How do you use splits?
So I'm not a sabremetrician by any means, but I was wondering if you guys had any thoughts on how to use a player's split stats for a given day, to determine which player has the better chance of providing more value for a given day.
So an example would be a player like Adam Jones versus Dexter Fowler and who would have a better game tonight with their respective matchups.
Say Jones is playing versus a lefty, on the road, in a night game while Fowler is playing at home, versus a righty, in a day game. Is there some formula that you would use to even get a rough idea of who's the better bet? Would you weigh their past 2 weeks more than any individual split since the sample size is so small?
Any thoughts on how people would approach this would be great.
s.park
2 comments
|
0 recs |
Do you like this story?
Comments
Best splits work I've seen done was in The Book.
Matt’s written some articles using that process. Let me see if I can find some… Well here’s a really good one:
http://www.fangraphs.com/blogs/index.php/estimating-hitter-platoon-skill/
My guess is that the splits are so fine that regression would beat it out
Certain things, like platoon advantage, do exist and can be quantified. Other things, like day/night situations, are probably affecting players (at best) at too small a level to predict.
For reference, last season the league was essentially the same in terms of day and night games (unsurprisingly). This could mean that players are affected in similar ways such that the overall average is the same, or it could mean there isn’t much of an effect.
I would project splits as mentioned in The Book or in the link Sky provides above. Do it for platoon splits primarily. That would be the way to go.
Marlin Maniac, a Florida Marlins blog
Writer, Beyond the Box Score
Writer, Baseball Propsectus Fantasy Beat
Writer, Heater Magazine

by 

















