I’ve spoken at some length recently about the juxtaposition between low index volatility and extremely elevated single-stock vol, a conjuncture one popular strategist described as “manna from heaven” for the dispersion trade.

I took readers on a visual tour of that trade last week. If your modus operandi’s selling index vol to fund longs in single-name vol, this is a really great environment.

The other side of “much dispersion” (to channel the imagined internal monologue of Kabosu) is low correlation. And a corollary of low correlation is suppressed index vol.

Goldman’s Ben Snider brought this up in his latest. “Three-month implied average stock correlation [fell last] week to a record low,” he noted. “As a result, implied volatility for the average S&P 500 stock of 40% now registers 2.8x the level of implied volatility for the index, also a record.”

The figure on the left shows you the record low for implied correlation. The figure on the right the record high for the ratio of single-name vol to index. If you were to overlay modeled dispersion PnL on the right-hand cart, you’d get a decent fit.

With that as the backdrop, imagine a scenario where you’re in the high-vol names, long, short or both, and you’re long index vol as a generic hedge against a broad-based, “nowhere to hide” selloff. You’d have a problem.

“The single-stock vol is bonkers [which is] why vol dispersion traders have been printing it the past few months, but the risk in these AI high-flyer single-names and themes is that index downside hedges aren’t helping for those dying in the momentum unwind,” Nomura’s Charlie McElligott remarked.

That unwind’s acute commensurate with crowding and leverage. Most of what’s been working stopped working of late, including and especially the semi trade, and that played all sorts of havoc in the long/short factor space.

The figure above shows you the spike in realized vol for Goldman’s Momentum factor which, as Snider explained, stating the obvious, “largely reflects the AI trade, with tilts towards semiconductors and tech hardware stocks and away from software firms.”

At a glance, the spike circled in red (my annotation) may not look existential. To understand the problem, you have to picture it without recessions. Simply put: Momentum factor volatility is the highest in at least half a century excluding NBER-designated downturns.

That’s way too much “chop” and if it continues, the unwind will too because as Snider went on to point out, data from Goldman’s prime desk suggests hedge fund positioning’s “still very elevated relative to the last few years,” even as it’s off the highs. “Extreme volatility,” he cautioned, “encourages further position downsizing, creating a vicious cycle.”

The table above, from McElligott, gives you a granular look at what’s going on. What’s worked over the past year isn’t working lately and vice versa.

“The AI unwind has been causing short-term performance carnage for investors, because they’re all in the same stuff,” he wrote. “There’s just so much leverage applied in these concentrated single-name longs and shorts, and [it’s] piggybacked by retail in singles, options and leveraged ETFs.”

Again, you can’t have too much daily “chop” in a setup like that. “Our Momentum factor moved by more than 2% on four of five days [last] week,” Snider mused, obliquely underscoring the point.

Charlie doesn’t do oblique or otherwise dance around the issue. “The daily two-way velocity in single-names and crowded themes where both longs and shorts are blowing through risk budgets makes too much sense unfortunately, as bunches of the biggest names in global equities are 100-vol assets and that means mechanical de-grossing,” he wrote. “In theory, running market-neutral with low / no net should help you during a de-risking selloff, but during a violent momentum unwind, the reality is that ‘your gross is your net.'”

That brings us full circle. Thanks to record low correlation and the accompanying index vol suppression, longs in index vol (or SPX puts) aren’t going to be much help unless and until everything sells off at once, an outcome that’s hard to get during earnings season when the “micro” tends to trump the macro.