Are Backtest/Optimization accurate?
Are Backtest/Optimization accurate?
18 Sep 2019, 20:53
I am curious how accurate a backtest and optimized parameters are? I ask this because I don't know if I can trust the performance, As I will backtest 1 year with positive results, then increase time of backtest to longer time period say 2 yrs.. And it will give different results for the 1 year than it did when I just tested for the 1 year by itself. I can see this through the deal map... backtest 1 year trades are there, backtest for 2 yrs, some trades are missing. So it leads me to think its not accurate. If you could you please help shine some light on this matter for me, that would be greatly appreciated!
Thank you.
Matt
Replies
ClickAlgo
19 Sep 2019, 10:10
Hi, P,
This is one of our products the customer is using on a trial basis, I asked the Matt to post on here, I think a cause for this is that with different time periods there will be different trades opening at different times, so it will not be the same. The question here is why a 1 year backtest shows certain results, but when he runs a two year backtest it shows different results from the 1-year. if optimization is running then the results will be different as different parameters are being used.
Matt,
Can you now send an email back to us at support@clickalgo.com with the following information?
- Your parameter settings
- The backtest results for 1-year
- The backtest results for 2-years
- A description of the problem.
If you need help exporting your cBot settings take a look here.
https://clickalgo.com/ctrader-how-to-take-screen-shots
Thanks.
@ClickAlgo
matt92
04 Oct 2019, 00:48
Wow, I am just terrible! I was actually awaiting email responses to reply to this thread, I must have missed them.
I now know exactly what the issue is. It has nothing to do with cTrader backtesting. It is an issue on my bot.
You see, because my bot is a renko bot - it cant be backtest. But it was coded so it would build simulate a renko chart in backtesting, so that it could actually be backtested.
The issue was it was incorrectly building the renko charts, which is why I got different trades when backtesting 1yr and 2yrs back.
I was told this:
the bot does its best to simulate Renko data from tick by tick data, sometimes the back testing data isn't valid and on tick change can be 100 tick change on the price due to missing data.
So it was on my end!! Sorry, and thank you both very much for the replies, I greatly appreciate it!
Please please, can we get renko chart back testing?! xD
@matt92
PanagiotisCharalampous
19 Sep 2019, 08:39
Hi Matt,
To be able to give advice you need to share with us the cBot code, cBot parameters and exact steps to reproduce these issues.
Best Regards,
Panagiotis
@PanagiotisCharalampous