So what does profit history have to do with pair selection? In Pair Trade Finder, the Matchmaker table shows the total profit of all the trades taken in the last year for a particular pair, and the average profit. Some pairs with high correlation and with many trades recommended by the Finder lost a lot of money, and some with lower correlation and also many trades made good money. Shouldn't profit history also be a criteria in selectiing pairs? If so, shouldn't the software have that field as part of the Watch Item (Pair Analysis) page?
I really do keep it simple, ive got pairs from several different industries, most of my pairs are above 90% correlation, won't trade anything lower than 80%, I don't like to see correlation strongly trending down, and the ratio chart has got to look like its in a trading range or overextended, I don't want the green line to be a small distance from the blue line compared to previous divergence's, then most importantly just keep making trades. I think the early sized losses in this journal were related to going against the above rules plus extreme market conditions. If you don't know what pairs to add to your watchlist after doing a backtest, try and think objectively, like the top 3 pairs by correlation from each industry or something like that, just stay consistent with whatever you do. I wouldn't rely too much on profit history for pair selection, I would just stick to pairs with high correlation in the same industry, because you could have a great pair, however one trade was negatively affected by surprise earnings, however you wouldn't have taken that trade because of upcoming earnings but the backtester doesn't know that. I have found however pairs with high correlation tend to produce more consistent profits than lower correlated pairs, it isn't random, if you sort the backtester correlation column from highest to lowest then lowest to highest you will see a pattern.
Johnnysharp, could you please clarify what exactly you mean by "overextended"? Do you mean new high/low or something else? My gut feeling is that if I see a chart where the spread is now beyond its levels in the past few months (in other words, a new high or low) it would be a risky trade. But I do not have the statistics/experience to back this up... it might be the other way round. What is your experience?
A couple of days ago there was a divergence between the Dow and Nasdaq. Dow up and Nasdaq down. Wouldn't that be a good pair opportunity? What symbols would one choose? Nasdaq is the Q's, what would be the Dow. WB
I mean look at the distance between the blue line and green line on your ratio chart, is it of similar distance than previous divergence's, I want to see it at least somewhat the same if not more, I don't like a ratio chart making new highs or lows, because that means your fading the trend.
Yes as Gabe mentions you would use the ETF's QQQQ & DIA, however Ive found pair trading profits with ETF's to be low compared with stocks because of their constrained movement and high correlation, there are 1000's of good stock pairs I wouldn't even bother with the ETF's.
Well for me its definately been worth the investment, very happy with my results so far, I posted all my trades ive taken in a spreadsheet on pg71 of this journal if you want to see my performance to date.