Are there any people here who are very quantitative/mathematical who can tell me what they think of 3.3% losses per trade? I'd like to know if anyone thinks this is reasonable.
It is reasonable if your win loss ratio and win size to loss size allows it. It isn't reasonable if your system could potentially have 30 losses in a row. How much of a draw down are you willing to accept? The amount you risk and your probability of ruin is all mathematical assuming good data and allowance for outliers. Systems with low standard deviations may do ok with a 3.3% risk. If your system is highly variable, it may not work. How did you choose 3.3% anyway?
Well, my risk reward ratio (would actually fluctuate) is about 1:2. Meaning I'm generally aiming for profits twice as large as my losses. I came up with the 3.3% partly because it would give me a 33% drawdown with 10 consecutive losses. If I filter my trades, I'm figuring my losses might not even get up to 10 consecutive losses. I know I'm questioning what seems obvious, but sometimes I like to test ideas with real-time prices to see if I could possibly be wrong even when I suspect I know the answer beforehand. Then I know with better certainty.
From Kelly formula (http://en.wikipedia.org/wiki/Kelly_criterion), the best bet is f* = (bp-q)/b, where: f* is the fraction of the current bankroll to wager; b is the net odds received on the wager; p is the probability of winning. If you receive 1-to-1 odds on a winning trade, i.e. b=1, then for obtaining 3.3% bet rate, we have 3.3% = p-q=2p-1 which gives p=51.65%, it mean that if you want to bet 3.3%, then the winning rate should be no less than 51.65%. If your winning rate is just 51%, then the best bet rate is (bp-q)/b = p-q = 2p-1 = 2*51% -1 =2%.
Wait, do you mean if your loss on every trade is 3.3% or if you risk 3.3% of your money? I meant if I permit myself to LOSE 3.3% on every trade.
If I don't like how the trade is behaving, I cut it loose regardless of the profit or loss. Fixed arbitrary numbers don't work for me.
Kelly formula originated in gambling in which how much you bet is how much you lose in a losing bet, so what the formula gives can be understood as the optimal amount you can lose in a trade.
Great! Thanks for that. I have a respect for that idea since I read Williams (can't remember full name) used the Kelley formula to win a contest. If I make it big, maybe I'll give you a first sneak peek at my method. I think I know what you mean, but for me, I find my judgement could get me out of a trade that would have worked out eventually if I just get out because it "doesn't feel right".