Sunday, February 22, 2004

In his new book, The New Bill James Historical Abstract, James ranks the top 100 players in baseball history at each position. In the pages preceding the player rankings, he explains his ratings system, including a brief discussion of the limitation of using raw Win Shares as a player rating system. In addition to the player’s career Win Share total, James uses the player’s three best seasons, five best consecutive seasons, career Win Shares per 162 games, as well as a time line adjustment and a subjective element.

I’m not sure that I see the value in using the player’s three best seasons as an indication of his career value, although this would slightly favor Mattingly and Perez. Just as I don’t believe that the player with the greatest cumulative statistics should automatically be rated highest, I don’t feel that a few seasons of greatness should outweigh a career of consistency. I am going to use the five best consecutive seasons, however, as well as career Win Shares per 162 games, but I’m also adding two of my own measures.

While I understand that evaluating a player based on his career Win Shares per 162 games is intended to measure how truly valuable he is as a player (when he plays), and does not subject him to be discriminated against based on such factors as the decision of his manager whether or not to play him every day or to play him only 140 games per season (possibly to preserve his longevity). However, what this fails to take into consideration is that a player’s value is based on his contribution to winning games over the course of a championship season.

A player who plays 162 games at an equal level of performance to one who plays 140 games is, obviously, about 15% more valuable in that particular season. Therefore, what also must be considered is a player’s value per season over the portion of his career that he is a full-time player. Steve Garvey played a full season of 162 games in seven separate seasons over the course of his career. The other three players combined to do so only once (Mattingly, 1986). This fact must be taken into consideration to the extent that playing in additional games increases a player’s productivity. Obviously, the flip side is that the lack of rest may actually detract from his performance, but the statistics will bear this out.

Therefore, the first statistic that I’ve added in my analysis is the player’s average Win Shares per full-time season, with the one caveat being that it should count against a player if he wasn’t able to maintain a full-time job at least into his mid-30’s, barring unforeseen circumstances such as major injuries or, most obviously, premature death. Both Garvey and Perez continued as full-timers into their late-30’s, while Mattingly retired after playing his final full season at the age of 34, which I’ll consider acceptable. Hernandez, however, played his final full-time season in 1987 at the age of 33, a little early by my standards, so I’m going to throw his 1988 season, in which he played in only 95 games, into the mix. Incidentally, Mattingly played in only 102 games in 1990, and Garvey in only 100 games in 1983, seasons that were not strike-shortened, but are considered during their primes for the purposes of this analysis, so including 1988 should not severely damage Hernandez by comparison.

In this particular analysis, all four of these players’ careers consisted of between 12 and 14 full-time seasons, which certainly provides for ease of comparison. I did, however, have to make one adjustment due to the fact that each of these players’ prime years were affected by strike-shortened seasons: Perez in 1972; Perez, Garvey and Hernandez in 1981; and Mattingly in 1994 and 1995. To account for this, I made adjustments based on the extent that each of these seasons were abbreviated, thereby reducing Mattingly’s full-time seasons from 12 to 11.59, Hernandez’s from 13 to 12.63, Garvey’s from 13 to 12.68, and Perez’s from 14 to 13.62. As it turns out, this didn’t make much of a difference, as you can see that each of their season totals was reduced by 0.32 to 0.41 of a season. I was mildly surprised by this, but I guess I should have realized that Garvey and Hernandez, who were affected by only one such season, were, in fact, affected by the season that was cut short the most (1981).

The second additional statistic I’m using is Win Shares earned during “prime” seasons, thereby adding a second cumulative statistic to offset my otherwise greater emphasis on win share rate statistics. However, I also want to point out that, in the seasons that are not counted in each of these players primes, Perez played in a total of 679 games; almost twice as many as Garvey’s 355, more than three times as many as Hernandez’s 196, and almost seven times as many as Mattingly’s 98. I feel strongly that a player’s decision to continue to play beyond his prime, which accounts for 464 of the 679 games in the case of Perez, should not be a major factor in evaluating the relative merits of the careers of potential Hall of Famers. Therefore, the advantage of looking at Win Shares earned during prime seasons is that it places a greater value on a player whose prime lasts longer than normal, but does not reward a player who continues to accumulate Win Shares even though he is past his prime and, obviously, not performing at a Hall of Fame worthy level.

Next, I'll get to the actual numbers...that is, the comparison of Win Share rates and totals. Thanks for bearing with me.

No comments:

Post a Comment