I have done some quick comparisons between high correlation and high cointegration with PTF 3.01 and I am seeing much better results with cointegration. Using the same groups of stocks/ETFs with a correlation of 80% or greater the average P/L for the group is about 1.5%. For high cointegration (.90 or greater) the average P/L is about 3%. This was just a quick comparison using the same groups of stocks and ETFs.
BD1, what I can't figure out about the new PTF, is if I should ignore all I did before and just rely on cointegration? For example, before cointegration I used to look for this in my pairs: 1) Correlations above 90% 2) Ratio chart should show wide trading range or overextended 3) Correlation in an uptrend 4) +/- and % from mean charts at historical extremes But now with cointegration, what's the point of any of these factors? Do they matter? How do you do it?
Basically I am just replacing correlation with cointegration in my process. I am using high correlation (.95+) as my first criteria. But then I am still apply all the other criteria like you mention in your post: - a good ratio chart - looking at news/earnings for stocks - etc. I am just using cointegration as a better starting point. It should be better if the math and theory of reverting to the mean is correct.
Ive tried talking to the guy. I was trying to get him to sell it but he suddenly stopped emailing me. Too bad its pretty quick.
Has anyone else experienced problems running PTF 3.01? When I run the back tester and then try and find pairs it runs much much slower than the 2.99 version. I ran 2.99 on my old computer and it too 2 min and 20 seconds to run with 10 stocks to find the 45 pairs. The new version on a new computer took 34 min and 15 seconds to run the same data. Just wondering if I did something wrong in the installation or setup. PTF has not responded to my request for help.
My PTF 3.01 is running fine. If anything, slightly faster than 2.99. I have been running lots of backtests of multiple stock groups. No issues other than I got a memory exception when I tried to backtest 5 stock groups at once. So I now just backtest 3-4 max at the same time.
Yes, PTF seems to have a memory leak when running the backtest. I've left it running overnight and in the morning it already consuming over a gig or RAM and crashes.