Speed & Quality of Backtesting with SQ
6 replies
Davey
7 years ago #115032
Hi,
When using MT4 as the engine for backtesting – how quickly does the backtest happen?
I ask because at the moment with MT4 (using every tick on all available least timeframes) to backtest my strategies it can take up to 12hrs to cover 8 years data with up to 220 trades per year.
Is SQ any quicker for a detailed backtest, using every tick, even when using MT4 as the backtest engine?
Thanks in advance,
Dave N
clonex / Ivan Hudec
7 years ago #136556
Few seconds
mikeyc
7 years ago #136557
SQ backtest on real tick data takes between 2 and 30 seconds on my i7 depending on building blocks selected. Same backtest on MT4 can take hours if not days.
Davey
7 years ago #136559
Thanks for the replies everyone. That is a lot quicker then…
Why is an MT4 backtest so slow and SQ so much quicker then?
Threshold
7 years ago #136561
MT4 uses only 1 CPU core. SQ uses Java and multi-threading.
mikeyc
7 years ago #136570
MT4 uses only 1 CPU core. SQ uses Java and multi-threading.
And the MT4 backtester hasn’t really had any work done on it to improve it, or speed it up for 10 year or more. MT4 is a very backward and outdated platform. I sometimes suspect this is a selling point to market maker brokers “Don’t worry the platform is so bad, no client user will ever make money from you using this”. 😉
Threshold
7 years ago #136574
Its good for automated trading though, just not testing. A few hedge funds use it and built their own API and files into the platform for custom money management and other tools to use it for their systems.
MT5 has faster testing speeds and uses multiple cores (I think) but has separate issues of its own. MT4 has more potential than they give it, metatrader dev team needs to step up their work and show it some love.
Viewing 6 replies - 1 through 6 (of 6 total)