Last Friday, when the market was down only 2%, we presented readers with a note which promptly became the most read piece across Wall Street trading desks, which was written by JPM’s head quant Marko Kolanovic, who correctly calculated the option gamma hedging imbalance into the close, and just as correctly predicted the closing dump on Friday which according to many catalyzed Monday’s “limit down” open.
Given that the market is already down ~2%, we expect the market selloff to accelerate after 3:30PM into the close with peak hedging pressure ~3:45PM. The magnitude of the negative price impact could be ~30-60bps in the absence of any other fundamental buying or selling pressure into the close.
We bring it up because Kolanovic is out with another note, one which may be even more unpleasant for bulls who, looking at nothing but price action, were convinced that after the biggest two day market jump in history, the worst is behind us.
In the just released note, the head JPM quant warns that a large pool of assets controlled by price-insensitive managers including derivatives hedgers, Trend Following strategies (CTAs), Risk Parity portfolios and Volatility Managed strategies, which is programmatically trading equities regardless of underlying fundamentals, is about to start selling equities, “and will negatively affect market in coming days and weeks.” For good measure, he casually tosses the word “crash” in the note as well.
By way of reference, JPM notes that a good example of how price-insensitive sellers can cause market a disruption/crash is the price action on the US Monday open. It says that technical selling related to various hedging programs, in an environment of low (pre-market) liquidity indeed caused a ‘flash crash’ on Monday’s open. S&P 500 futures hit a 5% limit down preopen, and then a 7% limit low at 9:31 and 9:33. The inability of hedgers to short futures spilled over into large cap stocks that were still trading and could be used as a proxy hedge. Had it not been for the futures limit down event, the selloff would likely have been worse as indicated by the price of the index implied by individual stocks. The figure below shows the S&P 500 futures, SPY ETF and S&P 500 replicated from
the largest stocks that were trading near the market open.
Kolanovic correctly takes credit for his prediction and notes that “in our Friday note we forecasted end-of-the-day selling pressure due to option gamma hedging. We saw similar price impacts on Thursday, Friday, and Monday (pushing the market lower into the close) and an upside squeeze on Wednesday. Our estimate is that up to 20% of market volume was driven by hedging of various derivative exposures such as options, dynamic delta hedging programs, levered ETF stop loss orders, and other related products and strategies (note that levered ETFs have gamma exposure of only ~$1bn per 1%, i.e., much smaller than that of S&P 500 options). We estimate the cumulative selling pressure from options hedging during the market selloff to be ~$100bn. Options gamma is expected to remain substantially (in excess of $20bn) tilted towards puts while the S&P 500 is between 1850 and 2000.
The figure below shows Put-Call Gamma assuming current open interest and different spot prices. JPM expects high volatility to persist (should we stay in this price range) and cause quick intraday moves up or down, particularly towards the end of the trading day.
According to the quant, it is not only derivative hedgers who are pushing the market around like a toy with barely any resistance: :in fact, there is a much larger pool of assets that is programmatically trading equities regardless of underlying fundamentals.”
It is these investors who, “in the current environment” are selling equities and “will negatively impact the market over the coming days and weeks.”
Trend Following strategies (CTAs), Risk Parity portfolios, and Volatility Managed strategies all invest in equities based on past price performance and volatility. For instance, in our June market commentary we showed that if the equity indices fall 10%, these trend followers may need to subsequently sell ~$100bn of equity exposure. These types of ‘price insensitive’ flows are starting to materialize, and our goal is to estimate their likely size and timing. These technical flows are determined by algorithms and risk limits, and can hence push the market away from fundamentals.
This is where it gets scary for the bulls who thought we may be out of the woods, and that the crash was behind us. If Marko is right, as of this moment we are merely in the eye of the hurricane:
The obvious risk is if these technical flows outsize fundamental buyers. In the current environment of low liquidity, they may cause a market crash such as the one we saw at the US market open on Monday. We attempt to estimate the amount of these flows from three groups of investors: Trend Following strategies (CTA), Risk Parity portfolios, and Volatility Managed strategies. These investors follow different signals and have different rebalancing time frames. The time frame is important as it may give us an estimate of how much longer we may see selling pressure.
So, how much longer may we see the selling pressure?
1. Volatility Target (or Volatility Control) strategies provide the most immediate selling as a reaction to the increase in volatility. These strategies adjust equity leverage based on short-term realized volatility. Typical signals are 1-, 2-, or 3-month realized volatility. Volatility target products are provided by many dealers, index providers and asset managers. Volatility targeting strategies also became very popular with the insurance industry. After the 2008 financial crisis, many Variable Annuity (VA) providers moved from hedging their equity exposure with options to investing directly in volatility target indices (e.g., 10% volatility target S&P 500). It is estimated that VA issuers have ~$360bn in strategies that are managing volatility; some of these use options to manage tail risk, some buy low volatility stocks, and some invest in volatility target strategies. We estimate that strategies that are targeting a particular level of volatility or managing to an equity floor could have $100-$200bn of assets.
Assuming that, on average, these strategies follow a 2-month realized volatility signal, we can estimate their selling pressure. 2M realized volatility increased over the past week from ~10% to ~20% (i.e., doubled), so these strategies need to reduce equity exposure by up to ~50% to keep volatility constant. This could lead to $50-$100bn of selling, and it likely started already this week. There is often a delay of 1-3 days between when a signal is triggered and trade implementation, and positions are often reduced over several days. We think this could have contributed to the ‘unexpected’ selloff that happened in the last hour of Tuesday’s trading session. While these flows may continue to have a negative impact over the next few days, they would be the first to reverse (start buying the market) when volatility declines.
2. Trend Following strategies/CTA funds have an estimated ~$350bn in AUM. We modelled CTA exposures in our May and June commentaries, and estimated flows under different scenarios for asset prices. In particular, under a 10% down scenario in equites we estimated CTAs need to sell ~$100bn of equities. In our model, the bulk of selling was in US markets, some in Japan and relatively little in Europe. S&P 500 futures did underperform Europe (by ~3%) and Japan (by ~2%) over the last two trading sessions (European hours), which may indicate that CTA flows have started to impact equity markets. The rebalance time frame for CTA strategies is typically longer than for volatility control strategies. CTA funds may act on their signal in a period that ranges from several days to a month. We believe that selling from CTAs may have just started and will continue over the next several days/weeks.
3. Risk Parity is one of the most popular and (historically) successful portfolio construction methodologies. Risk Parity allocates portfolio weights in proportion to assets’ total contribution to risk (a simplified version, called Equal Marginal Volatility allocates inversely proportional to the asset’s realized volatility). In a survey of quantitative investment managers (~800 clients in US and Europe), we found that ~50% prefer a Risk Parity approach (vs. 15% for traditional fixed weights (e.g., 60/40), 20% Markowitz MVO, and ~20% active asset timing). Estimated assets in Risk Parity strategies are ~$500bn and ~40% of these assets may be allocated to equities. Risk Parity portfolios may also incorporate leverage, often 1-2x. Risk parity funds often rebalance at a lower frequency (e.g., monthly, vs. daily for volatility target) and use slower moving signals (e.g. 6M or 1Y realized volatility). The increase in equity volatility and correlation would cause Risk Parity portfolios to reduce equity exposure. For instance, 6M realized volatility increased from 11% to 15% and a modest increase in correlations would result in approximately a ~20% reduction of equity exposure. Based on our estimate of Risk Parity equity exposure, this could translate into $50bn-$100bn of selling over the coming weeks.
In summary, JPM estimates that “the combined selling of Volatility Target strategies, CTAs and Risk Parity portfolios could be $150-$300bn over the next several weeks. Rebalancing of these funds may appear as a persistent and fundamentally unjustified selling pressure as these funds execute their programs. In addition, there may be a positive feedback loop between all of these sellers – Gamma hedging of derivatives causes higher market volatility, which in turn leads to selling in Risk Parity portfolios, and the resulting downward price action invites further CTA shorting. All of these flows pose risk for fundamental investors eager to buy the market dip. Fundamental investors may wish to time their market entry to coincide with the abatement of these technical selling pressures.”
* * *
In other words, if JPM is right, yesterday and today are merely the eye of the hurricane – enjoy them; tomorrow is when the winds return full force.