Credit: By Cam Hui, CFA :
Josh Brown had a fascinating post which postulated that the quants pose a significant systemic risk to market volatility:
Josh referred to a post by Dominique Dassault at Global Slant and he drew the conclusion that it may all collapse as it did in 2008:
I think that Josh missed the point. The incident that Dassault referred to was not the crash of 2008, in which quants made erroneous assumptions about their model inputs (house prices don't fall). Instead, she was referencing a little known quant meltdown that which occurred in August 2007 in which equity quants got into a crowded trade when someone tried to liquidate - in size. I wrote about this episode before (see Are quants the victims of their own success? ). The chart below shows the HFRX Equity Market Neutral Index (in blue) and a different strategy (in red) that I was involved in which used some very different factors that did not land us in the crowded long.
From discussions with former colleagues and other equity quants at the time, the meltdown was very ugly. Long-only quantitative accounts with relatively low turnover portfolios suddenly saw relative performance tank to -10% to -15% against their benchmarks in a matter of days, as per the above chart. Moreover, factors that were uncorrelated by design, e.g. value vs. growth, all suddenly saw their return correlations converge to 1.
How models blow up
Dassault explained the fatal design flaw of these models in her blog. She had been interviewing with a leading hedge fund manager about 10 years ago who expressed concerns about the stability of quant models:
My circle of quants at the time of the August 2007 were not the Citadels and DE Shaws, but traditional long-only quant shops around Boston like SSgA, GMO and so on. The results were the same. Everyone blew up.
At the time, most of the equity quants more or less had the same approach to modeling, though the details of the models were different. They were multi-factor models with a little bit of growth, a little bit of value, sprinkle in some momentum (fundamental in the form of estimate revision and earnings surprise), technical in the form of PRICE momentum, usually some form of relative strength with a one-month price reversal. They might toss in other exotic ingredients like insider trading activity or buybacks. These multi-factor models used uncorrelated factors by design, so that when you combined them, they were supposed to give you a stable alpha.
Then they overlaid extensive risk control. Dassault explains what she saw at her hedge fund:
As a general rule, hedge fund alphas tended to be shorter lived than long-only alphas, but everyone had portfolios risk controlled 18 ways to Sunday. Yet they all blew up in August 2007.
Andy Lo explains August 2007
Sometime after that fateful month in August, Andrew Lo of MIT made an extensive study of the topic. A subsequent paper by Amir E. Khandani and Andrew W. Lo asked the question: What happened to the quants in August 2007? Evidence from factors and transactions data . Here is the abstract:
Earlier iterations of this research postulated a large seller coming to liquidate what amounted to a crowded trade, otherwise known as the Unwind Hypothesis. In other words, the majority of quants were in the same set of stocks despite the apparent diversity of their models:
Fast forward to today. Dassault postulates a crowded long by the fast money crowd and their positions and returns have been exaggerated by leverage. What's more, the assets are concentrated in just a few very large funds:
Add in a dash of excess conviction and hubris and you get a potential time bomb:
1. Strong Conviction…aka Over Confidence +
2. Low Volatility +
3. High Levels/Low Costs of Leverage [irrespective of Dodd-Frank] +
4. More Absolute Capital at Risk +
5. Increased Concentration of "At Risk" Capital +
6. "Doing the Same Thing"
…Adds up to a Combustible Market Cocktail .
She concluded [emphasis added]:
Spotting the crowded trade
So far, we just have a "this will not end well" story, but we have no idea of what the crowded trades are and what the trigger for an unwind might be. Here is where things get more speculative ( and comments from anyone who is closer to the situation are invited) .
One of the crowded trades are algos that suppress market volatility, either by design or as a side-effect. Bloomberg reported that 2% stock market moves have disappeared in 2015 and Business Insider reported that the market hasn't seen a 1% in eight straight weeks:
This kind of low realized vol environment has made stars of managers who run risky long-tailed strategies that pick up pennies in front of steamrollers. Consider, for example, this account about a fund which holds cash and sells put options on stocks that is being marketed as (*shudder*) a fixed-income alternative. The lack of recent downside equity market volatility has made this strategy a stellar performer and put up a return of 12.7%, triple the stock market.
In a separate post , Dassault calculated the risk-adjusted returns of holding US equities in early 2015 and the results were extraordinary (recall that the Sharpe ratio = (return - risk free rate)/volatility):
To put these extraordinary Sharpe Ratios, the long-term Sharpe Ratios realized by legendary investors like Buffett, Robertson and Soros were in the 0.7 to 1.1 range. Yet, a simple buy-and-hold strategy yielded 1, 3 and 5-year figures better than these investment giants!
These results are highly suggestive that equity volatility has been artificially suppressed in some fashion (no it probably isn't the Fed as QE programs had been well under way for years before volatility dived), As Captain Kirk might say, "Someone, or some thing , is suppressing market volatility to make it seem that stocks are safer than they are." Which brings us back to the Josh Brown comment earlier:
Could the culprit for low vol be the black-box algos themselves? Another characteristic, or side-effect, of the low volatility environment is the stock market appreciated steadily without a 10% correction since 2011.
I have written about this theme many times before (see the Jim Paulsen study indicating the steady market uptrend is breeding complacency and my own work at Why I am bearish (what would change my mind) ). But now, the steady uptrend is starting to crack.
This is the daily SPX chart. Note how the RSI indicator (top panel) has not flashed an overbought reading over 70 and oversold reading of below 30 in all of 2015. The VIX Index (bottom panel) has been steadily declining. Both of these indicators are signs of a compressed volatility environment. On the other hand, the uptrend is starting to labor as the SPX index is losing momentum and appears to be rolling over.
The longer term 20-year monthly chart shows that the MACD histogram fell to a negative reading in January 2015, rose briefly and fell back into negative territory in March. These are the typical signs of a loss of price momentum cited in the Paulsen study. Every past instance in the last 20 years has resolved themselves in bear phases.
If these "black box algos" have been suppressing volatility either as a side-effect or by design, but they need a steadily rising stock market to make money, then could these indications that these models are starting to "knee-bend"?
We can make an educated guess. An examination of the returns of the HFRX equity market neutral hedge fund index*, which is how many of these algo strategies would be classified (though there would be other equity market-neutral strategies in that index) shows that these strategies have been struggling in 2015 with flat returns, though returns were positive in 2013 and 2014. The evidence is tantalizing and suggestive, but not conclusive as returns were not exactly stellar in past years.
Don't panic on an algo unwind!
If I am correct in my hypothesis that these algos are both suppressing equity volatility but require a steadily rising market to achieve returns, then the day of reckoning may be near.
The damage level depends on how crowded the trade is. If the trade isn't very crowded, then we may see a quick "flash crash", where these strategies blow up and the market returns to normal within a day or two. On the other hand, if the unwind becomes a "margin clerk" market that requires a liquidation period of several weeks, then we may see a LTCM or 1987 style crash and snapback.
Should we encounter such market turbulence, my inner investor believes that the best thing to do is nothing. Even if we suffered the worst case of a 1987 event, the market came back to normal within a few months. On the other hand, those who panicked and blinked got hurt very badly.
*Astute readers will ask why a market-neutral strategy requires market direction to make money. I was the analyst who initiated and wrote the BoAML Hedge Fund Monitor, in which we developed a heuristic to reverse engineer the betas of various hedge fund strategies. We found that despite the name, equity market-neutral strategies were generally not market neutral and took directional beta bets.
Disclaimer: The opinions and any recommendations expressed in this blog are solely those of the author. None of the information or opinions expressed in this blog constitutes a solicitation for the purchase or sale of any security or other instrument. Nothing in this article constitutes investment advice and any recommendations that may be contained herein have not been based upon a consideration of the investment objectives, financial situation or particular needs of any specific recipient. Any purchase or sale activity in any securities or other instrument should be based upon your own analysis and conclusions. Past performance is not indicative of future results. Mr. Hui may hold or control long or short positions in the securities or instruments mentioned.
See also Premarket Biotech Digest: JNJ Earnings, Celgene Making Deals, BTDs And FTDs on seekingalpha.com
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
