Data quality question

New comment
 
I ran a 12-month backtest on one of my VPS, which gave the below results with 100% data accuracy:

and also ran the exact same backtest on another VPS, which gave the SAME RESULTS, but has worse data quality apparently:

Can you please help me understand what is happening and how this is possible? 
Can I trust the results on the 2nd VPS even though it appears to have low quality data?
 
Your topic has been moved to the section: Expert Advisors and Automated Trading
Please consider which section is most appropriate — https://www.mql5.com/en/forum/172166/page6#comment_49114893
 
@traderjim42, given that your rating is currently low, you will not be able to embed images. Instead, attach the images as files with "+ Attach" link.
 
traderjim42: I ran a 12-month backtest on one of my VPS, which gave the below results with 100% data accuracy:
and also ran the exact same backtest on another VPS, which gave the SAME RESULTS, but has worse data quality apparently:
Can you please help me understand what is happening and how this is possible? 
Can I trust the results on the 2nd VPS even though it appears to have low quality data?

I will assume that you are testing with real tick data. If not, let us know.

Here is a supposition ...

Say for example that your broker only supplies 6 months worth of tick data.

However, on your first VPS, you have been running it for longer than a year, and so it has kept up with the downloads, and accumulated more than a year's worth of tick data. So, the back-test runs without issue.

On the second VPS, let's say you only recently set it up, so it had no previous data, and since the broker only ever supplies 6 months worth of tick data, that is all you have on it. So when you run your back test, you only have a 50% quality.

Obviously, this is just an example to explain a possible reason for the discrepancy.

 
Fernando Carreiro #:

I will assume that you are testing with real tick data. If not, let us know.

Here is a supposition ...

Say for example that your broker only supplies 6 months worth of tick data.

However, on your first VPS, you have been running it for longer than a year, and so it has kept up with the downloads, and accumulated more than a year's worth of tick data. So, the back-test runs without issue.

On the second VPS, let's say you only recently set it up, so it had no previous data, and since the broker only ever supplies 6 months worth of tick data, that is all you have on it. So when you run your back test, you only have a 50% quality.

Obviously, this is just an example to explain a possible reason for the discrepancy.

Thanks Fernando

I am using [redacted] (Demo) from Australia, and using 'every tick based on real ticks'.

That explanation is probably most likely what has happened. However, why would I get the same results on both tests (identical) when one test has 100% data quality and one test has 12%? 

 
traderjim42 #: I am using [redacted] (Demo) from Australia, and using 'every tick based on real ticks'. That explanation is probably most likely what has happened. However, why would I get the same results on both tests (identical) when one test has 100% data quality and one test has 12%? 

Because in the absence of real tick data, the Strategy Tester creates virtual ticks based on the M1 bar data, which can come quite close to the actual tick data.

So if your EA strategy is not too dependant on minor tick variations (for example, opening positions on the start of a new bar, etc.), then the final results could be very similar or even the same.

It really depends on your strategy and how it manages the positions. If your strategy is robust enough to not be affected by minor tick variations, then that is a good thing.

New comment