Perhaps the thread has gone cold because there has been too much 'after the event' posting, and not enough swapping of live ideas that we could share? In recognition of that comment, AMAT/LRCX & DTV/CMCSA are worth looking it today...
Damn! Sometimes pairs revert too quick - missed out on AM/IILG. Managed to get some SNHY/ZOLT That'll throw the cat amongst the pigeons....
A few that make sense to me at current levels (full disclosure - in all of them): DE/CAT - Giving myself room to add layers. Momo is strong in CAT, but the valuation disparity along with the price divergence over the past few months makes this one set up for a decent countertrend move. PEP/KO - Barrons pump piece on KO appears to have gotten this one out of whack. COP/CVX - Pullback in what appears to be a potential trend pair. On merger arb side, WYE/PFE still has about $1.15 left in it. Note this pair should be done on a 1:1 ratio because of the offer .985 of PFE plus $33 in cash. Also in SGP/MRK, but only about $0.70 left in that one...using a 7:4 ratio there because of the terms of the deal. Both have said they expect the deals to close at end of the 3rd quarter or in 4th. My guess is sometime around earnings.
not sure if it has been discussed here already or not. does anyone use log(p1/p2) instead of spread = p1- beta * P2, the validation and diff between these two?
cipherscribe, cooper, and to others who are NOT using PTF... I make this distinction because I'm not using PTF either and, I'm interested in thoughts from those that are trading having already created a universe of pairs to select, and wait for entry/exit signals from them to trade. My concern is that you can backtest and then extrapolate from this pre-selected universe, however, you will expose yourself to 'survivorship bias' and only analyse data from stocks you have pre-selected because they worked historically. This is dangerous because the relationship may break down suddenly. You can backtest stop losses on the trades, but again, that may interfere with the profiability of your system. My only solution is to apply discretionary analysis to selecting the pairs, and to apply a wide stop loss in order to avoid outlying disasters. I'll throw open the figure of 4% stop loss as a topic of conversation. I also note that whichever data I look at, there is one consistency, which is that the Win Prob is around 65-70% It seems to me that the key to getting a system working is figuring out how to systemise a stop loss that doesnt erode returns. Any thoughts?
I am not using PTF either. Basically, I web scrape the tickers from yahoo complete industry page and iterate the combinations. there I got my potential pair lists (historical good performace), assuming history info including characteristics of the pairs(leading, lagging, quick reversing). I am not worrying about relationship breaking down, I update the list every month, finding most of them are original pairs I have. So I would ask you to give me the evidence.
Cookie, Many thanks for your response. My pair filtering approach is similar to yours, and I'll be very happy to swap pair ideas with you off-thread. I suspect we (along with most on here) are using different entry/exit criteria based on a systematic method. Therefore, I appreciate that different people deal different ways with stop losses etc. To answer your question I would guide you towards UPS FDX. This looks like a great pair until it breaks down in March 09, and ever since then it has trended the other way. In other words, little oscillation since then. Furthermore, one could have backtested it previously to March and it would have encouraged confidence. My signals would have got me out with a 5.32% loss from the intial March move and then another two losing trades on the trend reversal.
I would really appreciate anyone who can give a hint on typical annualized returns of pair trading of stocks. I've read so many posts this weekend through this thread yet to find one.
I've got to the point where I don't even give too much attention to past profitability......I find it can be a little misleading. For me, I put around 80-90% emphasis on the ratio chart. I look over the past year and then 3 months. I purely look for sidewards action and a little volatility (These are usually the ones with highest past profit anyway, but this is beside the point). I could give you 500 pairs with 100% hit rate and huge profitability. I still wouldn't touch these with a 10 foot pole. In the last year we have the advantage of witnessing a bear, bull, and consolidation types of market. If relative value between two stocks holds true through these periods (sidewards ratio chart), then they will probably continue into the future. The last year, if ever, has proven the robustness of the strategy. I agree and think stops are the hardest part..... Unlike momentum trading, the further pairs move against us the higher probability they will revert, assuming no fundamental change (If trading the trend, the further it moves against, the more likely that particular trend is over). Technically on no news, we should be continuing to hold and add to postitions. But when is enough enough? It's hard to get an idea of Risk/Reward when we don't have a pre-defined exit. But again, if we pre-define an exit, then we are cutting short pairs when they theoretically have the highest chance of reversion. Further to this point...... If a pair changes from sidewards to trending, then it will ALWAYS work against us..... Unless you can swing trade pairs (which I don't like)... If the relationship breaks down, we have a 50/50 chance of it breaking down in our favour. However, it is what happens after the fact that presents our problem.... The current ratio will always outpace the MA. Trends in favour will be closed out pretty quick as the current crosses the MA......However, if it moves against us, then MA will begin the dreaded "chase" of the current.. Therefore when a relationship breaks down, the times the breakdown moves against us the loser will always be significantly larger than winner. Which puts a large amount of pressure on our winrate and av win... Personally i've had to remain quite discretionary in these instances. I guess you develop a "feel" for pairs after some screentime. Sometimes i'll add to a position, other times i'll close straight away. Contrary to what I first believed, I think having a sound fundamental grasp can provide a huge edge in pairs trading. Some pairs which have had huge moves against me, i've used some discretion and taken the view it's just an over-reaction to innocuous news. I've then held and they came good. Yes some may argue that is a system in itself and I could outright long/short these stocks, but then I wouldn't be able to write in the "pair journal"