molanis.com/forum - Tools for MetaTrader : The place to ask about the best expert advisor builder, expert advisor downloads, and expert advisor programming

Just curious what other people consider a good PF for an EA on back test results? I've read before that 1.5 and above is considered good which seems to make sense, but I'm personally thinking 2+ would be ideal. Maybe I should be asking 'What's the lowest PF you would go live with? And over how long was it backtested?'


Hmmmm...

nolube

Based on my worthless memory and a heap of informal secondary research: if you have complete and accurate historical data, and are able to achieve 90%+ modeling quality, I say 1.5 is good and 2 is excellent.

Keep in mind that most MT4 dealers don't offer tick charts, so MT4 does not "naturally" log tick price data. So with "stock" price data, MT4 approximates intra-bar price action using open prices, or using its lower available time frames. Can you see where this is headed regarding accuracy? Google Birt's Tick Data Guide if you care to read more. Beware of gaps and formatting errors within the data. You will also need some kind of tick logger indi, script, or EA to keep the data current.

I stopped backtesting some time ago in favor of forward testing--partly because many indi's/strat's are not backtestable, but also because maintaining quality historical data is such a pain in MT4.

I would go live with any consistently profitable EA regardless of profit factor. After all, profit is profit and it can be compounded by increasing lot size as long as it's in the black.

Thanks RJo, and sorry for the late reply. Fair points, although most of the strategies I've been working on trade off H4 charts so from what I've heard this negates a fair portion of the inaccuracies. Forward testing is the way to go I agree, but back testing has been very useful for me to learn how actually code the things without too many mistakes! Once you have these skills, then forward all the way!

Cheers