Monday, October 16, 2006

Internet Baseball Awards: MVP

Last year, in a rare convergence of my mind with the collective of the BBWAA, the actual MVP awards went the way I would have voted them, or close to it. ARod was my choice as AL MVP, and while I gave a very slight edge to Derrek Lee in the NL, it was tough to argue with the choice of Albert Pujols.

I will not go through ten guys and give my opinions on the minutiae of their ranking; instead I’ll just discuss the candidates for the top few spots. In the AL, there are a number of worthy candidates. Looking solely at runs above a replacement level hitter at their position, the top guys are Derek Jeter (+72), Travis Hafner (+71), David Ortiz (+70), Carlos Guillen (+65), Joe Mauer (+64), Manny Ramirez (+64), Grady Sizemore (+62), and Miguel Tejada (+61). The ones who stand out here in my mind are Jeter, Ortiz, Hafner, and Mauer.

Jeter hit 347/413/487, 123 RC in 692 PA. Ortiz hit 313/409/631, 137 in 677. Hafner hit 313/439/667, 124 in 554. Mauer hit 318/433/507, 108 in 600.

While I am hesitant about how much weight to give to the information, Fangraphs does a great job of providing us with WPA data. Jeter was +5.98; noted “Clutch God” Big Papi was +8.04; Pronk was +4.44; and Mauer was +2.36.

How big are these differences, compared to our basic evaluation of each player’s offensive contributions? AL teams averaged 4.97 runs/game this year, which equates to a R-W converter of about 10.2 r/w. So I will multiply the wins figures from Fangraphs by 10.2 to put them into runs, and compare to runs created above average (an average HITTER, not position-specific):
These figures certainly bolster the standing of Jeter and Ortiz while dropping that of Hafner and Mauer.

One question that needs to be answered is “how bad of a shortstop is Jeter”? I am no fielding maven, and so I have to look at other people’s rankings, many of which are not yet available for 2006. I gather that a conservative estimate is that he is 10 runs below average at shortstop. But remember that Ortiz has no defensive value whatsoever, and while we have assumed that everyone is equal defensively (and the big assumption that all positions have equal offense + defense value, which may well not be true), remember that I have lumped 1B and DH together. So if we impose a penalty for DHing rather then playing first, I think that defense is a wash.

That leaves the “clutch” factor as the only thing left to be considered. But I don’t think it really helps Big Papi that much. He was 22 runs better through WPA then we expected; Jeter 19. And I can’t justify giving him the MVP award on that.

The wildcard is Johan Santana, who had an outstanding season. Jeter is +72 RAR and +53 RAA, Santana +82 RAR and +50 RAA. Santana’s runs are more valuable, as they occur in the lower run context that he creates. But giving Jeter credit for the “clutch” runs edges him back ahead. And so, in a close call, I see it:

1) SS Derek Jeter, NYA
2) DH David Ortiz, BOS
3) SP Johan Santana, MIN
4) DH Travis Hafner, CLE
5) C Joe Mauer, MIN
6) SP Roy Halladay, TOR
7) SS Carlos Guillen, DET
8) CF Grady Sizemore, CLE
9) SS Miguel Tejada, BAL
10) 3B Alex Rodriguez, NYA

There are two players who have gotten significant MVP hype who do not crack my ten are Frank Thomas and Justin Morneau. Quite frankly, I am mystified (but not particularly surprised, given the fact that some people campaigned for Shannon Stewart a few years ago) that those guys are even considered candidates. Thomas is a great player, a surefire Hall of Famer, and one who has oddly been praised at the beginning of his career but underrated in the middle to end. Certainly Thomas’ run from 1991-1997 or so was brilliant, but he has been a very good player since then, albeit one who gets hurt a lot. But it seems to me as if his reputation was not at that level—it seemed that the perception was that Ken Griffey was still very good but always hurt, while Thomas was mediocre and always hurt. So I am glad to see the Big Hurt getting respect at this late stage.

Much of his candidacy stems from his status as the leading player on a playoff contender, and in fact he was the most valuable Oakland hitter. But a DH simply cannot be the MVP while creating 96 runs. Thomas ranks just thirteenth in the league in RG, and eighteenth in Runs Above Replacement (PRIOR to applying a positional adjustment). His WPA of +3.19 is only 3 runs better then would be expected. Big Frank is a valuable player, but not one of the top ten in the league.

Then there is Justin Morneau. On the surface, the reason for his candidacy seems to be the same as Thomas; big bopper on a division champ. He also has the glistening 130 RBI, second only to Big Papi. But in the sabermetric measures, it is no less mystifying. He ranks sixteenth in position-adjusted RAR at +47. His 7.3 RG ranks twelfth, certainly underwhelming for a first baseman who is staking an MVP claim. His +4.46 WPA is nine runs better then expected. But even if we give him credit for the full nine runs, moving him to +56, he still trails all of the hitters on my ballot, as well as Jermaine Dye, Vladimir Guerrero, and tied with Jim Thome. In fairness, those four are only separated by two runs, pretty negligible. But he’s still a long way away from Jeter, Papi, and Pronk.

What befuddles me is that Batting Average is perceived at least to be the holy grail at which the traditionalists worship. And Morneau’s teammate is the first AL catcher ever to win the BA title and the first ML catcher to lead both leagues. And yet Morneau may be an even bigger MVP candidate. I forget where exactly I saw it, but a study once showed that the most important factor in historical MVP voting has been RBIs. Morneau certainly has those, and that seems to be what is driving his candidacy.

In the National League, there are four players who I think are in the running: Albert Pujols, Ryan Howard, Miguel Cabrera, and Carlos Beltran. Let me do a chart of some pertinent data rather then writing it out:
“EXP” is the difference between estimated WPA runs and RC above average. Looking at this, it is tough to give it to anyone other then Prince Albert. He’d be the choice based on just going by the basic numbers, he has the best “clutch” performance, and he’s a better defensive first baseman then Howard. I jump Beltran over Cabrera because he is a fine defensive outfielder while Cabrera apparently is not all that at third.

1) 1B Albert Pujols, STL
2) 1B Ryan Howard, PHI
3) CF Carlos Beltran, NYN
4) 3B Miguel Cabrera, FLA
5) 1B Lance Berkman, HOU
6) SP Brandon Webb, ARI
7) SP Chris Carpenter, STL
8) SP Roy Oswalt, HOU
9) 2B Chase Utley, PHI
10) SS Jose Reyes, NYN


  1. The MVP voters did not always have an RBI fixation; it's more of a recent phenomenom. Since the founding of the modern AL MVP award in 1931, the American League has had 7 MVPs who made the top of the RBI charts and yet did not have that great an offensive season: Jackie Jensen (1958), George Bell (1987), Mo Vaughn (1995), Juan Gonzalez (1996), Juan Gonzalez (1998), Ivan Rodriguez (1999), and Miguel Tejada (2002). If Morneau wins, he would be the eighth, but also the sizth in the last twelve years.

  2. How are you defining replacement level? Comparing Jeter and Ortiz, you have Ortiz +14 in RC and -2 in RAR. So that means a repl SS is just 16 runs worse than a repl DH? That seems like far too small an adjustment. I'd think a replacment SS is probably 25-30 runs below a replacement 1B/DH.

  3. Thanks to Greg for the historical perspective on RBIs and the MVP.

    Guy, my definition of replacement level starts out at 73% of the league runs/out (equivalent to a .350 OW%). Personally, I think this is probably too low, and that 80% would be better, but for a long-time at least .350 was sort of a consensus in the sabermetric community and I made the decision that I would not fight that, and in working up the stats for my website would use the “established value”.

    Anyway, from there it is different for every position, based on a ten-year offensive positional adjustment. I believe that you participated in the great thread over at Tango’s blog about offensive/defensive positional adjustments, so you understand the strengths and weaknesses of this approach. Anyway, I lump 1B and DH together, as they draw from a similar pool of players (and this eliminates the absurdity of having the DH replacement level lower then the 1B replacement level). And that adjustment for 1B/DH is 119% of the league average. For shortstop, it is 86% of the league average.

    So in the 2006 AL, where the average is 4.97 runs/game, a DH is judged against a replacement who creates 1.19*4.97*.73 = 4.32 runs/game. A SS is judged against a replacement who creates .86*4.97*.73 = 3.12 runs/game. A “game” is 25.5 outs. If you had a shortstop and a DH who each made 1/9 of their team’s outs, the difference would be:
    (4.32-3.12)*162/9 = 21 runs
    Now if I had used .8, which I feel is more proper, it would be 24 runs, which is in line with the 25-30 runs that you speculate.

    You can feel the 21 runs is too few, and that’s fine, I can certainly see where you’re coming from. But what you cannot do is simply subtract the difference between Jeter and Ortiz’s RC difference and RAR difference. That would work if they made the same number of outs. But Jeter made 16 more outs then Ortiz did.


I reserve the right to reject any comment for any reason.