Brewers First-Half WPA Review, Part 1

by Cheeseandcorn on Jul 11, 2011 12:39 PM CDT

Win Probability Added isn't a particularly useful stat for measuring player performance - it doesn't account for defense and a lot of baserunning, and it can't distinguish between a pitcher who strikes out the side in order and one who walks the bases loaded before getting out of the jam.

But it's one of my favorite stats because it's one of the few that meaningfully measures how much a player's performance matters, which is a lot more closely related to our experience as fans than just the performance itself. So I thought it'd be a good lens through which to examine the Brewers' first half. I decided to measure it on a game-by-game basis because a) you can just find the season-long totals here; and b) I think the game-by-game tally will be probably be more interesting anyway.

I'll start with the game MVP and LVP tallies for the season. (I'll have some hitting-specific stats and the best and worst of the season tomorrow, and then I'm undertaking a foolhardy attempt to measure fan anguish in a post I'll put up on Tuesday.) One note: Batting is included in pitchers' WPAs - in most cases, it didn't significantly affect their totals.

Brewers WPA MVP Leaders

15 - Prince Fielder
11 - Rickie Weeks
7 - Ryan Braun, Shaun Marcum, Randy Wolf
5 - Yovani Gallardo, Carlos Gomez, Chris Narveson
4 - Jonathan Lucroy, Casey McGehee
3 - Corey Hart, Mark Kotsay
2 - Marco Estrada, Zack Greinke, Mike McClendon, Sergio Mitre, Nyjer Morgan
1 - Erick Almonte, John Axford, Yuniesky Betancourt, Tim Dillard, George Kottaras, Josh Wilson

Brewers WPA LVP Leaders

9 - Kameron Loe
8 - Yovani Gallardo, Casey McGehee
7 - Rickie Weeks
6 - Yuniesky Betancourt, Marco Estrada, Chris Narveson
5 - Corey Hart, Randy Wolf
4 - Zack Greinke
3 - John Axford, Prince Fielder, Carlos Gomez, Mark Kotsay
2 - Erick Almonte, Zach Braddock, Shaun Marcum, Nyjer Morgan
1 - Ryan Braun, Craig Counsell, Tim Dillard, Sean Green, Danny Ray Herrera, Brandon Kintzler, Jonathan Lucroy, Takashi Saito

There are 8 players who have appeared in a Brewers uniform this year but have earned neither an MVP or LVP: Brandon Boggs, Mark DiFelice, Mat Gamel, LaTroy Hawkins, Wil Nieves, Jeremy Reed, Mike Rivera, and Mitch Stetter.

Of course, many of those MVPs and LVPs are for pretty inconsequential performances - a couple of meaningless hits in a lackluster loss or an 0-4 day in a blowout win. I wanted to measure the number of games that truly killed or lifted the team, so I decided to make the cutoff a .250 WPA - a quarter of a win added or taken away. Here are the standings for those games:

Most games with .250 WPA or above

8 - Prince Fielder, Randy Wolf
5 - Yovani Gallardo, Shaun Marcum
4 - Chris Narveson
3 - Mark Kotsay, Nyjer Morgan, Rickie Weeks
2 - Yuniesky Betancourt, Zack Greinke, Casey McGehee
1 - Ryan Braun, Tim Dillard, Carlos Gomez. Corey Hart. George Kottaras. Kameron Loe, Jonathan Lucroy, Josh Wilson

Most games with -.250 WPA or below

6 - Yovani Gallardo, Kameron Loe
5 - Marco Estrada
4 - Randy Wolf
3 - John Axford, Chris Narveson
2 - Zach Braddock, Zack Greinke
1 - Erick Almonte, Tim Dillard, Sean Green, Corey Hart, Shaun Marcum, Mike McClendon, Takashi Saito

This is more for fun than anything else, so I won't comment much on it. There were a few things that jumped out to me, though:

- Prince Fielder has put the team on his back, dog. He has eight games where he was worth at least a quarter of an expected win, and no one else has more than three. His clutch-hitting woes from last year are regressing back to the mean, and it's been really fun to watch.

- Ryan Braun is the yin to Prince's yang: He only has one massively game-changing hit to his credit (the pinch-hit homer versus the Marlins), but despite coming up in tons of high-leverage situations, he's only been the LVP once. It's probably a good indication that he's doing a lot of table-setting and BMIRing, which is still extremely valuable.

- These numbers obviously don't come close to measuring how well someone is actually pitching, or should be pitching. But they do show what a general non-contributor Zack Greinke has been. Leaving aside his defense-independent stats, his WPA numbers are mediocre overall and terrible for an ace.

- Yovani Gallardo is either fantastic or awful - not much in between. So, basically an extreme version of what he's been his entire Brewers career.

- The less said about Kameron Loe, the better. Wait till tomorrow, though: Those numbers are even less kind to him.

Anything you see in there, or any questions about game-by-game WPA numbers? I have a spreadsheet of data to play with if there's something in particular you're curious about.

Tweet Comment 16 comments | 1 recs |

Do you like this story?

Comments

is there any way to statistically control for the likelihood that a given player, based on his position, will have a significant impact on winning/losing? for example a starting pitcher v. outfielder, closer v. pinch hitter, & c.

by sauvage on Jul 11, 2011 5:09 AM CDT reply actions

I suppose you could look at league averages

Which would be simple enough to figure out on a season-long basis, but a lot more time-intensive on a game-by-game basis.

Since WPA doesn’t measure defense, positions wouldn’t necessarily have any impact on your WPA output – it might be more productive to look at the spot in the lineup instead. As for pitchers’ roles, relievers (and especially closers, whose usage is essentially uniform across the league) are pretty similar in their impact: You either maintain a lead (or a narrow deficit) and get a marginal WPA boost (the amount goes up as you get later in the game), or you blow it and you’re almost instantly the LVP.

The best stat for measuring all this may be Leverage Index – I’m sure some interesting stuff could be done (or already has) with that, but I’m not mathematically skilled enough to tackle that.

by Cheeseandcorn on Jul 11, 2011 8:22 AM CDT up reply actions

No defense

It’s too bad that WPA doesn’t account for defense, because it really makes Yuni and Kotsay look much better than they should.

However, it is interesting to see that Kotsay has as many +0.25 WPA days as any position player not named Fielder despite not being an everyday player. If you normalize the WPA MVPs by games played for position players (which gives you the percent of games a guy played in that he was the MVP by WPA), you get:

WPA MVPs/GP:
Fielder – 8.7%
Morgan – 5.6%
Kottaras – 5.3%
Kotsay – 4.8%
Wilson – 4.0%
Weeks – 3.3%
Yuni – 2.4%
McGehee – 2.2%
Hart – 1.6%
Lucroy – 1.4%
Gomez – 1.3%
Braun – 1.2%

If you take the two guys who haven’t appeared in more than 25 games out (Wilson and Kottaras), it’s surprising that Kotsay is actually 3rd on the team. Also note that although Morgan has appeared in 8 fewer games than Kotsay, they both have the same number of PAs (175).

Of course, as you noted, WPA doesn’t exactly equate to value in any sense of the word. Just the fact that Braun is, by the normalized measure, the everyday player with the lowest rate of WPA MVPs per game should make that pretty clear.

by JLHanke on Jul 11, 2011 9:52 AM CDT reply actions

That's a great addition to the stats

I thought about calculating it by percentage of games played, but I got lazy, so thanks for doing that. I have some more hitting-specific stats to post tonight that shed a little more light on the position players’ WPA value. They make Braun look a lot better than these do – and Betancourt a lot worse.

As for Kotsay, it’s worth noting that he had more +0.25 WPA games in the four games of this weekend’s Reds series (2) than he had had in his 58 games over the entire rest of the season (1).

by Cheeseandcorn on Jul 11, 2011 10:02 AM CDT up reply actions

Sample sizes!

I would think WPA correlates pretty highly with BA w/RISP. That’s just about the only thing it’s measuring for all position players.

Founder of the BCBCU - Est. 2011

by ecocd on Jul 11, 2011 1:17 PM CDT up reply actions

That and BA in close-and-late situations, yeah.

Like I said, it’s not going to tell you much about who’s actually good, or help you predict who’ll be good going forward. It just helps put a number to the idea that “I feel like that guy has come up with a lot of huge hits for us” or “That guy always gets out in big at-bats.”

by Cheeseandcorn on Jul 11, 2011 2:10 PM CDT up reply actions

3rd List

You have the MVP leaders, and the LVP leaders, but, if you take this concept to fruition, the “true” leaderboard is the aggregate of the two. For example, Prince has 15 MVPs – 3 LVPs, resulting in an aggregate “score” of 12. Taking just some of the top guys…

Prince – 12
Braun – 6
Rickie – 5
Marcum – 5
Gallardo – 3
Wolf – 2
Gomez – 2

If nothing else, it corrects the list for wild swings between MVP and LVP performances, like Gallardo or Wolf.

by mpbMKE on Jul 11, 2011 1:43 PM CDT reply actions 1 recs

This is what I thought, also.

It really shows in the second list as well with the pitchers. You can see, in the differences between the two lists, the ratio between great pitching performances (usually, for 2.5+ WPA, keeping the opponent’s score low when your own offense is quiet) and terrible performances.

From those differences, Marcum (5:1) and Wolf (2:1) have been the only two dependable starting pitchers. If you NEED a win, they have been the best bets, while the other three have been basically a wash. I am indifferent to this fact, I guess, since it’s better than you’d expect from your #5 starter, but much worse than you’d want from your ace and your #3.

by OlyKnows on Jul 11, 2011 2:48 PM CDT up reply actions

I'm a little shocked

at the lack of WPA love for Plush, but I think that’s because he’s usually getting on base in critical situations, not driving runs home.

by Archibaldcrane on Jul 11, 2011 2:21 PM CDT reply actions

Flawed

There’s no way Axford deserves to be recognized for three LVPs and no MVPs.

by nullacct on Jul 11, 2011 2:44 PM CDT reply actions

Closers don't really get much credit with WPA.

If they do their job, they get a slight uptick, but if they blow it, they become the LVP for the game.

Pujols is the Barack Obama of baseball.

by sjlee on Jul 11, 2011 2:49 PM CDT up reply actions

Exactly.

Unless a closer comes in mid-inning or goes multiple innings, the highest WPA he can get in a game is .202, I believe. That’s rarely enough for an MVP, and it falls short of the (admittedly arbitrary) .250 standard I set.

But a good closer will show himself in the cumulative season-long WPA totals, rather than in game-by-game totals. That’s the case with Axford: He has a 2.15 WPA for the season, tops on the pitching staff and behind only Prince and Braun.

by Cheeseandcorn on Jul 11, 2011 2:58 PM CDT up reply actions

I agree, Axford's contributions are swept up in the WPA vacuum

since other people have to contribute to get the game to Axford, meaning that it’s hard for him to singlehandedly win the game. On the other hand, anything he contributes negatively will be weighted disproportionately

by OlyKnows on Jul 11, 2011 2:54 PM CDT up reply actions

So maybe the best measure

Would be tWPA/G ..?

by nullacct on Jul 11, 2011 4:16 PM CDT up reply actions

There's nothing wrong with it as long as you recognize that minimizing LVP is the measure of success for a closer

You could go through and find the probability of winning each game when he entered and create some kind of “potential WPA” measure which would measure how much WPA he could’ve lost in a given appearance. If a closer is only coming in on 3-run leads, he doesn’t deserve as much credit as one always coming in on 1-run leads.

Founder of the BCBCU - Est. 2011

by ecocd on Jul 11, 2011 4:49 PM CDT up reply actions 1 recs

Well-put. Rec'd

http://www.mlbsoup.com

by tcyoung on Jul 11, 2011 5:40 PM CDT up reply actions

Comments For This Post Are Closed

FanPosts

Community blog posts and discussion.

Recent FanPosts

+ New FanPost All FanPosts >

NL Central Standings

	W	L	PCT	GB	STRK
Cincinnati Red Reporter	26	20	.565	0	Won 1
St. Louis Viva El Birdos	25	22	.531	1.5	Lost 3
Houston The Crawfish Boxes	22	24	.478	4	Lost 1
Pittsburgh Bucs Dugout	22	24	.478	4	Won 2
Milwaukee Brew Crew Ball	19	27	.413	7	Lost 1
Chicago Bleed Cubbie Blue	15	31	.326	11	Lost 11