Different backtest results for demo and live account

PanagiotisCharalampous
16 Apr 2021, 13:44

Hi Sameh,

There are many reasons that can cause such a behavior different price feeds, different backtesting setup etc. You should check your backtesting results and figure out where are the differences. That would provide you with clues regarding the reason behind this.

Best Regards,

Panagiotis 

Join us on Telegram

findsameh
16 Apr 2021, 13:56

RE:

PanagiotisCharalampous said:

Hi Sameh,

There are many reasons that can cause such a behavior different price feeds, different backtesting setup etc. You should check your backtesting results and figure out where are the differences. That would provide you with clues regarding the reason behind this.

Best Regards,

Panagiotis 

Join us on Telegram

But isn't the price feed the same for both accounts? And backtesting setup and parameters are all exactly the same. All I do is switch from demo to real account.

PanagiotisCharalampous
16 Apr 2021, 14:21

Hi Sameh,

But isn't the price feed the same for both accounts?

Not necessarily.

Best Regards,

Panagiotis 

Join us on Telegram

findsameh
16 Apr 2021, 15:04

RE:

Not necessarily.

Best Regards,

Panagiotis 

Join us on Telegram

So basically, for the sake of this argument, if I test a bot and get a return of 110% on a live account and get 60% on the demo account; then I cannot know which result is more in line with the "real" historical prices because we don't really know what the real prices are? Just give me something in order to know how to adjust my results in accordance with this difference. I'm sure you'd agree that a return of 110% and 60% is quite a considerable discrepancy.

PanagiotisCharalampous
16 Apr 2021, 15:59

Hi Sameh,

The discussion is very general and I cannot help you without knowing what you are looking at. To help you more you will need to provide us with information that will allow us to reproduce what you are looking at. We need to know the following

1. The cBot code you are using
2. cBot parameters and backtesting dates
3. Screenshots showing the discrepancies.

Best Regards,

Panagiotis 

Join us on Telegram

findsameh
16 Apr 2021, 16:28 ( Updated at: 21 Dec 2023, 09:22 )

RE:

PanagiotisCharalampous said:

Hi Sameh,

The discussion is very general and I cannot help you without knowing what you are looking at. To help you more you will need to provide us with information that will allow us to reproduce what you are looking at. We need to know the following

1. The cBot code you are using
2. cBot parameters and backtesting dates
3. Screenshots showing the discrepancies.

Best Regards,

Panagiotis 

Join us on Telegram

I think sharing the bot code won't be useful at this point since we know now the historical prices do differ. You'll get different results if you backtested the bot I referred to in the first message. The difference varies from one bot to another. And do you think any change in code can cancel out the differences when the underlying data is not identical?

Here's what I can share for now (the screenshots).

PanagiotisCharalampous
19 Apr 2021, 08:23

Hi Sameh,

Unfortunately at the moment I do not have a live account with FxPro to be able to check. Did you try this with different brokers?

I think sharing the bot code won't be useful at this point since we know now the historical prices do differ

Usually this is not the case but you cannot exclude this from happening e.g. a temporary interruption of the feed on one of the two environments can cause a price gap, hence different price feeds. This is something you need to check since you have access to both accounts.

 And do you think any change in code can cancel out the differences when the underlying data is not identical?

No, you cannot expect the cBot to work the same way if the underlying price feed is different.

Best Regards,

Panagiotis 

Join us on Telegram