EA Backtesting and Building

Moved discussion from eddiebs thread.

Hi Sheldon,

Interesting idea. I know the problem had to do with ranging markets and low volatility. But the backtesting results from the tester were very good, up to 800% This was not correct and was caused by wrong data in the tester After loading an other data set I got completely different results. I don’t have good data to back test the EA for more then 6 months.

Do you think making step 3 and 4 can also be automated using some indicator?

EA and scalping are both most challenge, i backtest many scalping strategy with ea, but fail to find good one. Current only PA work well for scalping.

Yes, step 3,4 can finish by ea,but hard backtest.
If let ea do it, below is my solution:

  1. indication: atr, 2 MA
  2. calc most trending pair with 2MA, such 5 , 10 ma,
    Less crossing number and pips moved distance per unit time, such as 20 days move 1000 pips, MA no cross.
  3. calc volatile with ATR

Backtest is hard as it use all the pair data.

If we judge by our eyes, it easy.
Such as most trending pair is usoil in the pass year


This was the result from the tester Just price crossing 3 MA

Hi to all guys!

At first u need answer yourself what is a trend and range market conditions in math language. And second one - if u want to digg into pairs correlation and picking a pairs for trade them just fill information about forex basket trading concept and software. I think is the best and handy way for trading multi pairs.

I am not sure how you understant my way to calc most trendure pair.
For example:

  1. gbpusd 20 days move 1000 pips, MA5 and MA10 crossing 2 times
  2. audusd, 20days move 800 pips, MA5 and MA 10 cross 3 times
    3)usoil 20days move 2000 pips, MA5 and MA10 no cross.
    Then we can say that usoil is most trendure, let EA run with usoil.

if ea know when to exchange between trend and rang, then ea is smart as human does. I don’t think it’s easy to let ea smart enough.