrelated? Boy, I just don't know. I wish I did, but I can say definitively what that's doing."
Curnutt started on a fixed-income research desk at Nomura in 1991, ran options through the financial crisis at Bank of America, and left in 2008 to found Macro Risk Advisors, an independent broker-dealer that does nothing but risk management and options.
The full interview is covered here so you can skip it. 54 minutes of audio, 23 minutes of reading.
Here are the 12 insights that matter.
👤 Guest: Dean Curnutt, CEO of Macro Risk Advisors, the independent broker-dealer he founded in 2008 to sell options and risk-management advice, and host of the Alpha Exchange podcast
🎙️ Host: Felix Jauvin, who hosts Forward Guidance, Blockworks' macro and markets podcast
📰 Published: 16 September 2026 on YouTube (Blockworks) · recorded 14 September 2026
🔴 YouTube | ⏱️ 54 min | ✅ Time saved: 32 min
Key Takeaways
Realized correlation among S&P 500 stocks is running at 5% to 15%, against 35% to 40% in a normal market and 75% to 90% in a crisis
There is no comparable period; the longest look-back that resembles it is about a year and a half
The top 10 stocks realized 45% volatility in a month when the index realized 40 points less
Microsoft's correlation to the other six of the Magnificent Seven was negative over that stretch
Bank desks sell a short-correlation product most investors never see, and Curnutt cannot rule out that it is causing the thing it is priced off
The dispersion trade is now being put on at levels with no margin of safety
Implied volatility is at the 10th to 15th percentile of its own history and still not cheap against what it costs to replicate
The VIX near 16.5 against one-month realized volatility of 9 is a wide spread, not a narrow one
A positive-carry hedge existed for a decade and stopped existing in 2022, when the long Treasury ETF fell 20% alongside a 19% fall in the S&P
$860B of new federal debt in four months, at 4% unemployment and a record stock market, has inverted the risk-off relationship
His claim is that the 10-year note, not the S&P, is now the source of risk
A 7% mortgage rate is restrictive and a 6% to 7% funding rate for Meta is not, which is why he thinks the Fed cannot slow this economy without breaking something
He expects the AI complex to prove highly correlated after the fact, and those stocks are more than 35% of the index
1. Three Decades of Vol
Jauvin opened by asking for the career path, and Curnutt gave a version organized around what each job taught him about mispriced risk.
He has been doing this for more than three decades, and describes the through-line as one question: whether uncertainty is correctly priced by derivatives. He started at Nomura on a fixed-income research desk in 1991, working for the chief economist at a point when the Fed's whole project was winning bond-market credibility and pulling the term premium down. He noted that the term premium is much higher now than a decade ago, and that there have been periods when it was higher still.
Two years of business school followed, and that is where he says the theory of option pricing took over. He was at Lehman Brothers in the late 1990s and at Bank of America from 2000 to 2008.
The founding insight for Macro Risk Advisors came out of watching the credit market and the equity market disagree in 2007. He spent his last year or so at Bank of America with people in credit, and by mid-2007 he was convinced they were seeing a world the equity market had not registered.
"And it kind of flies in the face of market efficiency to think that the credit market knows something that the equity market doesn't." His conclusion was not that one of them was wrong: "So, I would say that asset classes can respond to things at different speeds."
The second reason to leave was that the dealers had stopped being useful to their clients. "And I also say that by early 2008, it was very clear that the banks were kind of compromised in their ability to service their client base because they had legacy positions that they were just trying to defend and risk manage. They were in trades at very much the wrong prices."
He also thinks the crisis was mispriced rather than unforeseen. The housing bubble was widely discussed, and a few people — he named the ones from "The Big Short" — saw it. What investors were unprepared for was how the unwind showed up in option premiums, the VIX and credit spreads.
2. Correlation With No Comp
Jauvin set up the market as it stands: a much lower implied-correlation regime than the 2010s, when volatility ground lower and lower and the VIX printed below 10 in 2018.
"You hit the nail on the head that we live in this environment in which the correlation among stocks is we've never seen anything like it. There's no comp." He said you cannot go back five, ten or twenty years for a precedent — only one month, three months or a year, and the regime has been in place about eighteen months.
Realized correlation among S&P 500 stocks is running around 10% to 15%, dipping near zero and spiking only briefly. The two spikes he named are last year's tariff selloff, which faded when the worst-case tariffs were undone, and the early stage of this year's energy shock in March and April. For the most part, he put the range at 5% to 15%.
The historical benchmarks are what make the number strange. "In a benign market historically the correlation of stocks was probably 35 to 40%." In a crisis — LTCM, the financial crisis — "correlations on the order of 75 to 90%," and in the 2011 sovereign-debt and debt-ceiling episode he put it at 85% to 90%.
He explained what 85% to 90% looks like in prices. Take the top ten stocks of fifteen years ago across sectors: if each falls 1.5% to 2% on a day, the index falls about the same, because nothing offsets anything.
The measurement he built to show today's opposite is the sharpest number in the interview: "The one month realized volatility of the top 10 stocks in the S&P was 45%." Against the index's own realized volatility, "So it's basically a 40v spread of the single stocks to the index." He built the index on Bloomberg himself to calculate it.
The cause he cherry-picked to make the point is that Microsoft's correlation to the other six of the Magnificent Seven was negative over that stretch.
"It's a windfall for people that own the index." His own gloss on the metaphor: not only are the eggs not in one basket, they are not in the same neighborhood.
The risk is not the regime but the extrapolation. Low correlation is stealing volatility at the index level, and traders build positions on the assumption that three months to a year of behavior continues.
3. What Pushed It There
Jauvin asked what is actually causing it — the crowding of dispersion trades run by multi-manager hedge funds, or a fundamental change now that the Magnificent Seven are no longer the only trade in the market.
Curnutt said both are live questions and he cannot separate them. His own risk philosophy treats market risk as endogenous: created by the trades themselves.
His historical case for that is the run-up to the financial crisis. The growth of mortgage credit was itself part of what pushed credit spreads down, because derivative structures were being built whose delta hedge was to sell credit risk, and they kept getting bigger — a "tail wagging the dog sort of trade."
The institutional misreading he still marks against the authorities is the low VIX of that period: "And tragically in some ways, even the Fed, even the IMF, they misread the VIX of 10 in late 2006 as this sign of safety while it was everything that related to a sign of danger." The system of leverage was growing by selling volatility and credit spreads, and the carry it produced was reinvested into the same trade.
The modern version sits in a product set most investors never see. "There's a significant product set within bank desks called QIS, quantitative investment strategies." These are not listed options; they are complicated, large, packaged derivative exposures delivered to a pension fund or a hedge fund through a total return swap. A client can effectively flip a switch and be short correlation.
Clients of those products tell him that "short correlation lives and breathes" in most of them, and that "It's a carry trade that has done well." Whether the products are forcing realized correlation lower, or merely responding to it, he said is very hard to disentangle.
On the economic side, he tied low correlation to the stability of the data. Once the 2022 tightening cycle found its endpoint in early 2023, growth and economic statistics were stable, and that coincides historically with low volatility and reasonably low correlation.
He engaged with the AI-versus-non-AI explanation and found it insufficient. Even within AI the stocks are uncorrelated; even within AI there is a picks-and-shovels group and the hyperscalers, and he cited a Bank of America chart showing hyperscaler capital expenditure and free cash flow moving in opposite directions with Nvidia on the other side. He granted the observation and said "it still doesn’t explain the degree" of the readings he is seeing.
"So something structurally has changed. It's so hard to know. I just can't really explain it." What he says he does know is the consequence: carry has pushed the spread between single-stock and index volatility to all-time wides, "And it's forcing the folks chasing carry on the dispersion trade to put the trade on at levels that really have no margin of safety."
4. The XIV Lesson
Jauvin asked whether February 2018 — the day the inverse-volatility exchange-traded product XIV was wiped out — is the right analogy for a crowded short-volatility position today.
His answer was yes and no, and the "no" is about mechanics. The XIV structure was unique: about $3.5bn of capital across two products at the peak, a reaction function of twice the daily move, and a starting point of a front-month VIX future around 12.
"So the math was so unmistakable 12 to 18, right? 2x a 50% move is 100%." Everyone could see that a move from 12 to 18 in a day would zero it, and the number of VIX futures that would have to change hands on such a day was more than the market could handle on paper.
What kept people in it was the carry. The product bled lower through 2017 and delivered its holders a return above 50% that year, which is what set up the showdown.
The "yes" is that correlation carry and volatility carry are the same trade. The volatility risk premium, he said, is the basis of most investing: taking in carry from people who will pay it.
His framing of the whole business is insurance, and not only financial insurance. Geico, Allstate and the reinsurers — Swiss Re — are in longevity and mortality risk, which he described as effectively selling tail risk: you make money most of the time, the distribution is negatively skewed, and when it pays out it pays out a lot.
He is explicit that selling insurance is fine; the failure is in the sizing. "You also need critically to size your trade so that inevitably when that risk event does happen and you need a payout, you're still there to effectively reinvest at a higher VIX, right?" The point is to size it so that you "live to fight another day."
The people who got that right in 2020 were selling the VIX at 83, with the authorities on their side. His claim about crisis policy is blunt: once a real tail event forces a government response, its only objective is to get the VIX down. "It’s almost fighting a financial war not an economic war."
The best time to sell volatility is when it has peaked, which is precisely when most sellers no longer have the capital to do it.
5. Implied Follows Realized
The bridge from correlation to volatility is the part of the argument that makes the rest of it work, and Curnutt laid it out mechanically.
There is a correlation risk premium the way there is a volatility risk premium. Run implied correlation against subsequently realized correlation and the shape is the same: a consistent premium, punctuated by spikes in realized correlation that blow through the buffer the seller was paid.
The reason implied volatility tracks realized volatility is that the marginal price setter is not expressing a view. "The price setter is simply a Jane Street quant or a Citadel quant. They are reproducing the VIX by trading in the underlying."
That quant is trading the gamma of the S&P — the swings — so the level of the VIX is the cost of replicating it. When the underlying swings a lot, re-hedging up and down produces a lot of benefit and the VIX should be high. When realized volatility is as low as it has been, the VIX cannot easily rise, because the hedging portfolio is not producing anything.
"Implied correlation really can't be very high when realized correlation is this low." So implied correlation can be sold at unprecedented levels as long as realized correlation is lower still — which is exactly the condition today.
His summary of the state of play: the correlation carry trade is on and still working, but it is on at levels never seen, because realized correlation is at levels never seen.
6. Tail Funds Arrived Late
Jauvin turned to tail hedging itself. Curnutt started with where the term came from, and with a piece of allocator history he treats as a warning.
The phrase is relatively new and is an artifact of the financial crisis and, more than that, of 2011 — the sovereign crisis and the US debt-ceiling fight together. The lesson allocators drew was that the tail is not only a global financial crisis; other things can happen.
By 2012 the dedicated tail-risk fund existed, with a simple mandate: spend premium as an overlay on a long portfolio and prevent a repeat of 2011.
"You know what's interesting is that the growth of these came almost at exactly the wrong time." He described the institutional lag precisely: an event happens, chief investment officers of endowments and state-fund boards decide it must never happen again, approval takes six to nine months, a manager search takes another six, and the money lands in 2013 — the start of some of the lowest-volatility years on record.
The result is that dedicated tail-hedging funds are now largely gone as a standalone business.
What he thinks is left is a trade rather than a permanent allocation, and he was careful about what he is claiming: "I'm not big on the capacity to time markets. I think timing markets is very difficult." What he looks for instead is moments when the price of optionality is out of step with the set of risks actually present.
7. Past, Present, Future
He gave three tests for whether options are attractive, and answered all three for today's market.
The past is a percentile score, and the tool is arbitrary as long as you are consistent. He listed the cross-asset equivalents he watches: the MOVE index for Treasuries, the CVIX for currencies, volatility on the long-Treasury ETF, swaption volumes, and now a VIX for bitcoin.
The VIX's own range gives the scale: "So, we know the VIX has been between 9 and a half and 83. I'm not going to count the 145 it hit during the crash of 87." He dates the series from 1993.
Against the last five years, he put it at "like 15th percentile," and said three-month S&P volatility — a better measure than the short-dated VIX — sits in the same place. He warned that the look-back window decides the answer, so he uses several, one of which starts after the 2022 tightening cycle.
His blended measure is lower still. He publishes an index of five-year percentiles across five metrics — S&P volatility, VIX, long-Treasury ETF volatility, the currency VIX, and credit spreads with credit volatility — and "And you were in like the 10th percentile of these things."
The present is carry, and here the answer flips. A doomsday story does not move option prices: "You could tell me a incredible doomsday story. We just heard one this weekend. 10% chance of humanity's extinction. Boy, that doesn't sound good, right? The V markets aren't going to price that in." He added that 20% would not price it in either.
What the market prices is how the option carries, which is a function of realized volatility. He gave the numbers: the VIX around 16.5 to 17 against one-month realized volatility of 9. "That's a big spread. In fact you could argue that the VIX is too high relative to realized."
The verdict on two of the three: "So that's the past where implied V is low relative to its own history. The present says it's fair at best."
The future is the one he has no measurement for, but he said "the totality of uncertainty is significant right now." His phrasing was careful — not things that will go wrong, but "uncertainty inducers in markets" that can rear their head.
The same logic applies to Treasuries. Even with the Treasury Secretary in an argument with the bond market, "But if realized V continues to be relatively benign, people aren't going to bid up the price that much of Treasury bond options."
8. Insurance Is Never Free
Jauvin raised the question every long-volatility conversation reaches: how to hold the hedge without paying the carry, the problem Mark Spitznagel and others have spent years on. He asked whether the free hedge exists at all.
Curnutt's answer was an analogy, and he flagged it as sarcastic himself. A bank requires proof of homeowners insurance before it writes the mortgage; you call the insurer and they quote you nothing at all. "There's no way you could buy insurance for free. And it's the same in financial markets."
He allowed one real exception, from someone he had recently interviewed. Alec Litowitz, the founder of Magnetar, spotted a mispricing in 2006 between super-senior credit-default-swap protection and the equity tranche. "He literally did buy protection for free."
The contrast he drew was with the more famous version of the same trade. John Paulson bought super-senior protection and paid for it — at a price that turned out to be far too low, but he paid. "Alec literally found a way to in some ways buy a straddle and get paid in the process. That's incredibly rare."
The closest thing to a free hedge for ordinary portfolios was created by policy, and he blames Ben Bernanke for it. Post-crisis quantitative easing ran through 2011, 2012, 2013 and 2014 with the market doing well and the Fed still buying the country's bonds on the argument that core PCE at 1.5% had to reach 2%.
That suppressed rates and produced a strong negative correlation between the S&P 500 and long Treasuries. "So for years owning government bonds was a positive carry hedge." He put "hedge" in quotation marks himself, on the grounds that a hedge is a guaranteed payout and this was not one.
It stopped working in 2022, and he gave both numbers: the long-Treasury ETF fell about 20% and the S&P 500 fell 19% in the same drawdown. The conclusion he draws is the section's title: "So insurance you got to pay for."
Where he says the alpha actually is: in what you buy, a little in when you buy it, and — the part he stressed — in how you structure it.
9. Not a One-Day Trade
The second structuring question was systematic against discretionary: roll ten-delta VIX calls every month and accept the bleed, or wait for cheap moments and buy call spreads.
He answered with the conviction first: equity optionality and cross-asset volatility are "too low relative to the uncertainties," and he said he cannot time it.
His argument to allocators is about what they are protecting, not what they are predicting. "Look, if you've been longing the S&P, and most people have some version of that in their portfolio, you got a 22% compounded return for the past three and a half almost four years. For the past decade, you got a 15 plus% compounded return." That decade includes the COVID drawdown.
"To me, insurance is about losing money, right?" You root against your own flood insurance; you still want to own it at the right price.
The pricing case is the chain he had been building all interview: never-seen-before realized correlation holds down realized index volatility, and realized volatility holds down implied volatility through carry.
On the mechanics, he named what he owns: S&P volatility, VIX calls and VIX call spreads.
The systematic part is about not treating it as a single dated bet: "My call for doing it systematically was listen, just don't wake up one day and fire some premium at a trade. Put it on till further notice."
10. The Bond Is the Risk
Curnutt then walked through the risks he sees, starting by dismissing the category most people reach for first.
Geopolitics, on his reading, has been a poor trade for years. Russia, Ukraine, Iran, China — the market keeps going up through all of it.
The exception is when geopolitics reaches an asset that is inside the monetary-policy conversation, which is crude oil. "Crude is this asset that's got so many knock-on effects, plastics, fertilizer." He was careful to say he is not an economist and does not know the supply chain in detail, only that gasoline prices are up and it is feeding into inflation at the moment policy was trying to push the other way.
His larger concern is the long end of the Treasury curve, and the Treasury Secretary's handling of it. He called Scott Bessent brilliant and a hedge-fund manager, and then said: "I find his rhetoric recently to be incredibly unsettling."
The specific episode is the dismissal of Stanley Druckenmiller's op-ed, which Curnutt called "flippant at best." His objection: "This is the bond market telling you to be careful and you're telling us you have inside information."
The fiscal arithmetic under it is the reason the bond market is worried: "We're talking about it because in four months the US racked up $860 billion in new debt in four months. And that's in peace time with a 4% unemployment rate and a record stock market."
The conclusion is the inversion of the standard risk-off relationship. "We are in a completely inverted risk environment where the bond market is the source of risk to the stock market." And: "So, people say the S&P is the risk asset. I said the 10-year note is the risk asset and it's just not behaving well at a time when the politics of deficits are intractable."
On the Treasury Secretary he added that this is "unearned bravado is how I would frame it." He then made the same point against the other side of the trade: anyone betting on higher yields is betting against the house, and if they force an over-the-top intervention it would have lasting consequences for the integrity of pricing.
Jauvin noted the timing — the 10-year had just hit 5%, with the Fed meeting that week — and asked what the market wants from the meeting.
11. Cheap Credit for Meta
Curnutt knows the new Fed chair — Kevin Warsh has spoken at his events and appeared on his podcast — and his critique is about the limits of the job rather than the person.
His first complaint is communication. He thought the first press conference was fine and the second was not, on the grounds that a chair has to give the market some roadmap.
The claim he disputes is that inflation is a choice made by monetary policy. Curnutt's counter is the debt and the deficit: on that basis "he’s fighting with two hands behind his back" unless the fiscal side does something more credible.
What he says would be credible is a bipartisan deficit commission along Simpson-Bowles lines — and he immediately conceded it is not happening. "There's no space in US politics for deficit reduction."
He allowed the literal version of the claim and then dismissed it as unreasonable: the Fed could put the funds rate at 10% tomorrow and lower inflation. What he does not believe is that there is a reasonable set of actions available. "And I don't think the Fed has a reasonable set of things that it can do without some help from the fiscal side."
The conundrum he keeps returning to is that the same policy rate is restrictive for households and irrelevant to the largest borrowers. He agreed with Warsh that the mortgage market shows restrictiveness: "A 7% mortgage rate doesn't feel like a great deal right now. But 7% or 6% funding for Meta, that's a joke to Meta. That's the cheapest option Meta has bought in a long time."
His reason that cheap credit matters more than usual is what the borrowers are buying with it. The hyperscalers and the wider AI ecosystem are using credit to buy an option on capital spending with a very large potential payoff, which makes almost any cost of money look cheap.
That is the trap he identifies in the policy: "And so to slow inflation and to slow the economy maybe means slowing the capex trade," and doing that might require much higher rates from here.
12. The Correlation Event
The last section is where the market-structure argument and the macro argument meet, and it is the reason for the episode's title.
His forecast is about behavior, not direction: "I think we are setting up for a correlation event where they prove to be very correlated after the fact." The "they" is the hyperscalers and the AI ecosystem, which trade with almost no correlation to each other on a daily basis now.
The arithmetic is what makes it an index problem. Those top seven or eight stocks are more than 35% of the S&P 500, so if they fall together it goes straight into the index.
His phrase for the mispricing is that their correlations are "so unpriced" — the hedge is cheap precisely because the recent past says it is unnecessary.
That is why he owns S&P volatility and VIX call spreads rather than single-name protection.
He closed on the contradiction he finds most telling about the moment, recorded the weekend before the interview aired. The chief executive of Anthropic had called for the whole field to slow down, with a roughly $2tn listing ahead of it and others joining him: "We're about to see an IPO, a gigantic IPO, and the CEO is saying, I want to slow this company down."
Bonus Insights
Curnutt's summary of his own philosophy is that market risk is endogenous — created by the positions in the market rather than arriving from outside them. It is the link between his 2006 credit-derivatives example and his 2026 correlation example, and it is what makes him treat a quiet market as evidence rather than comfort.
Jauvin's framing of the change is worth keeping: for years the passive flywheel meant large market capitalization begat larger market capitalization and the Magnificent Seven were the only trade, and it now feels as though there are many more opportunities. Curnutt neither endorsed nor dismissed it; he said it is one of the two questions worth exploring and that he cannot answer it.
On the AI split he was willing to be beaten by the data. He accepts there is an AI component and a non-AI component to the index, and that within AI there is a picks-and-shovels group and a hyperscaler group. He simply does not think any of it accounts for a correlation reading of 5% to 15%.
Jauvin's closing plug was for Curnutt's own podcast, the Alpha Exchange, which he said he listens to regularly for the derivatives guests.
Curnutt's bottom line is that the cheapness of equity hedges and the danger in the market have the same cause: a correlation regime with no historical comparison is holding index volatility down, the trades that profit from it have no cushion left, and the shock most likely to end it is the one asset class that used to be the hedge.
Products, Companies & Tools Mentioned
Macro Risk Advisors (The independent broker-dealer he founded in 2008, focused on risk management and options)
Nomura, Lehman Brothers and Bank of America (His career path: a fixed-income research desk in 1991, the late 1990s, and 2000 to 2008)
Meta (His example of a borrower for whom a 6% to 7% funding rate is no constraint at all)
Microsoft and Nvidia (Microsoft's negative correlation to the rest of the Magnificent Seven; Nvidia as the picks-and-shovels side of the hyperscaler capital-spending chart)
Jane Street and Citadel Securities (The marginal price setters of implied volatility, replicating the VIX by trading the underlying)
Magnetar (Alec Litowitz's 2006 trade, which Curnutt says genuinely bought protection for nothing)
Geico, Allstate and Swiss Re (His analogy for the volatility risk premium: insurers selling tail risk that pays out rarely and heavily)
Anthropic (The chief executive calling for the field to slow down ahead of a very large listing, which Curnutt calls a unique fact pattern)
Bloomberg (Where he built the index that produced the 45% top-10 volatility reading)
Books & Resources Mentioned
Alpha Exchange (Curnutt's own podcast on derivatives and risk, and where the Magnetar interview ran)
The Big Short (Named for the handful of investors who saw the housing bubble's mechanics; his point is that most people did not see how it would show up in option premiums)
The MOVE index, the CVIX and the volatility of the long-Treasury ETF (The cross-asset volatility measures he percentile-scores alongside the VIX)
Watch the full episode:
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: