Just wondering if anyone else has dealt with this problem and also come up with a work-around. I'm wondering if I publish my workaround to this issue if the folks at BT would then break the workaround in version 5. I hope not... (Who knows, my workaround might not even work in version 5?)
Drop me a note if this is of interest to you. I don't think this should be made available generally as I have a feeling if that happens BT might defeat the work-around in version 5 intentionally...
...hmm.
I guess it's not of interest to other people after all. Maybe it's just me, but I find this odd as doing meaningful "apples-to-apples" comparisons by back-testing with selected parameter modification is meaningless in the context of changing spreads. If the spread cannot be relied upon to be consistent you might as well not waste your time trying to determine the effects of tweaking parameters. It may be that everyone has decided that the "backtesting results are meaningless at best" mantra is true.
Perhaps that mantra is true.
Or possibly, if one understands the weaknesses and takes appropriate steps to guard against them, backtesting can be a very useful tool. All that being said, if one has to talk to himself on a forum topic it's probably a safe bet that people don't care. So THAT being said, I'll just stop rambling now...
;-)
How to modify/fix the spread is already well publicised in various websites, including this one.
I've written code to modify the spread (and other variables) in the symbols.sel file.
But, you don't need to do that if you're just trying to "compare apples with apples" as you put it. Just perform your backtesting whilst offline; this ensures that the spread doesn't change between tests.
CB
I've written code to modify the spread (and other variables) in the symbols.sel file.
But, you don't need to do that if you're just trying to "compare apples with apples" as you put it. Just perform your backtesting whilst offline; this ensures that the spread doesn't change between tests.
CB
I would be interested to know more about your code. Due to the volume of my testing (3 machines backtesting for days at a time, plus 2 of them also dedicated to live trading, with another machine used to consolidate results) it is not enough to backtest while offline. I need to have the same spread on command at any time on any given machine. That's why I created my work around.
Sure would be interested to know about your findings with symbols.sel though. I thought it was the binary data in the .raw files that is important. But my my work around works with the *.raw and symbols.sel, but not on a fine tuned level like changing variables in symbols.sel. Where did you find the format for this file? Is it available online. I certainly would not be adverse to coding a direct approach with symbols.sel if I knew the format.
I've written code to modify the spread (and other variables) in the symbols.sel file.
But, you don't need to do that if you're just trying to "compare apples with apples" as you put it. Just perform your backtesting whilst offline; this ensures that the spread doesn't change between tests.
CB
i am a ea coder, and i want to know how to modify.sel file
thnaks
Hi Kelly -
You and me both.
There are two messages in this thread that seem to imply no problems setting the the backtesting spread to a fixed value, but neither has responded when asked how to actually do it. Cloudbreaker says that he can modify variables in symbols.sel including spread but has not responded when asked for more information. Maybe he just has not had a chance to answer yet. And Blogzr3 says that it's documented all over the place how to set the spreads the way you want but I and others have looked without finding anything. When asked for specific links, Blogzr3 also has not responded yet...
Hey there Cloudbreaker and Blogzr3... Could you please provide more details when you get a moment? Thank-you kindly, t.
How to modify the symbol.sel file?
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Just wondering if anyone else has dealt with this problem and also come up with a work-around. I'm wondering if I publish my workaround to this issue if the folks at BT would then break the workaround in version 5. I hope not... (Who knows, my workaround might not even work in version 5?)
Drop me a note if this is of interest to you. I don't think this should be made available generally as I have a feeling if that happens BT might defeat the work-around in version 5 intentionally...