• Registered: 2017-03-14
  • Posts: 709

Topic: Question about historical data (Data Sources)

Hello,

When comparing an EA's performance using two different back testing programs -- e.g. FSB and MT4's Strategy Tester -- I understand why it is important to use the same historical data.  I've seen numerous times how different Data Sources can result in quite different results.

But what if I've decided to forego using MT4's Strategy Tester because I prefer (and trust) the one in FSB.  That is, I'm no longer comparing an EA's results with two different back testers.  Is it still important for me to use my broker's data?  I mean, what is it about my broker's data that makes it a better Data Source than, say, FSB Demo data?

In Popov's video 6 -- "Importing Historical Data from MetaTrader" -- he recommends only using your broker's data.  I would like to understand what quality or property inherent in my broker's data helps the generator create a more robust strategy.

Appreciate any insight...

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: Question about historical data (Data Sources)

FSB Pro can create strategies for any data source, however the purpose of the strategies is to make profit on the real market. This is the main reason we need data for the exact server we are going to trade later.

We test the EAs on MT tester only for confirmation that everything is OK

Popov's Website

  • Registered: 2017-03-14
  • Posts: 709

Re: Question about historical data (Data Sources)

Thanks, Popov.

I was curious about what exactly differs in the data from server to server that is so important.  I mean, suppose my account is with BrokerA.  What is it about BrokerA's data that makes it a better Data Source than BrokerB's data when I am generating a strategy.   Even if some of the numbers differ a bit, why would that make a difference?  Since historical numbers differ from future numbers, then it seems we should be able to use data from either BrokerA's or BrokerB's server.

On the other hand, if we were comparing back test results using FSB and Strategy Tester then I do understand why we would test both using the exact same data and time frames.

Here's a scenario -- suppose I generate an EA using data from BrokerA's server.  Are you saying that my EA, developed using BrokerA's data, will always perform better on BrokerA's server than BrokerB's server?  Isn't it possible that my EA could perform better on BrokerB's server?

  • Registered: 2017-03-14
  • Posts: 709

Re: Question about historical data (Data Sources)

Though a bit dated -- this blog article suggests that one *not* use data downloaded from the broker's server for back testing purposes and lists a few reasons why.  I'm not advocating to follow the blog's advice -- rather, just exploring different view points since, obviously, the Data Source can affect EA test results.

http://mechanicalforex.com/2010/04/unreliability-of-backtesting-data-why.html