This may have been covered before so apologies if it has but from my own experiences, never throw anything anyway! (at least in the automated trading strategy sense of the word). 🙂
At least every few weeks I fill up one of my charts with 10 of my most recent strategies and have a run through the different markets and timeframes, there is always value to be found, even if it relatively small gain but with no losses and a tiny amount of time in the market, ( less time in the market = less risk )
Don’t get me wrong it takes some time and dedication but in my opinion well worthwhile. I’m still learning and undecided on whether a strategy that works across many markets is better than one that works exceptionally well on a particular market or TF but if the former is your thing then it may give you extra peace of mind to move it into demo live or if you’re feeling ready for the jump live ! 🙂
Good luck my fellow traders may your digital pockets always be brimming.
I do a similar exercise as you detail above.
I never delete either … I just checked and I have 5,930 .itf files in 242 folders stored on my PC! 🙂
Blimey GraHal !
Glad I’m only really just getting started, not so many to wade through yet 🙂
I think understanding the premise of a strategy is most important.
Testing a strategy on every market and timeframe on prorealtime is a huge time clog. You’re better off using some genetic algo software to do that for you, in minutes.
A strategy that works on multiple markets and timeframes is indeed better than a single market or tf. It adds to the confidence of the underlying premise.
If a strategy works on a 15 min tf but breaks on a 16min tf then it’s just curve fitted and needs to be dumped.
I definitely agree with the first part that you need to have an appreciation of how your strategy is working in market conditions and any basic strategy can be made to make a profit across all timeframes but working on the assumption that most people including banks and professional traders will stick with the most popular timeframes and multiples of these I’ve found the more regular TF’s still give me the best return so far and lend themselves to a certain amount of fitting.
Yes it takes me some time already to review the older strategies against the regular TF’s, I’m not sure I’d have the patience to test against every TF minute using ProRealtime unless there is a coding shortcut that would do it one go? What genetic algo software are you using to test your strategies TF’s in minutes as I would love to try that?
@Kovit @Monochrome @GraHal @robertogozzi @Nicolas
Nice topic Kovit. Likewise, I’d be interested to know how people test a code on multiple TF and multiple Indices (Dax, Nasdaq, DJIA, S&P500, etc.) in a few clics, beacause it takes all my free time to the point where at the end I forget what I was testing.
I wonder if our Masters and Moderators can organize, even anonymously, a poll of actual results of Algos run by the members (Real or Demo accounts) who are happy to share. No need to discolse how much money people are making or investing, just % Win Rate, Risk/Reward, Drawdawn, etc. Backtest with WF is good but, in my case, actual results have always been surprising (Up or Down) although I’m not fan of curve fitting. And for those who want, they can share one or two tips on indicators they use in their best winners.
Also, it would be good to organize Webinars, 1 hour, 1 or 2 times a week, where Masters and Moderators can share insighat on a specific topic and registred members can ask questions live.
Thank you, great Community!
how people test a code on multiple TF and multiple Indices (Dax, Nasdaq, DJIA, S&P500, etc.) in a few clics
Test on 10K bars only … depending on the complexity of the code, it often takes only 10 to 20 seconds for each backtest.
I’m not running complex programs, but when you have to change TF every 2 minutes and then change the Indice and record the results on excel in between… I’m sure there a more efficient way to do this in an automated way. By the way, I’m running everything on Demo account, which takes much more time than on my real account.
What do you think about organizing an anonymous poll on real performance of Algos? Just want to see on actual performance if there’s room to outperform buy & hold of major Indices on the long run…
Sounds a good idea!
I set up and hosted a spreadsheet for use by any / all members to log performance of Algos so we can pool our testing time etc … hardly anybody enters data.
Go ahead, open a new Topic, give it a go! New blood, new ideas! I’d love to see it work!
The enclosed screenshot is the result of my live algos running on a Demo account at IG… whish I had them running on real account… I’ve some losers that don’t show because of the size of the table.
You will notice two of them are running since 29th of Dec. and show a perf of roughly +1500€ (1€ DAX contract), i.e. investment of 15000 pts *.5 *.1 *1 = 750€, so each have generated +200% since 29th Dec.
Any body wants to dive?
@GraHal would you mind posting again the link to the performance spreadsheet? Thanks
Monochrome
Bonjour
pouvez vous nous détailler ce qu’est un logiciel d’algo génétique et comment se le procurer et s’en servir????
Hello
Can you tell us what genetic algo software is and how to get and use it ????
posting again the link to the performance spreadsheet?
I will, but I do feel that you should open a new Topic with a title somethng like ‘Algo Performance – Post Here Please’.
This Topic started off about methods to be adopted to find worthwhile Algos and so we should not hijack / mix subject matter as it makes if difficult for members to find specific information etc.
