In recent years, large segments of hedge funds had been struggling with poor performance, thanks to the low volatility environment we had been seeing in the markets. This was even beginning to lead to a lot of client fund redemptions.
However, in February we finally saw volatility make a dramatic shift and move heavily upwards. Unfortunately, this was not exactly the type of volatility that hedge funds were hoping for.
A segment of quantitative trading funds, known as CTA hedge funds, which use computerised trend-following systems and account for about 10% of all hedge fund investment, were down heavily in February. This was the worst monthly performance in 17 years. In fact, pure trend-following CTAs were down 11% in just the first week of February, the second worst weekly decline since 2007 and overall CTAs were down 8% which is the worst weekly loss since the data began in 2000.
Trend-following systems are always going to be long or short during a particular period of time. Therefore, during a prolonged period of having a significant bullish move, they will be heavily long and any quick reversal is likely to leave these funds particularly vulnerable. This will mean quick adjustments will need to be made and analysts at JPMorgan estimated that this difficult situation might have led to as much as $200 billion in equity selling during that time. As one can imagine, this amount of selling only exacerbates negative situations in the markets.
This reminds us of some comments made by Cliff Asness, co-founder of AQR Capital, who was writing last year on the tenth anniversary of the famous “quant quake”. He spoke of the left-tail risks involved in quant funds and pointed out that these short-term extreme movements are often due to so many funds being assigned to doing the same thing at the same time. This means that some of the big events are a liquidity crisis rather than a change in something fundamental. Liquidity events are more likely to just come and go, whereas real negative events are longer lasting.
Based on that, if part of the recent negative movements were amplified due to liquidity events, perhaps that would change how one looks at the negative movements that have taken place in stocks and other markets. Maybe there has not been such a major fundamental shift. After all, quant funds have now reached around 1 trillion dollars in AuM (assets under management), so the risk of huge levels of funds doing the same thing is extremely high.
Some articles have claimed the systems failed to do their job in providing a hedge during a crash. This is not true, as CTA systems will not perform well in a crash; they perform well in a sustained bearish move, like they did during the financial crisis. They are also a long-term play, so single months of performance do not matter to the wider strategy over the long term.
However, these funds only account for 10% of all hedge AuM, and there are other funds that have been performing well. Some macro funds were suffering under low volatility and are now benefitting from the upwards shift; enjoying their best month in a long time. In fact, even many quant funds have been performing well, since trend-following CTAs are only a relatively small part of that.
In any market there are winners and losers. Not all market conditions favour every strategy; some strategies will perform well in some situations and fail in others. In fact, even very sophisticated algos can do some odd things at times, which a human would not do. For example, we recall reading back in 2011 about how Berkshire Hathaway has great performance in weeks when a new Anne Hathaway movie hits the box office. These headline scanning algos are potentially misinterpreting what is going on and buying Berkshire stock.
Overall, no matter how great the performance of a fund, whether it be a quant hedge fund or global macro, it is always important to diversify. Although these unusual events may not happen often, they still do happen and that is enough of a reason to prioritise protecting your capital.
We would like to point out that although we have some very limited experience with algorithmic systems, we are by no means experts and this article only looks at the topic from a high level.
