from Amazon
Sunday, March 27, 2011
Base running lwts
I recently revamped my base running linear weights program. The results will soon be available on Fan Graphs, along with their other stats of course.
I am still working out any bugs which might be present and tweaking the methodology. I want to vet some results and methodologies with you guys so you can offer any comments and suggestions.
I don’t include SB/CS/PO numbers. I also don’t include advancing or getting thrown out on WP or PB.
Basically, I keep track of all base runner advances and outs (and not advancing of course) on batted balls.
Right now, the only thing I keep track of on ground balls to the infield is when a ground ball is hit to the SS or 3B, how often a runner on second (with no runners on 1st or 3rd) with less than 2 outs, advances to third (or home) or gets thrown out at third (or home).
I could track how often the runner on first stays out of a DP with a runner on second, less than 2 outs, and a ground ball to the IF (keeping separate track of balls to the left and right side).
For air balls, I keep track of all advances, holds, and outs on hits and outs, with each situation treated separately, and batted balls to each outfield position treated separately as well.
For air balls, I don’t distinguish the fielding location of a batted ball other than which fielder fields it (in other words, left side, right side or middle). And like I said, on ground balls with a runner on second, I only look at balls hit to the SS and 2B, and I don’t distinguish between the two.
I do a simple park adjustment for all batted balls, based on where in the OF (L,C,or R) it is hit at each park.
That is pretty much it.
So I might add in “staying out the DP” for runners on second and less than 2 outs with a runner on first.
I also might add in advances and outs on WP and PB, although I am worried that the numbers will reflect more of how often a pitcher threw a WP (or the catcher committed a PB) than how often a runner advances on a potential WP or PB. IOW, if player A advanced 5 times on a WP and player B advanced no times, I am afraid that the difference is mostly because the pitcher happened to throw more pitches that got away with player A on the bases than the fact that player A attempted an advance more often than player B. But I am not sure. It is always a judgment call on what things to include based on whether there is much of a skill element in the numbers. I suppose I can look at the y-t-y correlations in the WP and PB numbers for base runners, and if it is real low, I won’t include them.
Anyway, here are some preliminary results:
It looks like on a yearly basis, one SD for runs per 150 games is around 4 runs. IOW, the worst and best are plus or minus 5-6 runs per 150. For true talent, one SD is probably 3 runs (plus or minus 4 runs or so). I am assuming that the entire population comprises 3 SD. So, there isn’t a whole lot of value in base running. If we were to add in things that I am not measuring, we can probably add maybe 1 more run in SD. I think that plus or minus 5 runs (per 150) is a reasonable estimate for the limits on base running talent. Again, this is excluding base stealing. I am pretty sure that if you asked a baseball person, they would say that a great base runner is worth several wins. It looks to me like .5 wins (in talent, not observed performance) is reasonable.
One interesting thing is that on a team level, differences among players’ base running performance/talent really tends to even out. This is not surprising as almost all teams have a mixture of slow and fast players. It is almost impossible for a team to be great or terrible in base running.
Stuff you hear about a team like the Rays or Angels being worth 4 or 5 wins in base running is nonsense I believe. Here are the observed team base running totals for 2010. Keep in mind that these are un-regressed numbers. As with UZR, they do NOT tell you what a player or a team has actually done. If you want to estimate that, you would still need to do some regressing. If you wanted to estimate team (or player) talent or do a projection, you would have to regress even more. The reason for that (that you have to regress to get what a team actually did) is that if a team has good numbers, it tended to have had easier opportunities that my methodology is not picking up, and vice versa for teams with bad numbers.
These are absolute runs above or below average and not “per anything.” Again, they would have to be regressed to have any meaning in terms of what actually happened. Don’t ask me how much, because I don’t know.
TBA +18
OAK +11
COL +10
FLO +9
TEX +8
DET +6
CIN +6
ALA +5
NYN +5
LAN +5
ATL +4
SLN +4
SDN +2
CHA +2
MIN +2
PIT 0
NYA 0
PHI -3
HOU -3
CHN -3
CLE -4
BAL -6
ARI -8
SFN -8
WAS -9
BOS -9
TOR -10
KCA -10
SEA -11
MIL -11
Here are some of the best and worst base runners in 2010. There are few surprises. These are in total runs above or below average, and not a rate (like per 150).
Best
Elvis Andrus +7
Drew Stubbs +5.8
Carl Crawford 5.8
Carlos Gonzalez 5.2
Colby Rasmus 5.0
Angel Pagan 5.0
Juan Pierre 4.9
Chone Figgins 4.6
Brett Gardner 4.5
(Joe Mauer is +4.3! Is he that good a base runner or is that just a fluke?)
Worst
Adam Laroche -7.3
Prince Fielder -7.2
Billy Butler -6.1
David Ortiz -5.7
Aramis Ramirez -5.1
Ryan Howard -5
Vlad Guerrero -5
Jason Kubel -4.7
Jorge Posada -4.5
Casey Kotchman -4.5
That’s it for now!
Sorry if I missed it, but what is the reason to exclude stolen bases?