MT4 Backtest speed on different data sets

 

Hi,

I noticed some strange behaviour in MT4's backtest speed. The speed highly depends on the data used.


I'm using two different sets of data spanning several years. Both set's are rather high quality M1 data spanning the same time period. I'm using two installs of MT4's most recent build and I have enough RAM.

Set1 is pure 4 digit data and has slightly less holes than Set2.

Set2 switches from 4 to 5 digits in the middle of the data and has some more gaps overall.


Now the problem: Set2 backtests slow down heavily a few months after start of the fifth digit accompanied by heavy HDD work. It keeps slowing down until even stopping the backtest too early although the data is fine.

Set1 runs smooth and fast.


So what is the reason for this behaviour? The slightly more gaps? Or the switch from 4 to 5 digit's? And why is it swapping/working on HDD although there's enough free memory?


Thanks for your help.


skai

 
The Strategy Tester synthesises ticks based on volume info on the M1 bars . . if a set of data has high volume on it's M1 bars you will get more ticks synthesised. The heavy HDD action may well be MT4 reading the fxt file . . .
 
Second... Volume would be my suspect as well. The Forexite. Ltd (4-digit Data) has less volumes and therefore runs faster compared to fxdd's data (5-digit Data) which has allot more Volume info.
 

That makes sense. The fast set is indeed Forexite data with volume set to 4 in every M1 candle.


So a volume of 4 or less would then produce 4 Ticks at OHLC? As I'm never using timeframes below 1H and not too tight SL and TP it should be ok to set all volume values to 4.


What do you guys think?

 
skai:

That makes sense. The fast set is indeed Forexite data with volume set to 4 in every M1 candle.

So a volume of 4 or less would then produce 4 Ticks at OHLC?

I think that is how it works . . .
 

skai: What do you guys think? I'm not a big fan of volume so sounds good to me ;)

Reason: