Ode to the MVP

A reader, Brian Rodriguez, writes:

I used to think the MVP should go to the best player (e.g., a-rod this year), but lately I've come to believe that perhaps it should depend on how the team does. After all, a-rod's presence has his team finishing last this year. How many fewer wins would they have without him? It doesn't really matter because Texas would still finish in last, albeit with a worse record.

Perhaps we can examine how taking away a certain players "wins created" would affect his team's league standing. After all, helping his team get into the playoffs should be the #1 goal of any MVP candidate. So, to flip it around, last year's al mvp Ichiro played for a great team, but if you took away his wins created, perhaps the team would have finished with 110 wins instead of 116 or whatever it was. The mariners would still be in first place in the al, and that would argue against Ichiro's value as far as mvp goes. But a player whose play made the biggest difference in a team's standing seems as if they would be more valuable to me.

It's similar to the argument Branch Rickey as GM of the Pirates presented to Ralph Kiner after Kiner asked for a raise: "We finished last with you; we could have finished last without you!"

Interesting question. The funny thing is I used to think that who was voted the MVP should depend on how well his team performed. Now, I think that the best player should be elected, but more on that later. There is a basic problem with the award in that what the MVP should be is never clearly defined, or at least if it is, that definition is never really followed by the voting press. It's the most valuable player to whom? His team, the league, baseball, humanity, a higher being, who? The voters' answer to this has seemed to change over time. The MVP now seems only to be awarded to a star player on a playoff-caliber team. Ernie Banks won the award back-to-back in 1958 and '59, both years playing for a sixth-place team (in an 8-team league). Mark McGwire broke Maris' home run record and had an historically dominant year in 1998 but lost the MVP to Sammy Sosa, in part due to the perception that Sosa played on a winner and McGwire did not. Sosa's Cubs were 6.5 games ahead of the Cardinals in the NL Central. The Cubs finished second but won the wild card on a one-game playoff with San Francisco and then were swept by the Braves in the Division Series, scoring only four runs in three games.

Also, whom the voters would consider for the award has changed over time. At one point, a starting pitcher could be, and often was, voted the MVP. The last starting pitcher voted MVP was Roger Clemens in 1986. The thinking was that a) starting pitchers were given their own award, the Cy Young, in 1956 and b) they were not as valuable since they aren't everyday players. Starting in the '80s relief pitchers started winning MVP awards, but it seems to have been a fad since none have won one since Eckersley in 1992. Note that the season record holder in saves, Bobby Thigpen in 1990 with 57, only finished fifth in the MVP vote. However, there is a lot of talk this year about Smoltz getting some votes. Also, among position players RBI has been a statistic that voters seem to identify with MVPs, even when cognoscenti have anointed another for the award. There is also the odd phenomenon of players being snubbed with MVP-type numbers after having won the award (e.g., Giambi in 2001). Ted Williams was overlooked a number of times. It seems that the voters are looking for a break-out type player.

Brian's suggestion of a "wins created" statistic is highly topical given that this year Bill James finalized his Win Share system. The number of Win Shares per team is calculated by taking the team's wins and multiplying them by three. Then, these Win Shares are distributed to the team members based on their performance offensively and defensively. Defensive Win Shares comprise 52% of the total and are divided between the pitchers and the defensive position players. Given this arrangement, pitchers are seen as less valuable (i.e., receive fewer Win Shares) than position players. Perhaps this perception or realization (depending on your stance on this issue) is what led voters to exclude or overlook starting pitchers and then relief pitchers as candidates. Anyway, the Win Shares system is based on a player's performance above the replacement level whereas previous systems (especially Total Baseball's Total Player Rating) were based on his performance above the average player. The result is that a player's Win Shares are a good means to measure his value as a player. You could divide the player's Win Shares by three and in effect have the number of wins (above replacement level) that the player contributed to the team.

Now, back to the team-performance argument. If a player contributes ten to fifteen wins to a team, it does not matter if that team is first last or in the middle. Each player is helping his team win or at least trying to. How much more valuable is a guy who is having a tremendous year when the team around him is sucking eggs than a guy who is having a very good year on a good team. Let's use that as our premise and look at some MVP races.

Ichiro Suzuki had 36 Win Shares in 2001, which translates into 12 wins attributable to him. The Mariners finished 14 games ahead of the A's. How much was Suzuki really worth to that team? Ichiro had a very good year and was worthy of the award, but Jason Giambi had 38 Win Shares, led his team to the playoffs, and seemed to be the type of leader that the voters usually love (he was pointed to as the player that helped the most in turning around Oakland's season). But without Giambi (i.e., his Win Shares), the A's would have finished 89-74 and would have still been the wild card. Hmm, that does not seem to stick to our premise.

Let's look at McGwire again. Without his 41 Win Shares, the Cardinals would have been fighting Pittsburgh for last in the NL Central rather than having been an above-average team. Without Sammy Sosa's contributions (35 Win Shares) in 1998, the Cubs would not have gotten the wild card. But by the same token, without Mark Grace, Mickey Morandini, Jose Hernandez, Henry Rodriguez, Kerry Wood, etc., the Cubs wouldn't have made the playoffs either. Does that make each Cub who contributed to a win (since the Cubs won the wild card by one playoff game) more valuable than McGwire? I think not. The difference here is that McGwire had a statistically superior year to Sosa in 1998, whereas Giambi and Ichiro's 2001 season's were so close, it's harder to call. That sort of adds a little gray area to our premise.

Now let's look at Alex Rodriguez and Juan Gonzalez in 1996. This is pointed to as one of the worst MVP results in recent memory. Alex Rodriguez was unbelievable, as always, that year, and earned himself 36 Win Shares, but the Mariners finished 4.5 behind the Rangers in the West and 3 behind the O's for the wild card. Juan Gonzalez had a good year (21 WS), racked up a ton of RBI, and after Texas won the division, he won the MVP. Without Gonzalez, the Rangers would in theory have finished behind Rodriguez's M's, but according to James he was only the third most valuable player on his team (Pudge Rodriguez, 26 WS and Rusty Greer, 23). The problem with the "where would the team be without player X" argument is that the argument may work for too many people. It's a spurious argument in that it doesn't necessarily measure a player's individual value to the team, rather than how much a good player affects his team's record. It leads to poor decisions like Gonzalez. Also, in all these cases the effect of removing a player from a team assumes that a replacement-level player would be inserted. Most if not all of these teams had other players in their organization who could have filled in more capably than that.

Win Shares is somewhat problematic in that the calculations required to get the results are monstrous. It also uses seemingly every possible metric in every facet of the sport. Therefore, it's difficult to have a running total during the year. Baseball Prospectus has a statistic called VORP (Value Over Replacement Player). Here are the VORPs for the top-ten position players and top-ten pitchers in baseball:

Position Player
Bonds, Barry     90.4
Rodriguez, Alex  68.7
Giambi, Jason    58.1
Kent, Jeff       56.1
Giles, Brian     55.4
Thome, Jim       52.3
Edmonds, Jim     50.3
Soriano, Alfonso 50.1
Walker, Larry    49.9
Berkman, Lance   49.5

Pitcher
Lowe, Derek      66.8
Schilling, Curt  57.8
Moyer, Jamie     55.6
Johnson, Randy   55.6
Martinez, Pedro  53.4
Halladay, Roy    48.9
Washburn, Jarrod 46.4
Glavine, Tom     45.4
Zito, Barry      44.9
Oswalt, Roy      40.8

First, you may notice that the position player's value ranges are higher than the pitchers (defense is not included for position players). Second, you will see that Barry Bonds is playing in an entirely different league. I don't care where the Giants finish, if he doesn't win the MVP, it's a gross miscarriage of justice. Alex Rodriguez leads the AL, but a case could be made for Derek Lowe (who won't win because of pereception) or Jason Giambi. We'll have to see.

So where does that leave us? It is my considered opinion that the MVP award should go to the best player. Any and all statistical means should be employed to make that decision, but each statistic should be used only as a tool not the sole determining factor (that is, voters stop relying on RBI to pick your winner). If it's too tough to call, sure give it to the guy on the playoff team, but get the right guy. Not Gonzalez in 1996.