Formula Help-Projection for Quality Starts

[tcs2tx]

One of my leagues uses quality starts (QS) as a category. In the past, HQ has not made projections for QS. As a result, I have simply used wins as a proxy for QS in the calculation of dollar values. I realize that this was not the most accurate gauge of value, but I didn't have the time to derive a formula for projecting QS. If anyone has ever derived a model for projecting QS, I would appreciate sharing the formula.

In the meantime, I am trying to derive the formula myself. As a starting point, I have the data for 2004. It seems to me that the variables to take into account would be (i) ERA, and (ii) IP. I was thinking that the starting point for a model would be the following:

QS = x(1/ERA) + y(IP)

where constants x and y are unknown.

As a preliminary matter, I need help in how to actually test the model. Many years ago, I did some multivariable nonlinear regressions (the buzz words may be wrong due to the number of years that have passed) on a VAX using SAS. I would love to use Excel, but I don't see any built in functions for multi-variable analysis. I've seen a number of articles on HQ that discuss the testing of mdoels, but I don't recall ever seeing anyting on how to actually create and test the models (I realize that this goes beyond the role of HQ).

Can anyone help me out or point me in the right direction?

Thanks,
TC

[PD@HQ]

A lot depends on which Excel version you have and which DataPak tools are installed, but you can do a multi-regression by using the Tools | Data Analysis menu, then choose regression. The y variable (QS) has to be contiguous with all of the X variables. I'm not sure wheere you will get a data set that has QS in it, but if you have it, Excel can do the regression.

[tcs2tx]

Thanks!

Thanks for the quick reply.