Thanks Visaria, your point is well taken. I think I know what thread/posts you're talking about although I'd have to go look for them. From what I recall, I couldn't really get what he was saying, he was a bit opaque, but maybe was clearer to you? I think he talked about being more tactical than strategic in his choice of entries? Something to do with biology... will have to look it up again later.
Hi Carrer, So, let's say you're testing a system that goes short only and whose exit is not based on exiting at the close. Would you use 2 or 3 above to test it?
I don't quite understand. Did you mean a system which will only short and the exit is not based on the candle close but the entry is based on the candle open?
I misunderstood you originally to mean 'market close' by close. Sorry my bad. I meant a short only system and exit is based on candle close, entry on open, let's say.
If you are testing against an index it's better to test your edges against all 3 because an index is usually long-biased. Therefore, 1 (long only) would be more profitable than 2 (short only). Hence, testing on 'short only' especially on indices is not a good thing. Choice 3 would be more random compared to 1 and 2.
Thanks Carrer. Question for you and others: when you have edged tested your systems ala Acary, did you notice a certain stability to the win ratio for the random test? Did it usually come in somewhere around 30-40%? I just ran my random test over 100 iterations over a number of years and I curiously noticed that the win % is always in that range... Will try Acrary's 1000 iterations next, looks like my C++ code can handle it - 100 turns took about 3 minutes... so would have to wait a half hour before getting the results for 1 year.
Hi Maverick1, If you want to test the edge of your rules, then yes, use entry on break-out without any rules. And why select random entries and not enter systematically (on break-out, in your case) ? If you want to test the edge of your entry mechanism (breakout + rules), then you have to compare with entry on any bar (random or systematic, same comment). As said before, keep everything else identical (exit conditions) - but of course, in the end you will be measuring the edge of the entry mechanism in the context of your exits. So work the exits side to get a fair assessment of that edge (at least, test with an array of stop / targets, and I would suggest no "trade-management" - BE/trailing stop as this will further complicate the landscape). Don't forget to account for commissions & slippage in your tests, as in the long run this is what you have to overcome to just break-even.
If you have an edge on entry+exits smaller than commissions+slippage, Prudent Risk Management won't help you make money ... ... but with a real edge on entry+exits, I agree there is more to look at - position sizing as f(account size) being next.
Most traders focus on trying to pick winners and have very few losers. It's just human nature, but it is exactly this mindset that aids traders in losing money. The only edge that a trader has is the ability to employ prudent risk management. It's right in front of everyone, but almost no one sees it.