View Single Post
Old 05-30-2016, 04:21 PM   #254
DSzymborski
Rookie
 
OVR: 0
Join Date: Apr 2010
Re: MLB The Show 16 OSFM HYBRID V1 Roster Available Now

Quote:
Originally Posted by WaitTilNextYear
LOL. Developed the personality or the projection systems? Sorry, I should just keep my mouth shut from now on, ha.

On a serious note, thanks for all you do with projections. They've been a big part of making this game play like it should and the modding community has used them (free of charge I might add) for many years in some form or fashion. Now if you could just upload your projections w/splits earlier in the year or have FanGraphs add them to your main page there...........
Personality! lol. No worries, I'm a big boy and can handle criticism. I also recognize that I'm an obnoxious, thoroughly ridiculous individual and there are always people that will find me grating. Comes with the job!

Should probably address one thing in the thread. You'll sometimes see a weird split, especially in very small statistics because for FanGraphs presentation, I *round* the numbers. So if you have a say, a pitcher that ZiPS is projected to a very few number of innings, you'll sometimes get a wacky presented rate because 0.6 projected HR in, like 20 inning but rounded to 1 is a tremendous difference.

I tried early on to show the "true" rates, but then people got mad when they calculated rates from the presented raw rounded stats and they didn't match the real ones and freaked out about that. Explaining to people how a pitcher projected for 4 ER in 9 IP could have a projected ERA of 3.65 or something got people very mad, so I do it this way instead.

As for Odubel Herrera, ZiPS uses PBP data in both majors and minors and I was also surprised by his BABIP projection before the season. But really, given that it projected him to have a very high BABIP anyway, I'm not upset with how the projection has turned out.

Sometimes when ZiPS surprises me, I'm the one that ends up wrong. I spent half the preseason saying that Clevenger would crush his ZiPS and while it's very early, ZiPS is certainly beating me at that projection.

One thing that also should be noted is that in-season regression appears to be less than season-to-season regression as the data strongly support a kind of "stickiness" to some things that are harder to maintain over, for example, three or four seasons.
__________________
Dan Szymborski
[email protected]
ESPN, Twitter
DSzymborski is offline  
Reply With Quote