azmel, many thanks for posting the demo.
As already mentioned backtesting isn't really feasable as it's mind-numbingly slow (azmel, very useful command 'IndicatorCounted()'
) as the indicator obviously uses all bars in its calculations, and will get progressively slower as backtest 'time' passes... Luckily, it appears to use EOB trading so 'Open Prices Only' backtesting should be reasonably accurate, so I backtested from June '06 -> Present - and it's still took about 3 hours! btw I start from June '06 'cos imo, Metaquotes price-data up till Oct-Nov '06 is worthless, (too easy to make profitable EA's with, therefore suspect) but it's good to have a few months of 'easy data' to compare with 'real' data. That said, notice that the bulk of the profit has been made before trade '216' on equity pic. (Nov '06), and it generally struggles to maintain profitability thereafter - practically every 'profitable' EA behaves the same I have found, and imo, it's all down to the s*itty data, so the EA's are not really profitable (my interpretation obviously...)
But when all is said and done, it did maintain profitability and d/d wasn't horrendous (it wouldn't have survived on the recommended $2,000 though...), so it could have potential...
PS. - Tried Backtesting EURUSD, it opened 27 trades then stopped dead, but no errors - seems to be a bug of some kind. So I tested USDCHF instead. No HG, that's for sure...