DT-waw, this is a very insightful post. Thanks so much! In my experience every market has a distinct personality - shaped by the traders who trade it pre-dominantly - so its only reasonable to believe that individual sync of the strategy with the market makes more sense. I have always used TS and MC to develop strategies, even though I can employ MATLAB, however I always had this feeling that TS/MC are sufficient to catch a lot of low hanging fruits. Btw, have you ever used MATLAB in your strategy development or did you never bothered ? Thanks for sharing your experience.
Greetings, optimization lovers. To absolutely lay to rest all doubt about whether or not my trend following method will fail in forward testing, to refute any claim that I have made postdictive errorz or WFO peeking, I did a raw, old skool OOS test. I picked a stock, pulled 4 years of data, made ONE ATTEMPT to optimize all 9 degrees of freedom to my heart's content, then pulled in the remaining 9 months of data from Tradestation. 38% return in 9 months:
Picked that ticker at random? Really?... It went in a straight line to new highs, which means any trend system would have worked.
You are probably arguing with a high school dropout who has no concept of what he is doing. No serious person would throw an example like that in a trading forum. He is a joker to be ignored.
Lol. That is an interesting point - the underlying stock moved up during the out of sample period, which I looked at when selecting the stock. Thanks for pointing that out. However, I am pretty comfortable with what this proved - a robust trend following system optimized on 4 years of data did not break down in forward testing. Curve fit or not, it definitely caught the trend in the remaining 9 months. (Bill, I don't hate you or have any need to return your reviling bitterness. I am willing take you under my wing and show you how to make money.)
OK, here is teh portfolio of my best 4 algos, but with percent risk money management based on the risk stop size of each strat. The objective was to proportion the algos so each has the same risk, and max out a standard 2x equity margin account. For example, BAC has a 3.5% stop so it gets 71% of the account per position, BZH has a 5% stop and gets 50% of the account, etc. for a grand total of 200% when all 4 are counted. This creates slightly more than 3% risk per position. The new equity curve:
I seached everywhere in my tradestation 9.1, couldnt find this programm! How to open portfolio maestro???? Where is it? Thank you in advance for the reply.
It's a separate program that you download. Just go to Tradestation's website, log in to your options and check the box for Maestro. Then it will let you download the program.