Mats Bengtsson

Over fitted ideas might seem to work

Mats Bengtsson mib over the years

Over fitting your trading idea until it works in only the test environment is one of the biggest risks.

  • It might be over fitted so that it works on that special stock, but nothing else
  • It might be over fitted so it works on the time period used, but will not work on future data

For example, many programs will allow you to test the profit returned if you run your trading idea on one single stock, or against one single index. They also allow you to optimize your system against that stock or index. Doing that on the MACD signal described in the section on trading systems that does not return profit turns the system into a profit maker against for example SXGE. Using those parameters and running it against all stocks still does not return a profit, actually the loss after optimization has now been turned into a bigger loss.

You are thus facing a number of alternatives:

  • Decide that the system thus optimized is not over fitted, and will work in the future against that stock or index. This is unfortunately hard to tell, you have a system that on average makes a loss. Is this stock/index so much different than the rest that it will continue working for this item? Maybe. But how shall you tell? How will you keep your confidence up after 3 losses in a row? How will you avoid starting second guessing your system?
  • Decide that Technical Analysis is not for you
  • Decide that you will go through anyway, maybe it works although you seem to have the numbers against you
  • Decide to find a system that when working good on the stock/index you are interested in still works with profit on the other tradable stocks you could select to trade with. In this case, it is a lot easier to feel that you have got the odds with you, and that you can keep your confidence even after a number of losses. To do this, you need to test your trading ideas.


Never trade stocks without having tested your system. In fact, most trading systems are not profitable if tested over many stocks. Full story...

Do not invest in heating equipment without having compared the alternatives, not only to current situation, but also to each other. Full story...

About me Mats Bengtsson | Mats Bengtsson Site Map | Privacy Policy | Contact me | | | ©2003 Mats Bengtsson