I have some data. How can I use Excel to create a formula to predict future outcomes?

http://chandoo.org/wp/2011/01/24/tre...ting-in-excel/

http://lmgtfy.com/?q=how+can+I+forecast+in+excel

Is this an Excel question, or is this more of a math/statistics question independent of which programming language used?

If you already know how you want to perform the prediction, but aren't sure how to program that algorithm into Excel, explain the algorithm to us, and we can try to help you program it into Excel.

If you are still in the "how does anyone anywhere perform this kind of prediction in any programming language?" phase, then I would suggest that you need to solve this question first before worrying about programming it into Excel.

It is true that a multi-variable regression might be a useful tool to make these predictions. Multi-variable regressions can be performed in Excel using the LINEST() function if the function is "linear" or using Solver if the function is non-linear. However, trying to figure out how each input statistic figures into this kind of "deterministic" model is going to be a real challenge.

Through another question posed here (http://www.excelforum.com/excel-gene...ng-system.html) I recently learned about ELO rating systems. It seems to me (as someone who knows nothing about the science of predicting sporting contests) that this kind of system is the basis for most predictive algorithms of this type. If you are in the phase of learning about the science/math, I would probably suggest that you start here, become familiar with this kind of algorithm, then come back to your hockey problem and see if you can see how apply it. I would expect a generic algorithm something like:

1) Using the statistics you have, calculate a rating for each team.
2) Use those ratings to "replay" historical/YTD data for your league to see how well you can "predict" the historical results.
3) Apply the correction algorithms to each teams ratings.
4) repeat step 2 to see if the historical "prediction" improves.
5) continue repeating steps 2 through 4 until you either convince yourself that the predictive algorithm is as good as it is going to get, or until you decide the algorithm is hopeless.