Backtest data quality
I've been backtesting a US30 EA strategy. The results seem promising, so I'm testing over a longer timer period. When I expanded the test date range to 2 years, I hit a snag where the data quality, according to the Strategy Tester backtest tab, dropped to 20% - obviously not good. Their US30 data for 2025 gives 100% quality but 2024 was poor.
I got a demo account from another broker that gave me 100% quality for the 2 year time period = good.
A friend with a more powerful PC helped me out with an optimization (just for the last month) - but when I tried to reproduce the most profitable configuration on my system, I just got break even and a rollercoaster equity/balance graph, nowhere close to their backtest. I discovered that they'd used my original broker for the test, which I thought explained the issue - I can reproduce the result by swapping to the original broker.
But here is the puzzle: Basically, I can run the 1 month backtest on each broker and get wildly different results, with both claiming 100% quality.
Where should I start digging? Thanks.
whats to dig for? you can get very different results from 2 different computers, using the same broker account for many reasons.
You throw in 2 different brokers and those differences are much more.
Search this site for this subject, you will find there is many reasons for this. It is not a new phenomenon. Instead, it has always been there.
if your issue is about the data quality, that is a broker issue, and nothing to do about that, except deleting the data you have and re downloading them again, and hope that you retrieve more the 2nd time. If you have same bad quality, then, it probably means the broker does not have any data from that period and mt4/5 is creating virtual data. This is the same as having bad data imo. In this case there is nothing you can do to improve it.
If your issue is about the different results from the same ea, on 2 brokers, then see my previous response.if your issue is about the data quality, that is a broker issue, and nothing to do about that, except deleting the data you have and re downloading them again, and hope that you retrieve more the 2nd time. If you have same bad quality, then, it probably means the broker does not have any data from that period and mt4/5 is creating virtual data. This is the same as having bad data imo. In this case there is nothing you can do to improve it.
If your issue is about the different results from the same ea, on 2 brokers, then see my previous response.Thanks - Your initial response assumed different brokers and computers but it's the same computer and MT5 and EA but with different brokers. Both tests claim 100%, but with different results. It scares me though, because I don't know which is correct, and going live with either (given its US30) would be a huge risk. Especially, if its an issue with the EA. Maybe I'll have to run them side by side on live demo accounts and see which one performs best.
I have deleted and reloaded several times, but no change.
Thanks - Your initial response assumed different brokers and computers but it's the same computer and MT5 and EA but with different brokers. Both tests claim 100%, but with different results. It scares me though, because I don't know which is correct, and going live with either (given its US30) would be a huge risk. Especially, if its an issue with the EA. Maybe I'll have to run them side by side on live demo accounts and see which one performs best.
I have deleted and reloaded several times, but no change.
so i did understand you. You just didnt believe me. my initial response stands. i did NOT assume that you were on different computers. Both brokers could be both correct. Do some research, and you will find that 1 ea will make millions at one broker, but same ea on different broker will go broke within days due to the price differences mostly, but also coding of the ea.
btw. the fact that moderators arent "standing in" or correcting my statements, is a reasonable assumption that my responses are accurate.so i did understand you. You just didnt believe me. my initial response stands. i did NOT assume that you were on different computers. Both brokers could be both correct. Do some research, and you will find that 1 ea will make millions at one broker, but same ea on different broker will go broke within days due to the price differences mostly, but also coding of the ea.
btw. the fact that moderators arent "standing in" or correcting my statements, is a reasonable assumption that my responses are accurate.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I've been backtesting a US30 EA strategy. The results seem promising, so I'm testing over a longer timer period. When I expanded the test date range to 2 years, I hit a snag where the data quality, according to the Strategy Tester backtest tab, dropped to 20% - obviously not good. Their US30 data for 2025 gives 100% quality but 2024 was poor.
I got a demo account from another broker that gave me 100% quality for the 2 year time period = good.
A friend with a more powerful PC helped me out with an optimization (just for the last month) - but when I tried to reproduce the most profitable configuration on my system, I just got break even and a rollercoaster equity/balance graph, nowhere close to their backtest. I discovered that they'd used my original broker for the test, which I thought explained the issue - I can reproduce the result by swapping to the original broker.
But here is the puzzle: Basically, I can run the 1 month backtest on each broker and get wildly different results, with both claiming 100% quality.
Where should I start digging? Thanks.