how can I try to get the best possible data ?

New comment
 
I wonder about the accuracy of broker data on MT5. I made a lot of comparisons on different backtests carried out recently and my findings are quite disappointing.
I find that my backtests can vary greatly from one broker to another while having exactly the same parameters in my EA. In addition, I noticed that there could also be significant differences in results on the same broker between a normal tick backest and a real tick backtest.
In conclusion, it is the backtest in real ticks that is the most credible even if here too I notice small differences with my live account. And then the real tick problem is that the history does not exceed 2 years, which is insufficient in my opinion to carry out solid backtests.

Can a platform like Tickstory provide accurate data when this data differs from one broker to another ?

My question is how can I try to get the best possible data ? Please give me your opinion on this subject.

 
David CAUJOLLE:
I wonder about the accuracy of broker data on MT5. I made a lot of comparisons on different backtests carried out recently and my findings are quite disappointing.
I find that my backtests can vary greatly from one broker to another while having exactly the same parameters in my EA. In addition, I noticed that there could also be significant differences in results on the same broker between a normal tick backest and a real tick backtest.
In conclusion, it is the backtest in real ticks that is the most credible even if here too I notice small differences with my live account. And then the real tick problem is that the history does not exceed 2 years, which is insufficient in my opinion to carry out solid backtests.

Can a platform like Tickstory provide accurate data when this data differs from one broker to another ?

My question is how can I try to get the best possible data ? Please give me your opinion on this subject.

Before "blaming" the data, I suggest you first "blame" yourself for not doing the necessary research into understanding the situation.

  1. Forex, CFD's and Indices, are not necessarily centralised. Brokers have different liquidity providers which also means different data feeds.
  2. Brokers can also be dealing desks, which means their data feeds are also different from the liquidity provider.
  3. Real Tick data is very different from generated virtual tick data (which you called "normal"). You should learn the differences.
  4. Each strategy is different. Some are very sensitive to small changes in data and others not at all.
  5. Incorrectly coded programs can also be affected by such changes.
So, before you blame the data, please do more research.
 

Different broker have different liquidity provider with different interests and therefore they have different quotes.

Learn to handle it.

 
Fernando, Why so much contempt?
I want to say that I just made an observation of precision on the data and that I do not criticize the system. I have very good trading results thanks to algo trading but I always try to be as precise as possible in my backtests. I know very well that there are data variations from one broker to another or from one optimization model to another. I know there is also debate about whether it is really necessary to optimize with perfect data. I know traders who prefer to have imprecise data to backtest but this is not my case. But that is not the problem.
I wonder about the need to buy data (supposedly the most accurate) from the Ticksory platform (which is none other than Dukascopy data) or other software knowing that there are so many differences from a broker to the other.
This would not be a problem for me on a swing strategy where my Take Profit is often placed quite far, often at several tens of pips.
The problem is more on scalping strategy backtests because my average take profit is around 2 to 3 pips. In this case, an error of 0.5 pips can strongly modify the results of the strategy.
For disparaging messages, don't waste your time replying. It does nothing for the forum. And thanks to those who want to contribute to the knowledge of Metatrader 5.
New comment