I am constantly seeking to have one less thing to think about. I liked that line from Forrest Gump: "I had enough money to live on for the rest of my life and that was one less thing to think about"... Having a high frequency scalping method that has a high win ratio makes me feel like I won't have to go through drawdowns and I can paddle my canoe through any kind of markets. The idea is to transform the scenario of a period of losses alternated with a period of wins to a different domain, that of a period of high income alternated with a period of lesser income. That's worked for me, I'm decoupled from the market periods somewhat and that is one BIG thing I don't have to think about...
depends on your psychology. especially if you experience large drawdown with a strategy that youve only been trading for a week, you will probably throw it out in toilet. even if that strategy wins over long time. so its better to have a strategy that has more winning trades, for most people anyway.
You might consider what happens when you risk 58 % of your capital on a trade and lose three times in a row. Ultimately it is my ability to sleep at night that determines my choice of risk level. I use 0.5 % risk in one of my systems. The Kelly criterion model uses assumptions about risk and reward that are not the same as trading in real markets. In real markets anything can happen. My trading style is to test, test, test and find a security that shows a big long term (years) profitable price change. I lose about 80 % of the time. I experience one good winner in about 10 attempts. About 1 attempt in 50 is the big winner that I seek. The big winner might show enough profit to pay all my losses for my entire trading career plus a big profit.
Actually, to be more precise, it means that the amount to risk in order to achieve geometric equity growth is equal to .58 of your account (current bankroll). However, the actual amount to risk will depend on the asset price and stop-loss. Michael Harris clarifies this in the following paper where he also shows the relation of expectancy to Kelly formula: http://www.tradingpatterns.com/Kelly.pdf As it turns out, %Kelly is equal to expectancy/avg win. Ron
win more lose more = counter trend win less lose less = trend trading its the way it is "period" i am more comfortable with winning more losing more. i could not be a trend trader as long as i can make money fading the market. mb
So if you have a trend-following strategy, and it's generating more winners that losers (say 2 to 1)....what does that mean??? Can it really be just all luck, even through 50+ trades?
first what market? i would say that as you add data to your testing it will most likely go down to 30-40 % what you have most likely done is to over optimize on a small set of data. using this method it is possible to get 100% winners using any method. the demons only show up with about 150 trades, then try 250 trades and so on your % of winners will go down as you add data. also if you don't see this deterioration you maybe using an unrealistic entry or exit method using limits which the computer assumes will be filled but in real world will not. there are many many pitfalls that need to be avoided in testing. the most robust models will use a entry on the close of a bar, same for exits. once you have something like this you can then build using price levels. but to come out of the gate using price levels can be misleading. 50 trades is not enough trades to really know what is going on, in fact 50 trades i could easily get 90-100% winners using a simple moving average. mb
What do you mean by over optimize? I mean, what kind of changes lead to overoptimization? For example, I tried one strategy and then adjusted the stops and entry criteria values (the actual #'s) slightly to produce like 5 different equity curves that were very close in pattern and value. I think the one that did the best in that case was overoptimized b/c like you said the values were essentially "tailored" to that data set. On the other hand, I made a fundamental change to the strategy code and then made changes to the values to get 5 new equity curves. These five equity curves were also close in pattern and value to each other but pretty different from the first set. Is this also overoptimization? Also, as I add new live data to each strategy, I still notice that the second set keep doing better than the first set. Is that a good argument for saying that the second set is NOT overoptimized?
>>Is it possible to have more winners than losers? >> I don't care about a stupid ratio. Money management can turn a profit even if you lose more than you win. Fortunately, I have the first blessing but not the second bad luck.