The Other Fifteen

Eighty-five percent of the f---in' world is working. The other fifteen come out here.


Baseball sucks because of stats geeks no one likes

At least it's funnier than most of what makes it to SNL these days. (Hat tip: Yankees Chick)

2 Responses to “Baseball sucks because of stats geeks no one likes”

  1. # Blogger Official Ted Lilly Fan Club

    Hey Colin.

    You're smarter than us, we know this, and that's why we come to you with this. We're somewhat statistically inclined, but the projected IP of pitchers in projection models really baffles us.

    Something like IP clearly matters to an overall season wins projection. For example if Juan Mateo starts 8 games versus 8 Ted Lilly starts, that would have a huge effect on the Cubs win count.

    Do project systems take this into account? It seams that something like PECOTA or CHONE is great for individual projections (K/9, etc) and really a shot in the dark for counting stats (IP, GS) that have a huge effect on a team's performance.... like we said, baffled.

    Is this crazy talk or are we making sense. We'd email you with this, but we don't have it. Feel free to email us at tedlillyrules@gmail.com if you'd rather respond that way. -TLFC  

  2. # Blogger Colin Wyers

    I don't know that I'm smarter than anybody; I think the most that one could confidently say about me is I read very fast and retain reasonably well.

    Forecasts of playing time are generally nothing more (or less) than a weighted average of past playing time. [I go over how those weighted averages work in another post.]

    Here's where people tend to get really confused. Most forecast systems use Minor League Equivalences in their forecasts of players with minor league playing time. The forecasting system doesn't know - and really doesn't care - about the difference between minor league and major league playing time when it comes to forecasting playing time. And so you end up with far more innings projected than actually exist in baseball.

    Essentially, you need to regard all playing time estimates as hypotheticals - if a player was given the opportunity to play, how would he perform? And you have to do it this way, because a forecast of Juan Mateo to play 0 major league innings doesn't tell us anything meaningful, even if it is the most likely scenario. It allows you to play "what if" games.

    Now, if you want to know how well a team will perform, what you have to do is go through and figure out what those opportunities will be. That's why you need to couple the forecasts with a depth chart, like I did for the WAR chart.  

Post a Comment

Links to this post

Create a Link