i'm an expert long term backtester with tick data, from 2009 using birt scripts
in last 2 years i used with success tickstory, a faster way without doubt
to have faster and reliable backtest optimizations, i always try different ways to test my EAs
the last idea, with really good results, is the following:
using tickstory to download tick data, and build from that ONLY high quality hst files (instead of using the poor mt4 download center)
then, start backtest letting mt4 to generate the fxt from the new high quality hst
in that way, backtests are faster than standard ones on tick data, with really similar results (only if your TP and SL are not soo short)
excited from that, i tried to teach my EAs to trade only on bar opening, for example, on M1
well, in that case, no tick data are required. the EA works only on hst values, and backtest results with "open price only" method are about 20x faster, with exact results compared to tick data backtest (mantaining the logic of trading only on open bar value)
unfortunately, some little limitations appear:
the biggest problem is the minimum size of lot size, that passes from 0.01 to 0.1, in the generated fxt file
i already googled that problem, it's a related problem of a not connected mt4 terminal with broker (i use that way to avoid any interference, and it works always)
to solve that, i thought to simply modify minlotsize value in the generated fxt file of mt4 with FXT EDITOR
well, it works
BUT the problem is that if i open fxt and modify any value, after saving it, the backtest using that modified file will be terribly slow! like a standard bigger FXT file with tick data
so, to resume:
i create only high quality HST files with tickstory
from that, i let mt4 to generate FXT file
this new FXT file has unfortunately 0.1 minimum trade size. but it's really faster in backtest than any other fxt file generated with tickstory
if i try to open and modify that fxt file with fxt editor, to change the value from 0.1 to 0.01 size
when done, the backtest time of that modified fxt is terribly slow, like a standard full tick fxt file 99% quality
but my fxt file is just 100MB instead of 2000MB
it seems that fxt editor changes some other values of fxt files
please help guys!
i will be happy to share in details my procedure to let ALL have really faster backtest, more reliable than standard one from mt4 and really similar to 99% ones
a middle way, in practice
)