A trader that i respect a lot once said that a long-term sucesfull trading strategy will have more losing trades than winning trades, because that's the nature of the game. The key for him was to craft the strategy and its money management in such a way that when you have a winning trade, you make more money than you lose on a losing trade. So, I am testing two strategies. Strategy A makes let's say 1000$ over 25 trades over 20 days. It has 11 winners and 14 losers Another strategy B makes 2000$ over 28 trades over 20 days. It has 13 winners and 5 losers. (Obviously this is a small time period so more backtesting/live testing is required. ) My question is: do you agree with the statement or idea that most stratagies have more losers than winners and that the "edge" comes from having the winners make more than the losers lose? In that case, do you think strategy B (the one that made more, but had more winners) is just on a "lucky" streak and when it evens out to havig more losers than winners it will make less money? I know without specifics it's hard to say, but i'm really asking this as more of a philosophical question on the nature of winning/losing trade ratio as they relate to strategies.
Trend following strategies have typical winners / losers ratio below 50%. I know well performing systems with only 40% winning trades. Scalping strategies can have 70%-80% winners, because of a very tight profit target. In order to judge profitablity of a strategy winners / losers ratio alone is not significant. I am always looking at winners/losers ratio _and_ winners/losers pay off.
thanks for the insight. Just to clarify both of those strategies are trend following. So does that mean that as I continue to backtest (actually live simtrade), the second strategy will likely start to get a bunch of losers so the ratio will eventually got to below 50%?
That makese sense (about expectancy). How many consecutive trades do you consider an absolute minimum in order to calculate a meaningful expectancy?
A practical rule of thumb: You need so many trades that when you remove your largest win or loss your expectancy does not change more than x%. Pick your x to be at most 5.
I'm just more comfortable with a high percentage of winners than low. That led me into scalping. Here's a simulator that can show you what your win /loss size has to be for a given percent of wins [with kelly sizing] http://www.hquotes.com/tradehard/simulator.html
Thanks for the insight and the links!! So if kelly size is say .58 Does that mean you should put 58% of your available $ in one trade? Also, say I calculated the expectancy of both strategies, they both come out to about 1.4 However, one strategy traded twice as much during a given period. That means that one is better right? Since it trades more with same positive expectivity it will make money in longterm correct?
I don't use Kelly nor fully understand it. It is said by many traders to not apply to trading, it was developed for bets with only two outcomes and the argument is that a trade can have more outcomes than that... I use that simulator however to get a feel for how a strategy might do once I have an idea of the parameters. I like to simulate 100 accounts.. having said that, if you place a trade and a bracketed exit nearly every time you have a two outcome situation... I do believe that the kelly size is .58 of your account for maximum growth rate in your example. If one strategy trades more often and the wins and losses are the same size as the other strategy then more often is better.... for laughs simulate the default values for 100 accounts, it's all 50/50 but you will see that one account would be doing great ["hey everybody I have the grail"] and another would be doing similarly terrible ["trading doesn't work"] when both were following an entirely random path..
to me it boils down to what actually affects you. if you are a daytrader and being more often right than wrong is necessary for your selfconfidence, go for that. if you are a thickskinned bigball go for trend following with hits (well) below 50%. i think that most people actually care about their daily pnl and how much it varies. if you have that kind of mindset trade both strategies and work on a third, then a fourth ... you see what i mean. and let sharpe be your key figure, because no profit factor, hit ratio, payoff, treynor, mar, sterling, tells what you actually care for: the daily pnl, its "positiveness" and its variation. once you see that you don't care about frequencies, styles, hit ratios. you just go for the nearest way to increase your portfolio sharpe. so, to cut this long post: your initial question is shedding light on something completely irrelevant to professional investment operations. you are simply asking the wrong question. and it is wrong because any possible answer is irrelevant.