Quality of historical data - very important issue to discuss - please contribute

 

Here's something that's been puzzling me for a while, obviously when designing an expert reliable backtest data is essential. I used to use Alpari data but found it unreliable. I have since been using the download history from recent metatrader builds, have people found this data to be reliable ? I feel this is a very important thread and question as if experts are being built/optimized on unreliable data then obviously they will lose over time.. Thoughts would be appreciated.

Reason: