JPM Head Quant Explains How The Algos Traded The Brexit Crash, And Why He Sees "Elevated Risk"

When looking at the kneejerk devastation in the aftermath of the Brexit vote, JPM's head quant Marko Kolanovic said that he expects up to $300 billion in program selling, coupled with 5-10% in near-term downside to the S&P500. While Kolanovic was correct about quant and technical fund flows, what he likely did not factor in was the dramatic crisis response by central bankers who have now made it very clear their only mandate is to keep global equity markets disconnected from reality and artificially bid higher no matter the cost.

So what does he think happens now that the S&P has wiped out all losses from Brexit in the past three days?

Here is his explanation, released moments ago:

Flows and Price Action—Largely a Repeat of August 2015

In our note last week we discussed how the impact of Brexit would likely be similar to August 2015. Market price action and flows observed so far this week are fully consistent with the moves that followed August 24. The Figure below (left) shows S&P 500 moves over the four days starting with the “crash” day (i.e., August 24, 2015, and June 24, 2016). The “crash” day itself both witnessed futures hitting limit down during pre-market hours and a significant move on the day itself (-3.9% on 8/24 vs. -3.6% on 6/24). The following day’s move was again lower, largely driven by flows from convex strategies (e.g., CTA outflows, derivatives hedging). The bounce-back that followed on days 3 and 4 were also similar in August and this week (in fact, the market rallied more on days 3 and 4 in August 2015). We would like to point out that both in August and now, market realized volatility reset significantly higher, and market outflows from various “VAR-based” investors (volatility targeting, risk parity, etc.) followed in the days and weeks ahead. This contributed to the market bottoming only weeks after the crash (in 2015, the market bottomed on 9/28).

In the past few days we have heard various arguments about how market action this week was substantially different from August 2015. This was argued based on the perception of lower volume orderly move, market bouncing back (which was identical in August), and supposed pension fund inflows that propped up the market this time around. Below we will address each of these.

Total US share volume on 6/24 (18.6 bn shares) was higher than on 8/24 (18.3 bn). This was helped by the Russell rebalance, which provided extra liquidity (e.g., market depth dropped by ~50% as opposed to the ~70% drop on 8/24). Futures volumes on 6/24 were the highest since 8/24. On both days futures hit their 5% limit down pre-market, and the end-of-day moves were similar in size. The fact that the 6/24 move was not larger than the 8/24 move can be largely attributed to the lower S&P 500 option gamma imbalance as compared to 8/24 (over the 6/24 move gamma imbalance averaged ~15bn per 1% vs. ~30bn for the 8/24 move, which was an alltime high as we pointed out before the crash). As gamma exposure turned significantly short on Monday 6/27, it also contributed to a larger squeeze up on Tuesday and Wednesday (please note, these were even more prominent on 8/26 and 8/27 than this week).

How about pension fund quarterly inflows? We have come across what we consider wild estimates of pension fund buying on account of quarterly rebalances. A number of clients told us how they are hesitant to sell equities until these alleged flows are out of the way. The figure above (right) shows the quarter-end effect of equal weight portfolio rebalances (% of quarterly to day move that is reversed in the last week of a quarter). The chart shows that this effect largely disappeared over the past roughly two years. This means that quarterly rebalancing funds that allocate based on fixed weights (which would drive mean reversion) are likely similar in size to those that rebalance based on fixed risk budgets (which would generally do the opposite), washing out any quarter-end effect. It should not be surprising that pension funds are moving away from fixed weight rebalances as these were the worst performing strategies over the past two decades (see our Primer on Systematic Strategies, page 105). We have extensively analyzed in the past another effect—the “month-end reversion effect” (see our report here). We argued that the month-end effect is increasingly related to the option expiry cycle rather than pension fund rebalances (this can be shown by isolating month-ends that did not coincide with post-option expiry weeks). The effect was obviously present and even more significant in the August 2015 (and January 2016) market declines, so there is nothing different about it now. The expected market impact of the month-end effect this week was about ~40bps of market upside and can hence explain only a small fraction of the market move higher post the crash. In order of importance, in our view, the drivers of the midweek rally were the snapback of oversold (European) markets and short S&P 500 gamma squeeze, and then the lesser drivers of S&P 500 price momentum turning neutral (from negative), month-end-effect, unwind of hedges, and short covering. As most of these were present in August, too, they don’t change our view on higher realized volatility and expected outflows from VAR-based investors.

One notable difference between the August crash and Brexit impact on US markets is the reaction of implied volatility, e.g., the VIX. The August 2015 crash was largely unexpected by the market, and hence the VIX increased from ~13 in the week before the crash to ~40 on the day of the crash. The Brexit risk was fairly anticipated with significant buying of VIX products (e.g., VIX ETP exposure reached near record levels in the weeks before the event). As a result, the VIX moved to ~19 in the week before Brexit and increased only to ~26 on the event. As the VIX was pricing in a large move in either direction (for the methodology see appendix of this report), once the event passed, its contribution was mechanically priced out of the VIX (e.g., think of it as a ~6 point drag). In addition, many investors rushed to monetize VIX hedges, which resulted in large outflows from long VIX ETPs and closing out of VIX call positions (as well as new shorting of VIX ETPs and opening of VIX put positions). Overall, VIX ETP exposure dropped by around half. We maintain the view that we have not yet seen the highs of VIX due to Brexit and  related risks (increase of market realized volatility, upcoming earnings season, and geopolitical consequences including post Brexit shift in US election polls).

Fund Performance and Positioning: Trend followers (CTAs) were the best performing strategies on the day of the Brexit result. Due to long bond and gold exposure, and low equity exposure, these funds returned on average ~2.3%. As we argued in our previous report (see here), Hedge Funds were going into Brexit fairly long equities. Equity long-short funds drew down ~2% (consistent with our estimation of their equity beta of 0.45), and Risk Parity funds drew down on average ~1% as their long bond exposure was not sufficient to offset high levels of equity exposure. Despite the very strong performance of long-short equity Momentum (3%) and Low volatility (~3%) factors, Equity Quant Hedge Funds were still down on the day (~0.6%) given the likely net long equity bias and ~1.5% pullback of the Value factor. Interestingly, the equity exposure of long-short Hedge Funds (and Hedge Funds in general) did not materially decrease since Brexit. Discretionary Hedge Funds likely decided to avoid selling into more volatile/less liquid markets. We have also noticed investors adjusting trading and prepositioning for various end-of-day hedging flows (Hedgers are adjusting accordingly, as one could for instance notice the moves of S&P 500 and VIX futures after the 4pm market close on Friday and Monday as opposed to the more common 3:30-4PM hedging window). Finally, we would like to point out that the size of all equity short positions calculated after Brexit is near the lowest point since September 2015 (i.e., very little shorts currently in place).

The points discussed above suggest equity markets face elevated risk in the days and weeks ahead.