Positive Expectancy

Contest Update:

Up about 11% for the year (see chart below).  All I can do is follow the plan.  I say that over and over, but it is the best thing I can possibly do. Most people fail when they start to deviate from their plan.  This contest system may still fail, but I know it has historical positive expectancy.  That means it has been profitable in the past.  But, I need to follow the plan to reap the benefits of positive expectancy.


That is a good intro to step 8 in my continuing series of "Developing a Trading System"...


Comments and questions, as always, are encouraged...




Trading Process, Step 08 - Have Positive Expectancy



Many people look at a chart, "see" a few instances of a profitable pattern, and then start trading it.


Other people see a flashy new indicator being sold by someone who probably doesn't even trade.  Every example shown leads to profit.  It is the Holy Grail indicator!


Does either case sound familiar?  Unfortunately, that is what most people do before they trade - they find a few profitable examples that back up their thoughts (and conveniently ignore or hide the losing examples), and then start trading.  That is the wrong way to do it.


The right way to evaluate a strategy, whether you are backtesting or evaluating in real time, is to:


1) Have a statsitcially significant number of trades.  5 or 10 trades is not enough.  You want at least 50, preferably 100's or more.  In my contest account, I had over 500 trades evaluated as part of my research.  In my SFE Trading system (available on my website), I evaluated over 3,000 trades.


2) Have an evaluation period that includes bull markets, bear markets, flat markets.  A month or two of market action is never enough.  You really need years.


3) HAVE A POSITIVE EXPECTANCY SYSTEM.  Here is how to calculate it:
Expectancy = ((Probability of Win * Average Win) – (Probability of Loss * Average Loss))/(-Average Loss)
This number must be positive, ideally 0.2 or higher.  The higher the better, although if it is too high, I'd wonder about overfitting or  over optimizing, or some other backtest error.


If you have all 3 of these items, your chances of success go up a great deal.


Next: How To Backtest