Monday, October 1, 2007

The Real MVPs Revisited

I knew some people didn't like WPA, although I may have underestimated how strongly people felt about it. "The Real MVPs" has 35 comments so far, and only 14 of those have been me responding to people. There was also the BBTF discussion. Here's my favorite comment, from Anon:

"Interesting analysis. However, it is fundamentally unsound to include park effects in MVP considerations. For an award, players should be evaluated based on their actual performance rather than what they might have accomplished on a level playing field."

I cant say I follow that logic, but to each his own, I guess.

Since I don't entirely disagree with the complaints about WPA, I thought I'd go through the same exercise, replacing WPA with VORP. With the understanding that the excellent "On Baseball and the Reds" blog beat me to it on this for the NL, here goes. Once again, the fielding stats are a combination of UZR's midseason numbers, and THT's numbers.

Surprisingly close. THT hates A-Rod's defense (-10), although that's balanced out by UZR's +7 through the ASB. Magglio is at +3 and +8, respectively.

Rodriguez is actually only third in the AL in MLVr, but has about 30 more PAs than Magglio, and 40 more than Ortiz. MLVr is also before the positional adjustments. Interesting as well that Granderson and Ichiro are way behind in VORP, but jump into the top five with their fielding.

Regardless of what this list indicates, the only real race in the AL is whether A-Rod will win unanimously or not. Since the Tigers missed out on the playoffs, I think that's a possibility. Ortiz definitely isn't getting any first place votes (Side note: I'm looking forward to Papi getting a big hit this week, then all the "Ortiz had a down year, but he's clutch so he came up big when it matters" stories; .445 OBP, .621 SLG is some down year, as Neyer pointed out). Nobody on the Indians is getting a first place vote- maybe Vlad will get a couple (he's 8th on this list, 62.6/-1.9/60.7).

On to the race that's still undecided:

Wright's chances to actually win this award are slim to none. When I vote (don't worry, it won't be anywhere meaningful), I'm going with him. Did the Mets collapse? Obviously. But Wright hit .394/.516/.657 in August, and .352/.432/.602. in September. Really can't pin that on him.

Some notable omissions from the top 5:

Also, since a couple people didn't appreciate my exclusion of Ryan Howard last time, he's at 53.6/-6.1/47.5.

I'd love to campaign for Hanley to win the MVP, I really would. But every fielding number on him is really ugly. Time to move to CF, buddy.

Rollins is 15 runs behind Wright in VORP. I'm not saying these fielding numbers are perfect (although I'm pretty sure they're the best that are publicly available right now, please correct me if I'm wrong), but there's no way Rollins' glove is worth 15 more runs than Wright's (after VORP helps out Rollins for playing SS). I guess beyond the #s we're giving Rollins points for his team winning 13 of their last 17 games, rather than 5/17. But I try not to get too caught up in that- Rollins' .298/.333/.542 September line is right around his full year averages.

Also a factor is Rollins' January proclamation that the Phillies were the team to beat in the NL East. That has certainly helped build the hype around his campaign for the award, but does that really make him more valuable? It makes pretty damn cool, but I don't know that it makes the Phillies better.

Rollins' teammate Utley comes out higher in this ranking. He didn't exactly catch fire in the last month either (.301/.377/.496). Over the 28 games he missed in late July/August, the Phillies went 15-13, vs. 74-60 the rest of the season (about a .017 difference in W%). Not terribly meaningful.

If the Rockies win today, I think it probably goes to Holliday (RBIs!). If they lose, probably Rollins. More on that tomorrow though.

Photo: matttracy.com.

7 comments:

Sky said...

Here's the proper response to Anon regarding park effects:

"It takes more runs to win a game in Coors than in PETCO. Therefore the contributions of Holliday in Coors did not help his team win as many games you're imagining they did based on his raw numbers."

Ideally, WPA would use separate win expectancy tables for each park and you wouldn't need to make any other park effect adjustments. (David Appelman's implementing that change for next year.) But even so, it's not hard to take WPA and apply a park adjustment (because WPA will overrate hitters and underrate pitchers in hitters' parks), like you did. It's the right thing to do.

Vegas Watch said...

I didn't know they were making that change with WPA for next year. Good stuff.

Of course you have to adjust it- that guy's comment was just so unreasonable, I was amazed.

Anonymous said...

VORP would make more sense if you were comparing players who play the same position. Is it fair that a RF'er who has better numbers all around is punished for playing that position? Compared to a SS who was 3/4th the RF'er was, but has a higher VORP?

It just makes more sense to me that VORP should be used in comparing players who play the same positions. Unless I'm missing something.

I've enjoyed both your blogs on the MVP candidates. Always looking to learn, thank you.

Sky said...

It's funny how the same questions pop up in different places on the same day...

You can divide up the process of assigning value into three pieces:

hitting
position
fielding your position

To most people, combining the second two into "fielding" makes the most sense. But VORP combines the first two together, leaving you to find a fielding metric you like in order to complete the puzzle.

But yes, you have to account for position somehow. If you have a RF and SS who both field their respective positions at a league-average level and hit at a league-average level, the SS is much more valuable, because he's actually a better defender. Put him in right and he'll save you runs. Put the RF at SS and he'll cost you runs.

VORP isn't saying who the better hitter is, it's saying whose offensive production is more valuable, given their position.

Here's a good article posted today. Read comment #45: http://ussmariner.com/2007/10/01/wrap-up-part-one/

By the way, you wouldn't need VORP if the players were playing the same position. Just a straight hitting value and fielding value.

Vegas Watch said...

Sky- thank you. I struggle to explain the logic behind the positional adjustments to people, but that is an excellent reasoning. As is Dave's on the USSM post.

joe said...

i read the similar article that used wpa, and i noticed you used the same fielding stats. i believe these fielding stats are compared to average players, while vorp is compared to replacement players. it doesn't seem logical to compare batting stats to replacement level players and fielding to average players.

Vegas Watch said...

Thanks for pointing that out Joe- I meant to bring it up in the post but obviously forgot.

It's not that it's illogical, it's just that it's kind of rough. The problem is that almost all batting stats are vs. replacement, and almost all fielding stats are vs. avg.

Post a Comment