Quant Crash: Human-Robot Blame Game Escalates As CTA, Risk Parity Scapegoating Reaches Fever Pitch

I’ll confess I’m not 100% sure why so many people seem to be zeroing in on that last-minute panic bid on Friday afternoon as a sign that the equity rout is over.

I mean sure, better a dramatic stick save than no dramatic stick save, but Friday wasn’t the only green day last week and it wasn’t the only day that saw wild swings and panic buying.

So I guess it’s not entirely clear to me why anyone should read anything more into Friday than they did into Tuesday.

One excuse for celebrating the late-day Friday action seems to revolve around the notion that the systematic deleveraging is over.

But even that notion is thrown into question by folks who seem to be hell-bent on insisting that the deleveraging never occurred in the first place or if it did, it didn’t have a material impact.

This debate always devolves into the absurd after acute risk-off events and this time is no exception.

On one hand are the quants who claim they had nothing to do with it, and on the other hand are the numbers which seem to tell a different story. And the irony of ironies here is that quants are by definition all about the numbers so I guess when the rest of read these rebuttals of the “blame the quants” thesis we’re all kind of listening, nodding our heads and thinking “ok, but… numbers.”

And by “numbers” we mean the apparent fact that CTAs had one of their worst 5-day stretches on record through Wednesday…

CTAs

… and estimates from the likes of BofAML whose admittedly imperfect models suggest that CTAs and risk parity were effectively forced to offload $200 billion in equity exposure.

CTA

And I mean shit, everyone knows what happens (or at least what common sense dictates should happen) to risk parity in an environment where the stock-bond return correlation flips positive. It’s the same thing that happens to balanced equity portfolios that depend on the diversification inherent in a negative return correlation. Recall this from Goldman, as it relates to the week before last – i.e. to the week during with the above-consensus AHE print seemed to validate the inflation narrative and thus underpin the bearish bonds thesis:

As we have written before, the equity/bond correlation depends on the level, speed and source of bond yield moves. The recent rapid repricing of bond yields has been again difficult for equity to digest. Since the crisis, if US 10-year yields increase by more than 2 standard deviations in a 3 month period, equities have sold off alongside bonds. When rates rise too quickly, they can weigh on growth expectations and valuations for risky assets and rate vol can spill over to equity vol.

RateRise

So what happened to balanced portfolios and risk parity in the week that ended with the best y/y AHE print since 2009? Well, this:

Boom

Again, we’re trying really – really – hard here to avoid being “those guys” when it comes to adopting a habitual tendency to “blame the quants” and we’re not trying to suggest that an army of self-aware T-800s is out to kill us all, but when you get a concurrent drawdown in everything from stocks, to bonds, to commodities it’s difficult to imagine how that is not going to lead directly to all kinds of problems for these strats. I mean just look at this table from JPMorgan:

Problems

Either the numbers themselves are now conspiring to lie to us, JPMorgan can’t do math good, or that’s a goddamn bloodbath. In that same note, JPM contends that the simple math “suggests that both CTAs and Risk Parity funds have been at the core of the recent correction.”

And I mean I’m sorry, but that’s the same thing BofAML said on Tuesday. To wit:

Risk parity and CTAs unwound or are in the process of selling $200bn equities. While our model implements position changes in response to a given day’s moves on the close the same day, in reality, both risk parity and CTA strategies operate over varying horizons. But we expect actual rules-based risk parity and CTA strategies to implement significant allocation changes within a few days of when our model’s positioning shifts. If we assumed $200bn in rules-based risk parity strategies and $250bn in model-driven CTAs, then our models estimated $140bn of global equity unwinds as a result of Friday’s moves and another $60bn as a result of Monday’s moves (Chart 15). For perspective, over the same two days global equity index futures volumes across the largest markets was approximately $1.6 trillion. So if we were to assume the entirety of equity unwinds were completed, then it would equate to approximately 12% of the volume over the last two days. We expect that if risk parity and CTAs are still unwinding equities in the coming days, then it will be against a continued rise in volumes due to higher volatility.

And it’s the same thing Goldman said on Wednesday:

We think the bullish equity positioning has in part been supported by systematic investors, such as CTAs, vol target and risk parity funds, that have been attracted to equities by their low volatility trend, especially in recent months. As trend followers, CTAs (c.US$350 bn, Source: BarclayHedge) would in simple terms buy assets with positive momentum — and the recent negative short-term trend reversal in equities triggers selling. As Exhibit 16 shows, the net dollar CFTC long positioning in the S&P 500 has been correlated with the CTA beta to the S&P 500.

Exposure

Risk parity and vol target funds, which have combined AuM close to US$1 tn, have also likely increased their exposure to equities due to their low volatility — they use both realised and implied measures of volatility (and often also correlation). Based on a simple risk parity and vol target strategy, they are currently running very high equity allocations (Exhibit 17). Risk parity funds usually focus on longer-term measures (6m realised) while vol target strategies often use a shorter window, which suggests they are more likely to de-risk as a result of shorter periods of high volatility. Depending on where realised volatility settles in the coming weeks, selling from those funds might continue. For risk parity and procyclical multi-asset funds, the combined equity/bond sell-off can further increase pressure to reduce risk.

So as far as the sellside goes, there doesn’t seem to be much of a debate here about whether these investor types played a role. The only debate is whether they’re done selling and on that score, the consensus seems to be that they are.

See that JPMorgan table? Well that’s from a note that hit on Friday. This headline crossed at 3:32 ET:

  • CTA/RISK PARITY FUND UNWINDING MOSTLY BEHIND US, JPMORGAN SAYS

I guess what I would say about that is there are two positive ways to spin things, but they are mutually exclusive. Either:

  1. CTAs and risk parity didn’t exacerbate the situation and even if they did they definitely weren’t, as JPMorgan suggests, “at the core of the correction” in which case “the only thing to fear, was fear itself” applies when it comes to model-driven selling
  2. CTAs and risk parity did exacerbate the situation, but now we can relax because that selling is mostly over

But note that you can only choose one of those positive spins. You can’t have both. Either there was nothing to fear from CTAs and risk parity in the first place or there was something to fear, but for the time being we can stop worrying about it.

Again, what you can’t do is have it both ways, because if we didn’t have to worry about them in the first place, then it makes no sense for people to be buying on Friday on the basis of “news” that they’ve stopped unwinding.

If, on the other hand, we did have to worry about them deleveraging but because that unwind is “mostly behind us” we’ve got the green light, it raises questions about what “mostly” means, and it also suggests we’ll have to worry about them again later.

I think perhaps this is a good time to quote BofAML from a note out earlier this year which suggested that what’s far more likely to happen is humans, expecting model-driven selling, effectively try to front-run that selling thus making the fear of these strats more dangerous than the actual deleveraging they might be forced to implement:

When fears of CTAs driving the market lower become a self-fulfilling prophecy.

While our expectations of potential CTA equity deleveraging flows may not dominate volumes in isolation, a remaining unknown is the additional selling pressure from investors fearful of these model driven flows.

The fear of CTA’s rules-based, nondiscretionary selling flows in stress periods may cause other more fundamental and discretionary managers to also unwind which could then potentially create a negative feedback loop of successive declines in equity markets.

So I don’t know. What I do know is that both sides of this debate have devolved into the absurd. Because on one side, it flies in the face of common sense to think that these strategies have absolutely no culpability for anything despite being model-driven and despite being quite large. On the other side, blaming them entirely for a given risk-off event is equally absurd because obviously, there are other investor types in this equation and the whole thing has a reflexive nature to it that’s impossible to disentangle.

exterminate

For the counter argument to the “blame the quants” meme, we’ll leave you in the capable hands of Rusty Guinn:

https://twitter.com/WRGuinn/status/962457488514846722

https://twitter.com/WRGuinn/status/962457490737778689

https://twitter.com/WRGuinn/status/962457492772065281

https://twitter.com/WRGuinn/status/962457494986600448

https://twitter.com/WRGuinn/status/962457497322876929

https://twitter.com/WRGuinn/status/962457499302617088

 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

8 thoughts on “Quant Crash: Human-Robot Blame Game Escalates As CTA, Risk Parity Scapegoating Reaches Fever Pitch

  1. Thanks, excellent article. I can always tell when I’m learning a lot from your articles based on the number of times I read a sentence out loud to myself, parse it, digest it, decipher it…. and then discover the answer that I painstakingly arrived at was just a little further down. Also, exhibit A for why I don’t work on Wall Street, lol.

  2. Trend following CTA models each have added bells and whistles, but you can approximate approximate like 75% of trend-following CTA flows using a simple 25/50 day breakout to enter and exit while taking into account vol-adjusted position sizing from both CTAs and obviously risk parity.

    This is how I see the action over the past week.

    You have rising yields which created initial rise in vol. This caused rebalancing/minor develeraging last Friday. This in turn caused a new 25 day low, which signaled to close positions, leading to violent movement on Monday. Tuesday, panic buying by the BTFD crowd. Thursday I suspect was driven more by updated VaR models, with minor impact from lagging CTAs and rebalancing from CTAs.

    Granted this is overly simplistic. But I contend this explains well above 50% of price movement. If I’m right, the move is done. Institutional money now has rebalanced or hedged exposure, but there is no systematic issues that would lead to lower confidence in equities, and I don’t expect any further major decrease.

    What I don’t know is if the BTFD crowd will be back in force on Monday, but I’ll be buying VIX puts either way.

NEWSROOM crewneck & prints