Tickstory dukascopy tick data vs mql5 tick data backtest results

New comment
 

When this expert advisor was back-tested using tick-story dukascopy tick data with 100% tick modeling quality the profits were significantly higher compared to the

mql5 data with 100% tick modeling quality. My confusion is which back-test is accurate.


back-test Method: Every tick based on real ticks

Pair: EURO-USD

 
Benrashi Sagev Jacobson:

When this expert advisor was back-tested using tick-story dukascopy tick data with 100% tick modeling quality the profits were significantly higher compared to the

mql5 data with 100% tick modeling quality. My confusion is which back-test is accurate.


back-test Method: Every tick based on real ticks

Pair: EURO-USD

Yes, so deppending on what broker you use, you should backtest your bot in that tick data that the broker is providing you through the platform. I know that the broker wont allow you to have more than, idk, 1 or 2 years of tick data, but normally thats the max they allow to use. Using dukascopy makes sense if you trade on Dukascopy :)
 
Benrashi Sagev Jacobson:

When this expert advisor was back-tested using tick-story dukascopy tick data with 100% tick modeling quality the profits were significantly higher compared to the

mql5 data with 100% tick modeling quality.

The Beta MT5 platform and its data are really best suited for development work (staying ahead of the version curve, if you will) and maybe testing basic functionality.

You should never use the Beta platform and its demo data to test for profitability.

 
Benrashi Sagev Jacobson:

When this expert advisor was back-tested using tick-story dukascopy tick data with 100% tick modeling quality the profits were significantly higher compared to the

mql5 data with 100% tick modeling quality. My confusion is which back-test is accurate.


back-test Method: Every tick based on real ticks

Pair: EURO-USD

can only be accurate if you trade on the same broker as the backtest was done.
 
Alexandre Manuel Cruz Bastos #:
Yes, so deppending on what broker you use, you should backtest your bot in that tick data that the broker is providing you through the platform. I know that the broker wont allow you to have more than, idk, 1 or 2 years of tick data, but normally thats the max they allow to use. Using dukascopy makes sense if you trade on Dukascopy :)
While advising someone to strictly use their broker's tick data seems like a safe baseline, it completely ignores the mathematical reality of why the Dukascopy backtest is showing drastically different profit margins compared to the internal MetaQuotes history, the discrepancy has absolutely nothing to do with which backtest is more accurate and everything to do with the fact that this specific Expert Advisor is clearly curve-fitted to a micro-spread profile, meaning it relies on unrealistic fill conditions that simply do not exist in live retail environments. If an algorithm's profitability completely shatters just because the historical tick source changed, you do not have a robust trading system, you have a fragile, over-optimized script that will instantly collapse the moment the broker applies synthetic slippage or intentionally widens the spread during a news event
 
Amanda Vitoria De Paula Pereira #:
While advising someone to strictly use their broker's tick data seems like a safe baseline, it completely ignores the mathematical reality of why the Dukascopy backtest is showing drastically different profit margins compared to the internal MetaQuotes history, the discrepancy has absolutely nothing to do with which backtest is more accurate and everything to do with the fact that this specific Expert Advisor is clearly curve-fitted to a micro-spread profile, meaning it relies on unrealistic fill conditions that simply do not exist in live retail environments. If an algorithm's profitability completely shatters just because the historical tick source changed, you do not have a robust trading system, you have a fragile, over-optimized script that will instantly collapse the moment the broker applies synthetic slippage or intentionally widens the spread during a news event
True. Absolutely.
New comment