Matt Swartz and Eric Seidman released a 5-part series on a pitching statistic they call SIERA. The articles run from February 8 to February 12 at The Baseball Prospectus. What I love about the series is that it hits upon many key concepts in higher math literacy.
SIERA roughly squares GB% + SO% + BB% and then adjusts the coefficients of the resulting nine terms. (They actually use something a little different than GB%). They use algebra to find the product. Then they use statistical tests to determine which of the nine terms really matter. As someone with a mathematics and not a statistics background this example really clarified the t-test and p-test I have seen in other papers.
The authors then go on to use calculus in Part 3 to show that their model predicts improvement, but diminishing improvement as the SO% goes up.
SIERA better predicts (lower RMSE) ERA in the next season than other estimators like FIP. In addition it shows how it better estimates performance for pitchers that cause problems for other models. For instance a pitcher who does not allow many base runners, but allows a lot of fly balls (and hence homeruns) is not penalized as heavily by SIERA because it accounts for the fact that more of these home runs will be solo home runs. If you have a solid algebra background and know something about FIP and ERA I’d recommend that you read the articles.