If I may present a method I use to compare strategies, as it is difficult to compare when they are over different periods of time and vary so significantly in the percentage of wins and the win/loss ratios.
I find the best way to compare “apples” with “oranges” is to convert them all to a common factor or two. Thus I use the Mathematical Expectancy (ME) and the compound Internal Rate of Return (IRR). The drawdown is not as important as it doesn’t necessarily affect the ME or the IRR comparison, but it is essential as a test of one’s tolerance for risk. There is no fun in watching your account balance depleting significantly in the hope that the back tests will hold true and you will eventually come out a winner. This is the ultimate test of faith.
The ME is the win% x Risk/Return ratio – Loss%.
Thus, from the Detailed Report of the Zig-Zag Trader (which I have posted under Strategies), we get the following:
Win % = 66.3%
Avg gain of winning trades = 3486.74
Avg loss of losing trades = 1537.78
From this we get ME = 66.3% x 3486.74/1537.78 – (1 – 66.3%) = 1.17
From experience, anything over 0.5 is good. Figures like 1.17 are rare.
Now for the IRR.
Annual compounded IRR = (end $’s/start $’s^1/number of days – 1)*365 (if compounded daily)
From the Detailed Report we get:
start $’s = $10,000
end $’s = $670,080
n = no. of days between 11/5/06 and 10/5/16 = 3650
therefore 1/n = 0.000273972602
and IRR = 42% pa compounded daily.
This means your trading account would double every 1.7 years approximately.
I have posted a spreadsheet calculator. You may want to convert it to a single line to compare the calculations for multiple systems.
Thanks a lot Mike. Nice to see you again on forums. It’s a always a pleasure to have some feedback from guys who have been around trading for a while 🙂
I looked at your spreadsheet, great job! How did you make your statistical results of your strategy (nb of wins/losses)?
Hi Nicolas, thanks for your comments. All the data is from the back test report. If you mean statistics like the ‘t’ test, I don’t calculate them. Are they a significant comparator?
I were wondering what was the zig-zag strategy you were talking about. I just saw it in my “pending review” to-do-list stuff.
Well, i’m sorry but it is not accurate. I will post it in the forum instead of the library and I’ll explain in this new thread why.
Nevertheless, sharing your code and your potential good strategy (or indicators/screeners) is something I greatly appreciate because it is just how website like ours can exist and will continue to .. Thank you very much Mike.
I’ll check your new thread with great interest, Nicolas. Thanks for you honest and helpful feedback.
However, I am still wondering if you think that statistical testing of data is really useful and worth the effort for analysing backtest results.
Van Tharp has written a great deal about this and called it a System Quality Number but I think it is the t-score. Are you aware of his writings?
I opened the new thread here about your zigzag strategy:
http://www.prorealcode.com/topic/zig-zag-trader-strategy/
Gday Mike, do you know of any kick arse aussie code writers???
Paul
Gday Paul, great to see a few Aussies on the forum. The only Aussie code writer I know is learning, just like me. How long have you been trading with PRT?
Cheers, Mike
