Wall Street’s ‘fear gauge’ might be lying to you about last week’s market turmoil #NewsMarket

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Here, via BNP Paribas, is a fun chart from last week (high-res mercy link for mobile users):

Based on the CBOE VIX’s intraday peak of 65.73, the market event that has been branded the “Summer Selloff”…

…was, as wags have been keen to point out, apparently one of the most significant volatility events to have ever hit US stocks.

So was Monday August 5th really an event on par with the Covid-19 crash, the heights of the Notorious GFC, or Black Monday?

Maybe not.

In a note titled “Did VIX Really Hit 65 on Monday?”, published last week, Academy Securities strategist Peter Tchir unpicked some of the odd dynamics underpinning movement in the volatility measure often labelled Wall Street’s Fear Gauge. He wrote:

Without a doubt, the official VIX calculation reported a level of 65.73 on Monday at 8:34am ET. That level was posted and is now being used by many people to justify long positions in equities. The theory seems to be that we had an “epic” spike in volatility indicating panic, and that panic has since receded – hence creating a buying opportunity. Notice, that we chose to use the word volatility here, partly because many seem to use VIX and Vol interchangeably – which is not accurate.

In any case, normally we’d just leave this alone, but the fact that so many people are taking comfort in the “fact” (quotations used to indicate it isn’t really a fact) that we had a vol spike and it is over, makes me incredibly nervous.

Tchir’s beef with the reading — prompted by a reading of its methodology, and conversations with its creator Robert E. Whaley — comes down to three main factors, all of them wonky (his emphasis):

— The calculation uses the entire option chain – including very illiquid options, which seemed odd to us as it allows some largely irrelevant options to skew the entire VIX calculation.

— Increases in the bid/offer spread impacted the calculation. It would be one thing to use traded prices, but since most of the deep out of the money contracts (lottery tickets) rarely trade, the calculation can use a midpoint (I think it is more complex than that, but think that is close enough for the argument we are making). So, a deep out of the money option, quoted 1 cent on the bid side and 1 cent of the offer side and never traded, is counted. If, on a volatile day, the wages making the option prices (and most of the out of the money options are quoted by algos rather than humans for small size), decide to widen bid offer spread and now make the market 1 cent bid vs 5 cents offered, they just impacted the VIX significantly. PAUSE. It doesn’t take a traded value to move VIX. The seemingly “innocuous” effort for algos to avoid getting picked off (wider bid/offer spreads) on something that rarely trades, moves VIX. This is a big part of my issue with the 65 print on Monday morning.

…

There is one final piece to the VIX 65 puzzle that we have discussed here in the past. VIX only includes options expiring between 23 and 37 days. So, as the market has gravitated to 0DTE options, a much smaller percentage of total option trades involves trades that impact VIX. While daily and weekly options might be fun to punt around (they clearly are dominating in terms of flows), it left us concerned that when real hedging needs occur, and traders want longer dated options, the rush from zero day to vix eligible options could cause distortions.

Part of the evidence for something being off in the VIX is the spread versus VIX futures, says Tchir (who asks “So, I’m supposed to trust a calculation rather than a traded price?”):

These distortions, Tchir reckons, show there “was fear” but no “panic” last Monday — despite what the VIX highs might indicate. This, he suggests, is a bad sign for bulls:

There was no panic on Monday (nor should there have been).

As a current bear and a contrarian, it would have been nice to see panic.

People are saying there was panic and are buying the market based on that.

That scares the heck out of me! I’m going to stick with my argument that VIX futures and ETF flows tell the real story – some fear, and a decent amount of greed.

Other post-mortems stress the same point. The volatility spike “went above and beyond what any catalysts would imply,” said Rocky Fishman, a derivatives analyst at research group Asym 500 who thinks a sudden rush for short-dated S&P 500 protection probably exacerbated moves in the Vix.

Dispersion traders (anyone buying single stock volatility while selling index volatility to profit from the difference) who were caught offside also deserve some blame, Fishman wrote in a note published last week:

First, the spike in index option pricing around the market’s open was not matched by single stocks. Put options on the two largest SPX constituents, Apple and Microsoft, are typically far more expensive than SPX puts, but in the opening minutes of Monday’s trading September monthly 10% out-of-the-money SPX puts became more expensive than the comparable single stock puts. We have seen a similar differential with other single stocks.

Second, the time of day may have contributed. Prior to the market open, dispersion traders may have had their short index positions accrue mark-to-market losses, while the single stock market was closed. If this led them to cover short index vol positions, and then sell single stock option positions after the market opened, it could have contributed to a spike in index vol pre-open, and then a reduction in all vol levels (particularly single stock) once the market was fully open.

BNP Paribas equity derivatives strategist Bénédicte Lowe adds that vol-selling ETFs which had previously been keeping realised and implied market turbulence in check (indirectly contributing to the build up of leverage and extreme positioning by CTAs et al.) started enhancing both measures last Monday as stocks dropped.

In a note from April, Lowe’s colleagues had warned that a big enough market slump might lead to such a shift (“gamma overhang” being a fancy phrase for the massive growth of strategies shorting volatility):

If the Gamma overhang disappearing on the downside causes the market to move from a low to a high volatility regime, it could trigger a large volatility target outflow. This scenario has been absent in the recent correction, with SPX realized volatility remaining at low teen levels. Whilst our volatility target model shows we have already seen some outflows, these could be much more significant in a more volatile risk-off move. For outflows to accelerate, we would need to see a trigger and a large spike in realized vol.

This pretty much played out last week, BNP wrote on Friday:

[In April] we highlighted the massive growth in volatility selling ETFs and Mutual Funds. Our takeaway was that there was not an unwind risk from these trades, given they are mainly benign variants of covered call selling. However, we did suggest that this flow was depressing realized and implied volatility levels, which in turn builds leverage and risk. The gamma overhang for dealers from the selling flow is localized. We concluded that in the case of a rapid spot drawdown (similar to the one seen this week), the market would quickly flip from a regime of vol suppression to one in which volatility is much more reactive. We estimate that the dealer gamma profile turned negative at the start of the week, co-incident to the VIX highs.

The long gamma “overhang” from the volatility selling funds has been “almost entirely wiped out by the spot declines” over the past week or so, Lowe concludes.

But those positions will be rebuilt if markets stay relatively calm, depressing volatility on the way up but bottling up trouble all over again for the next crisis.

And so the cycle starts afresh.