I thought I would update you guys on what I have learned this weekend...
The system that Yuppie presents here I think has some serious flaws. I think he has been rather fortunate that he has not blown up his acct after 1800 trades.
Hi eegozi,
I wanted to make a few comments before you throw Yuppie's system under the bus! First, the fact that he has made that many trades, and it's up very handsomely means something is right. 1800 is a very large sample, and gives some credibility to his statistics (market conditions can change, and a system get better or worse - but that's another conversation). Let say the period contained all market types. If so, it looks like it works well.
Yuppie's original question was about MONEY MANAGEMENT. And it's a good question, especially when your win rate is below 50%.
For one this system has a probability that is worse than a 50:50 flip of the coin or throwing darts at the board to pic a trade (he says ~38-40% win rate). This would severely limit his ability to profit longterm because he is likely to have heavy drawdowns.
By heavy, what do you mean? 30%, 70%? eegozi, the drawdown is a function of POSITION SIZING, and related only
indirectly to win rate.
Doing MC Sim on his data shows that he should have an average of 9 losers in a row (at 1R each) and MAX of about 20 losers in a row (so -20R) and only about 5 winners in a row with a MAX of 13 winners in a row.
This is really useful information to Yuppie, and the kind that he can get from doing MC simulations. But it's how it's used that is important... This kind of information prepares the trader mentally for what might come. It's not used to calculate drawdown, because you must consider losses before and after a large streak.
Since each winner makes only 0.088R that is only 1.144R (13 * 0.088R = 1.144R). So we have a loss of ~9R and win of 1.144R. This is horrible and has a high probability of ruin.
No, no, no, no, no, no, NO! eegozi, the winners do not average .088R, that is the average PER TRADE! That includes winners and losers... His average winner is 1.72R. Expectancy is suppose to make things EASIER, not harder. Honestly, each time you do a calculation, you make the math needlessly complex. Just multiply the number of trades times the Expectancy. That's how many 'R' you can expect after that many trades. Then, if you want a rate of return, multiply that times the position size.
The BEST case scenario gives a +94% return after 200 trades. Even more importantly this data is using ONLY 1% capital at risk. The original post said he was using 5% risk which would compound the problems.
The average return, and the worst return are the ones he should probably give the most weight to. And yes, based on the results I listed in a previous post, a 5% position size had a very good chance of blowing out his account.
This is highly doubtful in my opinion that the Yuppie has not encountered this sort of drawdown scenario after 1800 trades.
If you look at his chart, he does have some pretty significant drawdown there...
Also, as we all know a drawdown of this magnitude on a "discretionary system" means that anyone looking at such losses would invariably have their emotions effect their trading and subsequently make mistakes in trading.
Well, that's true. But, according to my earlier calculations, Yuppie could run his system very profitably with a 1% bet size, and contain his risk of drawdown to a 5% chance of a 33% drawdown in 500 trades. Based on 1800 trades in 20 months, he has an average of 90 trades per month. If he would like to look at this problem from an annual perspective, that's 1080 trades per year, average. The simulator I use shows a 2% chance of a 25% drawdown over 1080 trades with a 1/2% (0.5) bet size. Personally, this would be my objective with respect to risk.
Now, what about returns? At the current level of trading, with the exact same trades he took before, he would 'expect' (1080 * .088R) or about 95R. Multiply this by a 1/2% bet size, and he could look for returns of about 47% per year. THIS DOES NOT EVEN INCLUDE COMPOUNDING.
This MC Sim is only a best case scenario IF one trades the system exactly the same as in these 1800 trades.
So, in conclusion I think the data is very likely flawed and/or this system is a terrible system despite the high frequency of trades. Sorry Yuppie, but i think we need more detailed data of your trades to be able to help you with your system.
In order for a system with this poor expectancy (R multiple of 0.088R) to do well he needs to have very high probabilty of a win. In order for a system with a this poor probability to do well he needs to have a high expectancy. However, you cannot reasonably trade a system with both poor probability AND expectancy despite the high frequency of trades.
Good luck
eegozi, I really don't see any reason to question his data - it is what it is. It's just one possible outcome from his system.
The concepts you mention are all important, for example winning percentage and expectancy. But it's important that they not be jumbled together. Expectancy already factors in the winning percentage (what you call probability above). When you say a system with a low expectancy and probability you are counting it twice.
I think what you are trying to say, basically, is that you don't feel his edge is enough. However, if his data is accurate I'm not sure I agree. If he can create similar results going forward (I'm still not clear if this is all back testing, or live trading)
his potential results would put him up there with the best hedge fund managers.
fastcar
