Should a betting system/model work on all data?

[WestWasWon]

I've worked on a few models using standard penetration and power rating systems. I've applied them to the Premier League, sometimes using the data from one season to create a rule, then testing it on other seasons. However when I tested the same model on the Championship over around 3000 games it was pretty ineffective. I was wondering if this means that the model is likely to be therefore invalid, and shouldn't be used on any league as a means of prediction.

[KThom]

Re: Should a betting system/model work on all data? Its one i've always wracked my brain about. I don't think you can ever achieve a "one size fits all" system. However, I've failed to find sufficient concrete evidence for me to discard a specific league or country from my analysis as I deem that to be fashioning data/ results to an extent. For me, a proper system should show potential profits regardless of league - as the actual make up of a league will change season on season.