Backtesting results differ on 2 different VPS
Backtesting results differ on 2 different VPS
03 May 2018, 22:50
Hey
Does proxy server that cAlgo is connected to somehow affecting the bactesting results?
I have slightly different results on 2 different VPS:
1. First is connected to Frankfurt-2:
2. Second is connected to Amsterdam-2
Also notice, in the second screenshot the Backtesting data slider (the first one) is not touching the left border meaning I can backtest starting from 2012 whereas in the first screenshot you can see that the max left slider position is 19/01/2014.
Yes, I have cleared backtesting Tick data in BOTH cAlgo installations (C:\Users\Administrator\AppData\Roaming\IC Markets cAlgo\BacktestingCache\Ticks) several times and started the backtesting all over again after software restart. Still results differ.
Any ideas?
Cheers :)
Replies
irmscher9
05 May 2018, 21:26
Could anyone from the dev team respond please? This makes me worried a bit. Today I've backtested another bot, same exact bot (I'ver triple checked it) and results again different.
I'm 100% sure both run on Tick data from server (Accurate).
@irmscher9
PanagiotisCharalampous
08 May 2018, 17:58
Hi irmscher9,
We checked your issue and it seems that the one backtest was executed using a live account and the other using a demo account. Live accounts connect to different servers than Demo accounts and as a consequence a different tick data repository. Therefore some discrepancies in tick data might occur. Maybe you can get in touch with the broker to check and fix the issue.
Best Regards,
Panagiotis
@PanagiotisCharalampous
irmscher9
16 May 2018, 11:19
Ok, here's what they said:
The demo accounts are simply to test out the features of the platform, and not to be used for chart analysis. Any real trading decisions should be made using the data on the live charts only.
So I guess I should only use LIVE accounts for testing from now on.
@irmscher9
... Deleted by UFO ...
ctid418503
05 May 2018, 11:34
I've noticed the thing about slider position and how far back you can backtest. It varies between tick data and M1 prices on my broker. When I first installed CAlgo I could only backtest with tickdata for a few days because the slider wouldn't let me put it futher back (play button was greyed out if I put it further back). Then later I can backtest from about 2013 with tick data. Downloading tick data takes a while, they are big files but I wish Ctrader/Calgo had more visibility of what it's doing because it doesn't instill confidence when one time you can backtest only a few days and then suddenly you can backtest for several years. More transparancy would be good!
I'm not sure about your different backtest trading results. Explanations in my mind are:
1. You were actually running different algorithms
2. You were running backtest with different parameters, maybe using M1 on one and tick data on the other.
3. Your broker is manipulating the price feed.
Probably 1 or 2. Ctrader doesn't help because there is not a lot of transparency in how it downloads historical data. Don't get me wrong, its still better than MT in this respect.
@ctid418503