The Real MVPs

Update: For all the WPA haters, I did the same thing with VORP here.

There are many problems with the MVP award. The main one, of course, is these guys. The second constant issue is the term "valuable". Everyone has their own definition of the word, which makes things quite difficult. Can it come from a guy on a losing team? What about if there are multiple candidates on the same team- does that make the individually less "valuable"?

The way I see it, WPA was created for this debate. The Win Probability Added definition from Wikipedia: "[WPA] is the difference between the Win Probability when the player came to bat and the Win Probability when the play ended."

Well that seems like a pretty good way of analyzing the situation, doesn't it? A measure of how much a player contributed to his team's chance of winning the game- that's just what we're looking for. So let's just look at the WPA leaders, and be done with it, right?

Of course, it is not nearly that simple. As far as I can tell, there are three main things that WPA does not account for:

-Positional value. Having a shortstop who hits .335/.420/.530 is a whole lot more valuable than having a first baseman with the same line. VORP accounts for this, but the WPA numbers listed on FanGraphs do not. To deal with this issue, I developed a system that takes position into account. The positional adjustments, per 600 PAs, are on the right.

-Park factors. Another thing that is considered by VORP, but not in traditional WPA. This is also pretty simple to take into account as I have looked at previously. Here are the per 600 PA adjustments for parks with the most dramatic effects:


-Fielding. Oh boy. Any system trying to account for fielding is going to be imperfect, but it would be unreasonable not to include something. I've used the Hardball Times' fielding stats that were released at the end of August, as well the UZR numbers that MGL released around the All-Star break. Since this is obviously far from complete, I also included UZRs from '05 and '06.

So now we have accounted for the three main inadequacies in WPA- positional value, park factors, and fielding. The only other adjustment needed is a small one, accounting for the leverage of the average situation the batter has come up with, in order to control for hitters who have come to bat in an inordinate amount of "high leverage" situations. All of these adjustments can be seen in the various columns below.

Finally, we can apply all of this to this year's MVP races. Let's start with the straightforward one:

AL MVP:

Rodriguez starts off with a huge lead in raw WPA, and does nothing to relinquish it. I think we tend to forget that Yankee Stadium is actually a pretty significant pitchers park, which helps him here. Also relevant is the fact that although he it taking a slight hit for his position, it's not nearly as much as RFs Ordonez and Guerrero.

I think the top four are pretty much what you would expect (although it's interesting that almost half of Ichiro's value comes from his fielding). Polanco, though, is probably not someone that would be expected to show up on this list. He really does belong here though. All of the numbers agree that he is an excellent fielder. More importantly, he's hitting .340/.389/.459 as a second baseman, which is quite impressive. If you put Polanco in pinstripes and throw a "2" on his back, he's in every MVP discussion. Of course, that's now things work, so he'll have to settle for being on this list.

Now, for the interesting one:

NL MVP


I understand that the names not on this list may be more interesting to talk about than those which are, but I'll get to that in a second. Interesting to note that all four of the top NL guys get huge boosts from their fielding. This is because the NL is so bunched up at the top- there are 9 guys within 1.2 of the WPA leader. When that happens, getting an extra 14-18 runs from your fielding vaults you right to the top.

Wright and Utley are neck and neck, and the winner will be decided over the final weekend. Wright suffers a little from the positional adjustment, but gets a boost because Shea is so tough to hit in, and obviously gets a huge boost from his fielding (THT's stats have him at +28 runs).

It's too bad that Utley missed so much time, because he's really had a remarkable season, both with the bat and the glove. The only blemish on his line is the park adjustment, and that's obviously not his doing. Because of the time he missed I didn't expect him to do so well in this analysis, but that just shows how good he's been when healthy.

I talked about Tulowitzki yesterday, so I won't get into that again. Interesting thing about Pujols is that he gets killed in the positional adjustment, but he makes it up (and then some) with his fielding prowess.

OK, now for the three guys who didn't make the top five, but have been included in most discussions:


Prince Fielder leads the NL in WPA. He then loses almost 22 runs because of the position he plays, and how poorly he plays it. That's the thing about this system- you can't just be a good hitter and have that be your entire candidacy. Fielder's only value is in his bat, and you would have to have an overwhelming year at the plate to be the MVP while playing first base poorly. While obviously impressive, Fielder's 2007 offensive campaign doesn't fall under the category of "overwhelming".

Holliday is somewhat similar, but his weakness is the park he plays, rather than his fielding skill (or lack thereof). He's been good (and those 131 RBIs are going to be real tough for the voters to pass up if the Rockies somehow sneak into the playoffs), but he really doesn't belong in the MVP discussion.

Rollins is definitely the most interesting omission, as he would have to be considered the co-favorite at this point (along with Wright) to win the award.

Is this really even worth discussing? As far as I can tell, the main thing in Rollins' candidacy is this stupid 20-20-20-20 thing. Who cares? Can we talk about how the guy walks less than half as often as Wright, while having a lower slugging percentage and getting destroyed in BA, rather than his possible inclusion in some gimmicky club?

The two factors that this table doesn't show, but I've included before, are fielding, and position. Rollins obviously has the positional value thing on Wright- with this huge number of PAs, there's about an 11 run difference between SS and 3B. But none of Rollins' numbers indicate he's anything other than an average fielder, while all of Wright's 2007 numbers are stellar. I just don't see how Rollins even comes close to making up the huge gaps in the hitting categories.

So there you have it- A-Rod takes it in the AL, with Utley and Wright battling it out down to the wire in the Senior Circuit. The official MVP award is quite different than this- it focuses mainly on batting average and RBIs, while ignoring useful batting statistics and fielding. It also won't be announced until mid-November, when everybody will be preoccupied complaining about the BCS anyway. Much more timely, fan friendly, and intelligent will be the Internet Baseball Awards, which you can vote on over at BP after the World Series ends. To give you an idea of how things work over there, Morneau finished 6th in the AL voting last year, which is probably all the proof you need that it's a superior system to the actual award.

Photos: Wright, A-Rod, Utley, Rollins.

Update: After reading the BBTF Discussion of this post, I wanted to mention a couple things I should have included.

I should have mentioned that UZR is screwed up for Coors, as MGL mentions in the comments of the post in which he released them. UZR had Tulowitzki at +14 runs through the ASB, which is probably too high. He is probably slightly overrated by this system in terms of fielding. It should be noted that a lot of Tulowitzki's value has come from his brilliant performance in"clutch" situations- FanGraphs has his BRAA as 13.86, but his raw WPA is 2.85- his performance in "Clutch" situations has caused his WPA to be 1.43 higher than would be expected.

It's also a fair point that I should have included runs in the comparison between Rollins and Wright, since there was a column for RBIs. This was an honest mistake (rather than trying to make Wright look better)- I was taking the numbers from B-R, and just started with doubles and moved down the line. I didn't notice that runs wasn't there, as it's not really something I take into consideration when comparing players (neither is RBI).

Wright has scored 110 runs, Rollins has scored 135. This is somewhat amazing, as Wrights OBP is 68 points higher. Part of the reason for the large discrepancy is Rollins has had an amazing 761 PAs, compared to Wright's 693. But a bigger factor is the guys who hit behind Rollins. Utley, Burrell, Howard, and Rowand have combined for an amazing 267 extra base hits. Which is exactly the problem with runs- if you put Wright at the top of the Phillies' lineup, he probably would have scored just as many.

Update 2: I didn't incorporate BP's fielding stats into my fielding numbers, which was silly. If I do, Russ Martin jumps to fourth in the NL. The AL top five stays in the same order.

NL MVP Watch: July Edition

Following up on yesterday's post on the AL MVP race, today I'll take a look at the NL.

For offense, my stat of choice is adjusted WPA. Adjusted WPA is a variation of the more widely used WPA (Win Probability Added). The only difference is I adjust it based on position, as explained here. The raw WPA data is taken from FanGraphs.

For fielding, I have used MGL's UZR, which can be found here. UZR is just one of many fielding metrics, but it is thought to be pretty accurate, and is released at points nthroughout the season.

The NL's top 10 in adjusted WPA, through games of 7/25:

Prince has had a great year, and at BetUS he is the favorite to win the MVP. But in reality, his raw WPA isn't all that much better than these other guys, and when his position is factored in, Prince has actually been the sixth most valuable hitter in the league.

(I know it is kind of screwy that so many Colorado guys are in the top 10, and I'm not thrilled with it either. I don't know how I should fix this with a WPA park factor adjustment; if you do, please e-mail me. There is also a problem with Coors for fielding, as MGL himself says here. If I were smart enough to fix this I would, but that is certainly not the case. Anyway, take the Colorado numbers with a grain of salt.)

The top 10 fielders thus far:

And combining all this, the NL's ten most valuable players to this point:




Now I know this list is weird and stupid and probably not correct, because of the Coors thing. And I think that explains Matsui and Holliday and Helton and their lofty places on this list. So let's ignore that (I'll deal with that in a second). Let's deal with the guy at the top of the list.

Troy Tulowitzki has an OPS+ (which takes park effects into account) of 100. He is, by this measure, exactly an average hitter. We can work with that, since both WPA and UZR are measured against the baseline of an average player.

First, he plays shortstop, which helps him. This is valuable regardless of park factors; he gets +0.31 in my adjustment system because of this.

Not only does he play shortstop, he does it really, really well. John Dewan, the author of The Fielding Bible, released his rankings for this year a week ago, and Tulowitzki is listed as the top SS. He is third among SS in THT's RZR rating, but has made about twice as many out of zone plays as the two people ahead of him, Reyes and Vizquel.

As I used originally, he is +14 in MGL's system, which leads all SS (both leagues). So, for the sake of this discussion, I'll put his UZR at +12, which puts him slightly ahead of the rest of the shortstops in UZR. So 1.2 more wins, added to his .3 from before, and we're at 1.5. Which, obviously, is very, very far away from the 4.38 I orignally had him at.

Troy Tulowitzki has been the most clutch hitter in the National League this year, and it is not close. FanGraphs has a stat called "Clutchiness", which measures this. This stat is the difference between a player's actual WPA, and what would his WPA would be expected to be based on his OBP and SLG. Tulowitzki's is at 2.15 in this category; the next closest is Adrian Gonzalez at 1.37.

Now, I'm not going to get into the whole clutch debate. But, when we're looking at how valuable a player has been, rather than if they have a natural "clutch" ability, I think the timing of their hits is very relevant. And Tulowitzki has certainly been clutch so far this year, as seen in the stats shown here.

I wondered how much of an effect Tulowitzki's extra inning performance had on his WPA, so I looked through his game log. In his 12 extra inning plate appearances, Tulowitzki has a WPA of 1.93. This I find amazing. In extra innings, Troy Tulowitzki- defined previously in this post as an average hitter- has contributed almost two more wins to the Rockies than an average hitter would be expected to. Contributing this is (all in extras) a game-tying two out double, a two-run homer, a two out RBI single, a two out RBI triple, a solo HR, and an RBI single.

(After going through this, and realizing the Rockies ended up losing most of these games, I remembered Kurkjian saying something about Tulowitzki on BBTN. Sure enough, it's here:

"Three times in one week, Tulowitzki hit a go-ahead home run in the ninth inning or later, and his team lost all three games. That has happened four times in baseball this year, once to Magglio Ordonez and three times to Tulowitzki. He is the only active player to hit three go-ahead home runs in the ninth inning or later in a loss in his career."

So there you have it.)

Anyway, if Tulowitzki has been worth two extra wins because of clutch hitting, we're up to 3.5 total. Which, even with the conservative assumptions I have made, puts him right behind Chase Utley and Aramis Ramirez in my top 10 most valuable players.

This has gotten completely out of hand, but to conclude, I'll say this: it is my belief that Troy Tulowitzki has been one of the three most valuable players in the National League thus far this year.

For the sake of simplicity, here's the non-Colorado top 10:

When I went through this exercise in early June, Utley also came out on top. I guess this happens when one of the top fielding second baseman in the league (the +6 UZR I use for Utley is probably conservative- Dewan and THT have him first and second in all of baseball, respectively) is hitting .334/.412/.582. Utley may not win the award- the Phillies only have about a 20% chance of making the playoffs- but there's certainly a pretty good chance he'll be deserving of the award at the end of the year.

Aramis Ramirez, who is having a good year despite the fact he never takes a walk, is helped out by a Clutch rating of 1.34; this hit plays a big role in that.

After hitting .248/.347/.426 over his first 35 games, Albert Pujols has hit .351/.455/.631 in 62 games since.

Pictures taken from here, here, here, and here. Stats taken from Baseball Reference and FanGraphs.

Thanks to The Big Lead and With Leather for linking to my AL MVP post yesterday.

UPDATE: Chase Utley broke his hand in the Phillies' game against Washington this afternoon.

AL MVP Watch: July Edition

About six weeks ago, I took a look at which AL hitters had been most valuable so far.

To do think, I used two statistics- one for batting, and one for fielding. For hitting, I used adjusted WPA, my varation of the widely used WPA (Win Probability Added). Adjusted WPA takes into account a player's position. A 4.00 WPA from a SS is more valuable than an equal WPA from a 1B or DH; this is shown in adjusted WPA. WPA data is taken from FanGraphs.

For fielding, I have used MGL's UZR, which can be found here. UZR is just one of many fielding metrics, but it is thought to be pretty accurate, and is released intermittently throughout the season.

First, the current top ten in adjusted WPA:

This is where Jeter reaps the benefits of his position, ad he moves from fourth in WPA to second in adjusted WPA. Unfortunately, you have to actually play defense. Which brings us to fielding, and the top 10 in UZR:



Some other notables:

Combining the two (and dividing UZR by 10, as 10 runs is about equal to one win, we get the ten most valuable players thus far:

A-Rod has what is probably an insurmountable lead. Even if Jeter played defense like Adam Everett (UZR of 11), he would only be in second (.6 behind Rodriguez). And his defense is much closer to Manny (-14) than Everett.

Sizemore's position and unbelievable defense allow him to vault from ninth in straight WPA to third in my rankings.

Unfortunately, there is no UZR for catchers, which is why Martinez and Posada have UZRs of 0. I found it interesting that the league's two top hitters (Magglio and A-Rod) both play outstanding defense.

Photos taken from Why Don't We Get Drunk and Blog, USA Today, here and here.

UPDATE: My NL MVP Watch is now up.

MVP Watch, Part 2: NL

Saturday I took a look at which AL players have been most valuable thus far this season. Today, the NL.

Check out Saturday's post for the specifics, but essentially I am using adjusted WPA do determine each hitter's value at the plate, and Ultimate Zone Rating (UZR) (found here) to measure their fielding value.


The top 30 in unadjusted WPA, through Saturday's games:



Among the top 30 are six shortstops. That becomes a big factor when these numbers are adjusted for positional value and defense:

The thing that jumped out at me here was the quality of the fielders. For the AL, the average UZR for the 20 players I looked at was -.05. In the NL, the average for the top 30 is 1.37. Among the six afformentioned shortstops the average is just .5, but without Hanley Ramirez and his -9 UZR it is 2.4.

After WPA, the positional adjustment and UZR (divided by 10, so it measures wins rather than runs) are added together, the most valuable players in the NL this season are...

(Quick note: you may notice that a few of the players on this "top 30" list haven't been particularly valuable (see: Young, Dmitri). This is the one problem with this: the top 30 in Value would look a little different, ad players who are 31-60 in WPA would take the place of some of the players who are 21-30 on the following list. Even so, I think the top 20 on my list would be the same as the top 20 overall.)

Despite being just 9th in unadjusted WPA, Utley vaults into first thanks to a combination of his superb fielding and the negative adjustments of the four unadjusted WPA leaders.

Despite relatively minimal hype, Tulowitzki comes out as the top shortstop.

The unadjusted WPA leader, Fielder, really gets killed by his position and his fielding. His -6 UZR is especially unfortunate considering the superior fielding of his peers.

If you have any questions, comments, or suggestions, please e-mail me at jnwheat@emory.edu, or post something in the comments section below.

Finally, I would like to thank Deadspin, BallHype and Baseball Musings for linking to Vegas Watch over the last couple weeks.

MVP Watch, Part 1: AL

Recently I did a lot of work with Win Probability Added (WPA). I developed a system to adjust WPA for position (as a 1.5 WPA from a C is more valuable than the same WPA from a 1B).

Then, a couple days ago, at The Hardball Times, I discovered that Ultimate Zone Rating (UZR) had been posted for the first two months of the year.

I thought it would be an interesting excercise to combine these two statistics and see which American League players have been most valuable to their teams over the first two months of the season. I'll do the NL tomorrow.

First, I took the top 20 AL hitters in WPA from FanGraphs.

Taking these, I then adjusted them based on position. Note that I have changed my adjustments so they are now based on PA, rather than AB. Here is how much each of these players' WPAs changed with this adjustment.

In short, adjusted WPA adjusts a player's WPA based on what an average player at his position would have done in the same number of plate appearances. A more detailed explanation can be found here.

Adding this number to each player's WPA gives us Adjusted WPA. But that is only half the story- fielding must also be taken into consideration.

UZR is given in runs. When adding it to Adjusted WPA, I divided it by 10, since, generally 10 runs is equal to one win.

Vlad has been so much better than everyone else in terms of regular WPA, it would be almost impossible for anyone to make up for it with the positional adjustment and UZR.

Grady Sizemore does makes a valiant effort, as he is first in UZR (+12) to this point in the season. This allows him to move from seventh in WPA to second in value.

The player most adversely effected? Wouldn't have been my first guess- Derek Jeter. The controversy about Jeter's fielding value has been well documented. Since he indisputably plays SS (not even Neyer could argue that point), his bat has some extra value in the posititional adjustment.

Then comes UZR. There are 584 players listed in Lichtman's ratings. Derek Jeter comes in 574th, at -7 runs.

To me, this combination of adjusting WPA and incorporating fielding stats is very interesting. Casey Kotchman, the Angels' first baseman, loses .47 becasue of his position, but gains almost all of it back with his UZR of 4.

I always thought this was one of the most difficult parts when comparing different positions. Not only do you have to compare each player's bat's value based on their position, but how valuable they are defensively at that position.

This is very difficult to do, but I feel this system does that pretty effectively.

If you have any questions, comments, or suggestions, please e-mail me at jnwheat@emory.edu, or post something in the comments section below.

Back with the NL tomorrow.