…to consider an EA a reliable one.
During the last few days I’ve been asked for more years of backtest for Crescendo. Yesterday we posted a backtest started Jan 2009 till last friday, so about 20 months.
To develop and backtest Crescendo EA, we used data coming from 2008.
That mainly because the Forex market has changed so much during the very last few years that most of the times something that was good in 2007/2008 is no longer good now. We all know that, isn’t it?
Let me tell you something on my experience as a developer and tester: backtesting and optimization doesn’t need to be done for very long ranges of time, but for significant periods. Periods with trending market, flat market, with high and low volatility, etc. The 20 months used for backtesting Crescendo EA, tell you a lot of things. Most of all because they has been made using tick by tick data (99% quality so very near to real time data). THAT is the real difference
The data that are usually used for backtesting (and that are used for common commercial EAs optimization) are very far from real data. That’s one of the reasons why most of the EA fail the real life test even if they perform great in backtests for a lot of years.
But we also have more than 7 months of live trading performances to confirm backtest performances that probably is the most important proof that you can have of the consistency of Crescendo.
So a part of Crescendo EA performances, in general don’t trust too much backtests, not even those made by you on your platform because is mainly the data that you use that are not good. Start trusting them only if after a few months of live trading, performances are confirmed and almost in line with them.