Cool new toy

by erik on Apr 17, 2009 9:07 AM EDT Comment 5 comments

Thanks to Dave Studeman, there's now a calculator that gives you a players win probability above a bench player. Pretty handy-dandy. Here's his explanation:

The only other win stat I know that exactly matches each team's wins and losses is Win Probability Added (WPA). WPA is the difference between a player's Win Advances and Loss Advances in each game. By subtracting one from the other, WPA compares a player's contribution to an average (.500) player.

Well, what if you want to compare a player to another level, such as a .350 player, or .400? It's a lot more useful to compare someone to a lower level than average, particularly when players have played different numbers of games. Unfortunately, the math for replacement-level WPA isn't exactly straightforward, because teams compile win advancements even when they lose games. The typical team has 1.65 win advancements in a win and 1.15 win advancements in a loss.

Never fear. I've decided to teach myself PHP and MySQL programming this summer, so I've created the following little PHP script to calculate a player's WPAB. Win and Loss Advances are available from Fangraphs and you can select your own replacement level (please use the decimal point—.350 or .400).

The calculator can be found here. Monkeying around with it, here are the top ten hitters by this measure over the last three years:

Player	+WPA	-WPA	WPAB	Win%
A Pujols	47.21	27.70	24.3	1.269
L Berkman	45.76	29.48	19.5	1.106
R Howard	48.66	34.97	17.2	.958
D Wright	46.48	33.26	16.6	.964
D Ortiz	43.09	29.66	16.5	1.017
M Holliday	47.42	35.44	15.5	.905
C Beltran	42.37	30.19	15.2	.970
V Guerrero	46.45	31.65	15	.940
M Cabrera	44.62	32.89	15	.924
M Ramirez	40.74	29.09	14.6	.967

In case you are wondering, I set replacement level to .380. Albert Pujols, the mang, the mangchine. A team of Alberts would go 210-0 every season.

Now for the laggards. Here are the worst five:

Name	+WPA	-WPA	WPAB	W%
J Castillo	19.00	-25.62	-4.7	0.085
R Cedeno	11.55	17.38	-4.6	-0.064
B Ausmus	15.97	21.44	-3.9	0.091
I Rodriguez	23.3	28.61	-3.1	0.214
Jack Wilson	21.39	26.4	-3.0	.206

0 recs | Comment 5 comments | Share

Story-email Email Printer Print

Comments

This is actually an awesome theory.

For WAR, you need plate appearances to convert from using average as the baseline to replacement level as the baseline. But plate appearances doesn’t work for WPA, because not every plate appearance is equally weighted (because of leverage). So you need something else to measure “playing time”. In this case, WPA+ + WPAneg does the trick, because it’s the sum of all (absolute values of) the little changes in WPA throughout the season. Think of it like separating all of a hitter’s good PAs into a bucket and all of his bad PAs into a bucket, and then recombining. A hitter with 8 WPA+ and 4 WPAneg rates the same as a hitter with 10 WPA+ and 6 WPAneg (both 4 WPA), but the second guy is more valuable (even if they both had the same number of PAs), because he’s been given more “leveraged playing time”, preventing a replacement-level hitter from hitting in important situations.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Apr 17, 2009 9:29 AM EDT reply actions 0 recs

Pujols can pitch?

"If Bowden was a general contractor, he'd build houses with nine bedrooms, six garages, no bathrooms, and half a roof."

by DyeLongJustice on Apr 17, 2009 9:58 AM EDT reply actions 0 recs

I think Pujols' WPA+ and WPA- is listed wrong.

It’s the same as Ryan Howard’s and doesn’t imply such a lofty WSAB.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Apr 17, 2009 10:18 AM EDT reply actions 0 recs

you're right.

I fixed it.

Beyond the Boxscore

by erik on Apr 17, 2009 10:58 AM EDT up reply actions 0 recs

CastiLOL

*sob *

The Basil Fawlty Moderating Strategy:
"We could run a nice blog here if we didn't have all these members getting in the way."
Kevin Frandsen should be with the big team.

by WalrusMan on Apr 17, 2009 11:35 PM EDT reply actions 0 recs