If you combine system you get the mean of the systems, both then it comes to returns and probability outcome. You can never get a higher score than the highest component. (A + B + C) / N Combining two systems with 30% and 70% prob give you a new combined system of 50% probability.
For the two strats A and Byou have 4 possible out comes. 1. A win B win 2. A lose B win 3. A win B lose 4. A lose B lose I am going to make one assumption that winners equal losers. So for cases 2 and 3 the net result is 0. So: 1. .36 2. .24 3. .24 4. .16 So your win rate on each trade made is only 36% but 48% is scratch and 16% is a full loser. Now I am guessing that you want the scratches thrown out(which people do when developing a system, but is extremely stupid). So that then give a ratio of .36/.16 or 2.25 of winners to losers. This equates to a 69.23% winning percentage.
it is clearly wrong. re-read the previous posts about one system 100% right, the other is, say, 30%, right. the two system combined will still have a winning rate of 100%.
trend2009, I am going to restate the question for you. I think I know what you are trying to do, but it looks like a number of people have interpreted it in different ways. Here is what I think you wanted to ask, please correct me if I am wrong. I have system S1 which enters the trade when condition C1 is met, and exits the trade when condition C2 is met. I also have system S2 which enters the trade when condition C3 is met, and exits the trade when condition C4 is met. The two systems are independent and uncorrelated. Now let's say I have system S3 with this rules: - enter the trade when both conditions C1 and C3 are met - exit the trade when both conditions C2 and C4 are met And the question is, would system S3 likely be superior (based on risk/reward) to either S1 and S2? Based on my experience with such "combos", the answer is "no".
Why wouldn't you just back-test the new system and get the REAL answer. There is so much left for people to guess on that it 'may' not be possible to come up with a 'quality' answer. However, I would be willing to bet .36 is most likely not the answer.
what?????????????????? Out of which a.. did you pull those assumptions? Seriously, I think you just clearly disqualified yourself from further participation...enough said...
In my original post, I did not mention stop and amount of profit for each trade. I am only interested in entries. in case the original post is unclear, now let us reformulate about this, suppose we only consider long trades: 1) system A generates its own signals to enter the market at the open of the bar, and exit at the close of the bar. if close>open, it is a winning trade; otherwise it is a losing trade (we ignore the case of open=close). statistically, if trading based on system A signals, the winning rate is 60%. 2) system B generates its own signals too, and enters at the open and exits at the close of the bar. it also has a winning rate of 60%. 3) system A and B do not correlate to each other. for example, system A is based on RSI, and system B is based on pinBar. (if you argue that RSI and pinBar are not 100% uncorrelated, then you are off the point). 4) now if I design a new system C, where the signals of C is from system A that is also agreed by system B. what would be the winning rate?
Well then I dont think you understand what you are asking without giving a p/l ratio per strat. Who is to say you make 10 bucks per win but lose 10k per loss. Still a win rate of 60% but a giant sucky strat.