What bothers me using any trading system is that I'm never sure if the performance is sheer luck (good or bad) or something real. Any ideas out there that can help me deal with this issue?
This thread might be a good place to start: http://www.elitetrader.com/vb/showthread.php?s=&threadid=36083&highlight=deactivate It will show you how to calculate a statistical confidence level, providing a more rigid and stastitical way to determine if the system is good, or whether it might just be in a lucky streak instead. -Eric
Backtesting the strategy over a significant time period to include all types of markets may also improve your confidence. I use Wealth-lab Developer almost daily to test trading different trading ideas. - Spydertrader
If your consistently profitable... Let someone else trade your method and if they can be consistently profitable regardless if they can duplicate your results or not... My bet is that you have an edge especially if the method is well defined (no subjectivity). There was a journal here at ET and the guy was doing well and shared his method but the performance by others wasn't profitable... That told me that he was the deciding factor and that his method didn't have an edge although at first glance it looked great. Later he started performing bad and ended his journal. NihabaAshi
EricP, Thanks for the writings. They are very helpful. For statistical purposes only independent trades should be sampled. So how do you determine the level of correlation between trades? Obviously you could have 20 nasdaq stocks all move in one direction in a day, and they don't give you 20 valid samples.
The mythical edge. If your a broker - you have an edge in trading, if your a specialist or MM - you have and edge in trading, if you sell books/courses/systems - you have an edge. You have no edge if you simply buy and sell stocks/futures all day long, because what do you have the others don't but want and are willing to pay for? Have you reinvented the wheel? Or have realized some esoteric pattern that thousands/millions of other traders, analysts, machines haven't figured out yet? Now thats not to say that you can't make money in trading, you can make/lose all the money you want to with discipline or lack of it.
Personally, I check for statistical confidence of a system on a stock-by-stock basis. In other words, you may have a good system that works well over time for IBM, but is not well suited for trading XMSR, for example. If you are interested in checking the statistical confidence of a system over a broad portfolio of stocks, then you might want to combine all trades closed on the same day, and check for the statistical confidence of your combined daily results. -Eric