Bad news everybody, according to my calculations, ( which I sincerely hope are incorrect) the classical pairs trading is dead. Some people would strongly disagree, but here is what I found:
Let's take a hypothetical strategy that works on a basket of etfs:
['SPY','XLY','XLE','XLF','XLI','XLB','XLK','IWM','QQQ','DIA']
From these etfs 90 unique pairs can be made. Each pair is constructed as a market-neutral spread.
Strategy rules:
On each day, for each pair, calculate z-score based on 25-day standard deviation.
If z-score > threshold, go short, close next day
If z-score < -threshold go long, close next day
To keep it all simple, the calculation is done without any capital management (one can have up to 90 pairs in portfolio on each day) . Transaction costs are not taken into account either.
To put it simply, this strategy tracks one-day mean reverting nature of market neutral spreads.
Here are the results simulated for several thresholds:
No matter what threshold is used, the strategy is highly profitable in 2008, pretty good throuh 2009 and completely worthless from early 2010.
This is not the first time I came across this change in mean-reverting behavior in etfs. No matter what I've tried, I had no luck in finding a pairs trading strategy that would work on ETFs past 2010. My conclusion is that these types of simple stat-arb models just don't cut it any more.
Let's take a hypothetical strategy that works on a basket of etfs:
['SPY','XLY','XLE','XLF','XLI','XLB','XLK','IWM','QQQ','DIA']
From these etfs 90 unique pairs can be made. Each pair is constructed as a market-neutral spread.
Strategy rules:
On each day, for each pair, calculate z-score based on 25-day standard deviation.
If z-score > threshold, go short, close next day
If z-score < -threshold go long, close next day
To keep it all simple, the calculation is done without any capital management (one can have up to 90 pairs in portfolio on each day) . Transaction costs are not taken into account either.
To put it simply, this strategy tracks one-day mean reverting nature of market neutral spreads.
Here are the results simulated for several thresholds:
This is not the first time I came across this change in mean-reverting behavior in etfs. No matter what I've tried, I had no luck in finding a pairs trading strategy that would work on ETFs past 2010. My conclusion is that these types of simple stat-arb models just don't cut it any more.
Using pandas + matplotlib? That's not a Matlab plot...
ReplyDeleteThat is correct. This calculation is done with python&pandas
DeleteMy impression is that the reversion has simply moved to intraday timeframes. Can't really compete with the "big boys" on this one.
ReplyDeleteI've got the same feeling, but no proof. Doing the same backtest but with hourly bars could provide the answer. Will give it a thought.
DeleteEven the big boys are having their problems. Check out the poor performance of any of the market neutral equity ETFs or mutual funds (like HSKAX, CSMN). Thanks Jev, that was interesting.
ReplyDeleteIt only tracks one day mean reverting nature of markets as you say yourself. It does not mean cointegration - thus reversion over more than 1 day won't work. It would be useful to publish cointegration market neutral strat on the same universe.jozef
ReplyDeleteYes, this simulation is not about longer-term cointegration. However, I would expect that strategies using a longer holding period have gone through similar performance deterioration.
Deleteif you check statarb this year, they were pretty good this year, YTD is the 2nd of all strategies, might worse than B&H(which is a common issue for all HFs).
ReplyDeleteNormally when do pair trading, you need co-integrated pairs, and sector ETFs are very bad choices both fundamentally and quantitatively. tickers from same sector/related business/common biggest holders, etc. Also It's not surprising that it outperforms in volatile period at all.
Pairs trading is a search for temporary mispricing. Why would you expect mispricing on highly efficient markets/instruments? The theory is true - it is impossible to profit from efficient markets. So many people/funds look on these instruments that the probability of mispricing is virtually zero. Especially if we take index based ETFs. Trading them is so simple. To do pairs trading on them you don't need to be smart, you just need to be quick - and that is what big guys are.
ReplyDeleteIf you are looking for pairs trading opportunities, you have to go to less efficient markets. Go to mid caps, small caps, emerging markets. Yes, you have to take higher risks there but that is what you get paid for - for taking risks. Be smart, take carefully calculated risks and you'll be rewarded.
Pair trading is far from dead.. I know many pair traders that are doing great using the IRIS Pairs Trading Platform. Lightspeed just announced that they are offering the software in this link http://www.bizjournals.com/prnewswire/press_releases/2013/03/07/CG72970
ReplyDeleteSimilar is the behavior, I observed in Indian and Brazilian stock markets. This is primarily because of the fact that the market vol have been relatively low. PT is not going to work if there are not huge deviations in stock prices. During relatively calm periods we need to decrease your threshold for trade entry and exit.
ReplyDeleteWhat you have discussed in PT using cointegration. There is another approach to PT which uses PCA. This methodolgy is likely to generate greater number of trades. If you want to have a look I have written a blog regarding the same:
http://quantcity.blogspot.in/2014/07/pair-trading-using-pca.html