BryanK
Posts: 140
Joined: 8/21/2005 Status: offline
|
First thing I have to say is "Wow." I was pretty amazed how accurate the engine was, particularly with respect to how stats are distributed across the individual players. I hadn't been expecting this for a fictional league, but with the exception of like 3 metrics, everything was dead on. Having said that, there are some places where the engine could be better for fictional associations: 1. Walks are too rare (i.e. the league average needs to be higher) 2. Homeruns are too frequent (i.e. the league average needs to be lower) Also, there appears to be some small issues about how strikeout totals are distributed across both batters and pitchers. Depending on how easy it is to tweak, it also wouldn't hurt to have there be a *slight* widening of the HR distribution in batters. (i.e. relative to the league average, the best should hit more and the worst should hit fewer) I didn't look at it in this data, but I'm eventually going to check the distributions for GS. Annecdotally, I remember the league leaders having more IP than I felt was normal, and based on some of the findings here I think it's because they probably had too many starts. All in all, the engine is amazing solid. Great job, Shaun! I've prepared some slides that make my points, and all the data I used has been summarized into excel files if someone else wants to take a look. For the slides, click HERE For the summary sheets, click HERE Note: For those unfamiliar with Powerpoint, you can double click on the graph to open up a sheet that contains the datapoints used therein
< Message edited by BryanK -- 4/17/2006 5:21:01 AM >
|