Backtesting Volatility System
Dear Traders,
I am pleased to start my career in Babypips in this thread. My journey to the Forexland is still in its beginning, I have been studying the Babypips school and forums and practiced with demo accounts, trying to find my trading personality and appropriate systems that would match it and prefereably also be profitable. I have read the thread "PSAR - that's all" through, and found it both entertaining and extremely educating. Thanks, Dale and the others, for both it and this thread.
I have some practical limits for trading caused by my daily job, which I am not intending to quit right now, so I want to check my trades once a day at a fixed point of time. This seems to leave only the volatility system for me of the Wilder's systems discussed here, so here is my small contribution about it to your great work. Sorry, I have not bought "the book" - yet, though I know it is sort of an entrance criterion here. Explanations for the volatility system were available in the internet, and it was not too difficult to understand.
I have read through this thread trying to find some backtesting data about the performace of VS in the long run, and tried to google that as well, but haven't succeeded. So I thought to do some coding in MT4 to see the results myself. But I did not end in going that path because of the algorithm to count the daily values is somewhat recursive, and having done some experimenting with EAs, I was quite afraid about the speed of the calculations - it might take ages to calculate the results for just one set of input parameters. So I decided to do it in a traditional way by coding in C++, which was also an easier coding task. There is a nice feature in MT4 to export data in a csv file, which I have so far done only for EUR/USD, 1999-2008. My program then eats this data and calculates the results. MT4 exports several timeframes in separate files, which makes it easy to compare them. Here are my initial results:
1) During this period of time, i.e. almost 10 years, there were a bit above 50 trades, making about 5 trades per year.
2) I parameterised the ATR factor and the number of ATR days used. For this pair it seemed that 3.1 and 8 yielded about the best results. Just following the rules, it calculated a total of about 5300 pips. The biggest drawdowns for single trades were around 350 pips. 12 of the trades went over 1000 pips, but only two of them stayed there up to the bitter end. My conclusion would be that at least for this pair it would be quite beneficial to use a stop at 1000 pips, once it is achieved.
3) There has been some discussion about how this system would perform on lower timeframes. I did some checking on H4, but the results are not worth mentioning here. I think that once I get some other pairs checked as well, I'll check H4, too, but just for curiosity and completeness. By the way, I had to use ATR factor above 4 there to get at least some kind of positive results.
If you folks are interested to hear more, I'll post results from other pairs as well once I have them. I just hope that my coding went correctly, at least my cross-checkings did not reveal anything alarming.
J.
|
Linear Mode
