Totally get you, most public trackers are more “marketing curves” than real data. 😅
For ML or serious EA validation, I’ve found that the most reliable approach is using raw broker tick/candlestick data from multiple sources and then normalizing it yourself. Some paid services provide clean, consolidated feeds that are consistent across brokers, and you can export them straight into Python/ML pipelines.
Also, always double‑check for gaps, session offsets, and spreads, even small inconsistencies can ruin backtests or model training.
It’s a bit more work upfront, but far better than trusting fancy dashboards.
You will have to be more specific on what you mean by data? if you are asking about data for mt5 that you can backtest an ea for yourself, then, you can find many posts on many forums that will suite this purpose. However, the results that you get will only be indicative of a broker that has the same data, so if you want data that is the same as or similar to YOUR brokers, then, only way to get that is to create your own data, and then do real testing, live testing to confirm that this new data is so similar or same as your brokers. I have done this myself and my price data, trade executions and closing prices are over 97% the same as my brokers -- But it took years to get it to what it is.
just search the fx forums and backtesting programs online, they have their own forums too, and learn how to use those programs to modify their data and create your own custom data.
• Consistent across brokers
That is the biggest challenge in your requirements.
You're in the States which is somewhat good in this regard.
All FX trades must ultimately be cleared in the global OTC interbank FX market. CFD's are banned. This creates at least some equality among price feeds of different U.S. FX broker-dealers. This is not a guarantee because broker-dealers can, and do, play games in the interim. Unscrupulous broker-dealers will move profitable retail traders to their A-book with liquidity providers, while losing retail remain in their B-book. If you're trading 50 to 100 standard lots, this isn't a problem but what retail trader actually does that? Almost none, so trade execution suffers in the A-book. The point of it is to protect the broker-dealer and its liquidity partners from winning retail traders, or at least to mitigate the damage. This is illegal and it happened to me some time ago which caused me to switch broker-dealers.
Outside the States, CFD's are popular and legal. A CFD broker-dealer maintains its own captive liquidity pool and is not connected to any global interbank market nor asset. Therefore, the odds of finding 2 CFD broker-dealers having the same price feed are basically null. The odds of finding 2 U.S. FX broker-dealers having similar price feeds is a bit better. Of course, the level of integrity of the people that you're dealing with remains paramount.
The only foolproof solution is to trade on a centralized exchange. There are 2 MT5 CME Futures broker-dealers in the U.S., and there are FX pair-derived futures contracts listed there. Order execution and priority is first-come first-serve... period.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Dear all,
I feel a little bit lost in this ocean of unverified data and I will need you. Have you any recommendations from people who’ve been burned before?
For my work — both for EA validation and ML training — I need something very different:
• Usable for long‑horizon backtests without weird gaps or broker‑specific artifacts
I’m looking for scientific‑grade data — something you can actually build models on without feeling like you’re training on illusions.
Thanks a lot for your help :)