I am looking to start a discussion on Pairs Trading. On my site I set up pics of the excel page and my charts so you can all see how it is supposed to work. With very limited info available on the subject I'm having to teach my self and would like for anyone interested to give me input/suggestions/comments etc. Here's the link: http://www.iwinfree.com/pairs.htm
Hi Dustin, I'm strictly an index futures trader, but I find pairs quite interesting. So, just some random thoughts about Pairs trading: One danger is liquidity. In using a longer timeframe, you are not basing a position on a single tick of one of the pairs during the day. Arbs will be watching the more popular and larger issues, and that errant tick will be brought back in line very quickly. I think that there's a certain minimum timeframe that someone can successfully trade a pair, and that's driven by the issues you are trading, the trade routing (ECN vs. Specialist). I like what you are doing by evaluating the value of a move back into the "normal" zone. I would want to make sure to backtest the action of the pair over a very long historical period. When choosing 1.5 std dev. over a 5 day time horizon, there is still a considerable possibility that this pair will continue moving outside of those bands. I'm assuming that you are using 2 minute bars, from the chart lower on the page. What I really like doing is taking something like a three or four year intraday history of something like 5 minute bars, and find all of the levels and their associated probabilities. Like, in your Excel sheet, let's say you have a 3.5 Std Dev close (or alternatively, say 5.03%). With a large amount of historical testing, and an understanding that this is say an 88% trade with a very small adverse excursion, I'd be confident on really leveraging that opportunity (much like being dealt an 11 in blackjack with a very favorable card count....). Other values, I'd size my position accordingly. The advantage to a pairs strategy is that you can pull one of the legs off the position if you want to work with your position timing. So, you can do a longer term trade, with some short term directional stuff as part of it. For example, say you sell the MXIM-LLTC spread, and it keeps moving against you. You can take off the short MXIM intraday and try to time its turn, all the while holding the LLTC long (presumably it's increasing somewhat with the general trend of the market--I'm assuming it's up for this case). This adds risk and commission/slippage, but can be an additional tool for finessing the trade.
Dustin My only idea for you is to maybe try to include the sox index as some sort of standard factor. I feel it would be easier to correlate to an index instead of another stock. Good luck RC
Thanks for the replies. I have made quite a few changes to the spreadsheet and it is working much better. I updated my website for those interested in this stuff. By the way the alert the strategy gave for LLTC/MXIM yielded $1.06 from the close Monday to the close today...RISK FREE TRADING WOOHOO! Or, maybe it was luck ;-) http://www.iwinfree.com/pairs.htm
just a thought.. when you find a disparity in the price levels of a pair and you put on a trade (hedge, arbitrage, whatever), you must believe that the correlation will come back to its 'normal' state. that is, the disparity must have come about due to a temporary imbalance of some sort, and will soon be corrected. the risk is that something fundamental has changed (or will change soon) and the divergence will continue, or increase. it sounds similar to people buying stock because it is less than its book value -- surely the stock must be worth more than its book value right? well, you are counting on others recognizing that fact *and* acting on it (i.e. buying/selling) to bring the market back into alignment. in the end, the strongest force on price is supply and demand.. XYZ can declare bankruptcy and negative earnings for the next 100 years, and every fact can scream that XYZ should tend to zero, but if everyone is buying and no one is selling, the price *will* go up. so my point is that it is not enough that there is a disparity -- you must be confident that *others* will exert buying/selling pressure to bring the prices back into line, so you can take your profit.
It's all about playing the odds that the pair will revert to the mean. Some pairs are better than others, and sometimes the trade just won't work. That is why it's important to find pairs with .95+ correlation numbers. I have no idea if this strategy will work, but these boring summer days are the perfect time to find out.
Dustin, I am a futures spread trader. The most successful traders I know of (in futures) are largely spreaders. I think it is a strategy that is more consistent than outright trading but also requires a much deeper understanding of the markets. I looked at your pairs trade. Using a very simple trading system over the last 2 years yielded 80 points in profit (before costs) with a max drawdown of 20 points. Not spectacular but a starting point. Also, keep in mind that correlation can be deceptive. You can have 2 products that are correlated >.9 but there is no edge whatsoever in fading divergences. The problem with trading spreads from a purely quant approach is occasionally they will blow out (see Long Term Capital Management for a perfect example). So a good spreader will follow the fundamentals/news of the products they are trading and know when to over-ride their model.
the "spread" traders i knew who were successful weren't consumed with the analysis side of the business but were the ones who didn't have a problem adding to losing positions. it seemed the less they analyzed the better the spread trader they made. in fact, just given my very small universe to sample from, those that weren't successful momentum trading or position trading were more apt to do well spread trading and vice versa.
Dustin I don't know if you're still playing with pairs, but EchoTrade is giving a seminar in October, and one of the areas of focus is going to be pairs trading. I talked with a guy at Echo today, and he mentioned that this type of trading has been successful lately, so there might be more of a focus on it than other subjects. The other topics look very interesting also. Here's the link for the seminar http://www.echotradeonline.com/resources/Seminar.htm