Here's how you can figure out your expected maximum losing streak:
Streak = ln(1/#trades)/ln(losing%). ln = natural log, found on most scientific calculators
Streak = ln(1/500) / ln(.65) = -6.21/-.43 = 14.4 or round up to 15 losers in a row expected in a sample of 500 trades (over a year's worth for my system). Even a system that wins 50% of the time can have a string of 9 losers in a row.
I may or may not be able to trade for the next few trading days due to other commitments so if you don't see a post, that is why. Of course with my luck, any system trades I am not around to take will likely be winners!
Net breakdown (contracts traded):
ZS $917(16)
ZS $917(16)
| RESULTS FOR DAY | |
|---|---|
| Contracts: | 16 |
| Net $P/L: | 917 |
| Wins: | 1 |
| Losses: | 1 |
| Win%: | 50 |
| Avg$Win: | 1559 |
| Avg$Loss: | -641 |
Posts
Glad you were able to hang in there and break the losing streak! As for the "natural log," I think I'll stick to the one's I can find in nature. ;)
ReplyDelete-AT
You're talking about a mechanical system of course where the entry/exit parameters can be strictly calculated via some backtesting program. Does it say what your anticipated # of consecutive winners should be? Have you experienced that side of the equation yet? For those of us trading a discretionary plan there is no such math. That can be a good thing, or a bad thing. My point?? Maybe its just that the mental stress of trading can be either expected (trading mechanically and HOPING that the math will indeed "work out"), or unknown (HOPING that we don't wilt under pressure because there's no math equation that says we ever have to have such a losing streak).
ReplyDeleteTrading Engineer at work ;)
ReplyDeleteA few days late as usual.!..
ReplyDeleteThis must assume a certain confidence-limit. E.g. with WR=50% the probability of 1000 losers in a row is non-zero (although thankfully vanishingly small).
It made my brain ache, which is a good indication that I may be wrong, but I think I disagree with YM-Trader. The calculation is valid because the outcome of *any* trade is probabilistic, the only assumption for the discretionary trader is that trade x is independent of trade x-1. So the criteria is about independence not mechanical/discretionary.
Yes, I have too much time on my hands. Until the Greek vote :-)
... and what I actually meant to ask was, do you know the confidence level (%) this calculation assumes? I am currently doing this 'long-hand' (calculating a few million equity curves while the kettle boils) and your few lines are much more elegant!
ReplyDeleteYM, # of consecutive winners can be calculated with same formula. Just substitute winning% for losing%. In my case, this is around 35% so works out to max of 6 winners in a row that can be expected over 500 trades. Actual results vary of course but my ZS system did have 6 winners in a row back in Feb. 2010 and 8 winners in a row in Dec. 2010.
ReplyDeleteIf your discretionary trading consistently produces a given win%, this calculation does indeed apply. But if it you have a mix of strategies you implement during the day then it would be difficult to calculate a max expected streak.
L&W, good point on the confidence level but beyond my intentions to be that precise. I think some calculus may be needed to determine that figure which would require me dusting off my college text books! Not gonna happen! Haha
Typical engineer, happy to make do with a gross approximation :-) Thanks anyway.
ReplyDelete