Same strategy when backtested on two brokers giving different results

[gvr321]

I backtested same strategy on two different brokers giving different results, in one it is profitable and in other broker it is showing loss, why this difference? Are backtesting results on MT5 are reliable or is it just waste of time?

Whether MT5 databases are reliable to be trusted being decentralized? or it is just that traders will spend their whole life on building algo strategies under false hope? What I mean to say is if the underlying asset is same and it is backtested on same MT5 platform, then it must give same results irrespective of brokers.

[gvr321]

gvr321:

I backtested same strategy on two different brokers giving different results, in one it is profitable and in other broker it is showing loss, why this difference? Are backtesting results on MT5 are reliable or is it just waste of time?

Whether MT5 databases are reliable to be trusted being decentralized? or it is just that traders will spend their whole life on building algo strategies under false hope?

I kept all conditions similar while backtesting on two different brokers.........

[Oleksandr Medviediev]

gvr321:

I backtested same strategy on two different brokers giving different results, in one it is profitable and in other broker it is showing loss, why this difference? Are backtesting results on MT5 are reliable or is it just waste of time?

Whether MT5 databases are reliable to be trusted being decentralized? or it is just that traders will spend their whole life on building algo strategies under false hope? What I mean to say is if the underlying asset is same and it is backtested on same MT5 platform, then it must give same results irrespective of brokers.

What ticker/pair/underlying are you testing? If it's FX - yes, you may spend your whole life backtesting under false hope.

NOTE: broker's tick history//data-feed, spreads, commissions, swaps, symbol specifications and execution conditions can differ considerably.

Possible long-term fix:
 -- build your strategy on D1
 -- stop optimizing every tick.

[Sergey Golubev]

gvr321:
What I mean to say is if the underlying asset is same and it is backtested on same MT5 platform, then it must give same results irrespective of brokers.

The results should be different fropm one broker to an other broker.
Why?
Because the data (quotes) is different (slightly but different), the number of ticks (which is used for backtesting) may be different, and it depends on the strategy coded to be inside EA (because most EAs are not using W1 and MN1 timeframe to trade so ...).

If you backtest EA with two different brokers and having same results so - ask the coder: which special coding technoque he used for the results to be same ones.

---------------------

• #20 - summary post 
• All (not yet) about Strategy Tester, Optimization and Cloud - summary forum thread
• Testing trading strategies on real ticks and the explanation is on this post. 
• some joking explanation: post #3

[Roberto Danilo Riccio]

gvr321:

I backtested same strategy on two different brokers giving different results, in one it is profitable and in other broker it is showing loss, why this difference? Are backtesting results on MT5 are reliable or is it just waste of time?

Whether MT5 databases are reliable to be trusted being decentralized? or it is just that traders will spend their whole life on building algo strategies under false hope? What I mean to say is if the underlying asset is same and it is backtested on same MT5 platform, then it must give same results irrespective of brokers.

Why make such a big deal out of it? If your strategy falls apart on two different data feeds, it’s probably not very robust, or you’ve overfitted it to that data. That’s one of the fundamentals of algo trading, even before blaming the platform.

Obviously, I’m not saying that backtests on two different data feeds should produce identical results, but a large and clearly noticeable discrepancy between the results is not a good sign.

Also, you should assess whether the historical data you were using in the first place was actually high-quality. If you start from a poor foundation, the other tests can only be worse.

[Muhammad Fa'iq Abdi]

Besides the data feed differences mentioned above, check two specific things in your test:

First is the broker's server timezone (GMT offset). If one broker uses GMT+2/GMT+3 (New York close) and the other uses GMT+0, your H1, H4, and daily candles close at completely different times. Any indicator, pivot point, or breakout level calculated on those bars will generate trades at different times and prices.

Second is your average trade profit in pips. If your strategy takes small targets (say 3–5 pips), a broker with just 0.5 or 1 pip higher spread plus commission will easily flip a winning curve into a losing one. If your edge is too thin to absorb standard broker spread differences, it simply won't hold up in live conditions.