Could I please have some advice?

I’ve got a pretty simple strategy. It is performing amazing well in back test on 1 min OHLC Forex data. The profits are pretty volatile, but it looks completely tradable.

However, I’m a little concerned. It is a little too profitable if you know what I mean. I’ve got some posts in OpenQuant’s to determine the best way to get some order book data into the database so I test it with 1 min signals, but executing on tick or order book data thereby crossing the bid / ask every time it trades (see here: SmartQuant Discussion • View topic - Forex tick data into OQ). But I’m told that IQFeed doesn’t support order book data (historical) for Forex ( SmartQuant Discussion • View topic - New IQFeed plug-in).

I’ve started to live test it into a OpenQuant’s but I’ve had some issues with crashes and disconnects. I’ll continue to test it demo to validate somewhat.

I think I can chop the performance in half straight away if I look at slippage and bid / ask spread. Even with that we are still looking at very good returns, but again a little too good; for such a simple strategy I wouldn’t expect this level of performance.

I’d greatly appreciate if I could have I could have some other ideas on how to validate if I’m looking at a backtest artefact here or if I’m actually onto something.

Thanks and regards,

drolles

I don't know anything about OpenQuant, but there are a million and one 'gotchas' of automated trading.

I don't have hours to write a lengthy post, so suffice it to say always forward test the strategy in a live market before trading it cash. Spend time on finding a good way to simulate forward tested trades. If your platform doesn't support this, find one that does.

Backtesting is more or less worthless the way most people try and use it.

Mike