Baseball fans love to argue. That much we can say with certainty. Where uncertainty begins is in the facts brought forward to support the arguments. Baseball is awash with statistics but a common mistake (the availability error) is to use the easy ones to make one's argument regardless of its relevance. Situationally, a fan can use batting average, home runs, RBI, ERA, saves or other easily available statistics to bolster his case. Alternately, more subjective criteria such as fielding ability, speed, clutch hitting and leadership are also used to contend that certain players are better. Sabermetricians have created many objective measures (OPS, VORP, etc) for player quality which, while sometimes used, have not caught the popular imagination, largely because of a lack of simplicity and comparability. Wouldn't it be nice to have a single, simple number that can accurately summarize a player's complete contribution during a season and that allows players to be easily compared? That is what Bill James the patron saint of sabermetricians has developed. It is called Win Shares. In Part I of this post we will take a non-technical look at this statistic. In Part II we will look at highs and lows over time and why we may be witnessing one of the greatest baseball players of all time.
As James describes it in his unique style Win Shares are "magnificently simple" in concept and "almost incomprehensibly complicated" in execution. The concept is top down in that it is based on the most important statistic in baseball - wins. Each team gets three times as many Win Shares as wins (it just works better that way) and then the system assigns these Win Shares to individual players based on their contribution. The primary factors for this assignment are runs created, quality innings pitched and fielding excellence.
Clarifying these primary factors is where the explanation gets complicated, so I'm not going to attempt that. For a full immersion into Jamesian logic and language you may want to check out his 700+ page opus appropriately titled "Win Shares". The first 100 pages are devoted to explaining just the formulae involved. Suffice to say that the system involves the use of almost every available baseball statistic, many long formulae, as well as the creation of some new concepts. Players are given credit for doing something extra (roughly, for being better than expected) in every aspect of their game and this is added up into win units or shares of a team's total wins in a season. Because there are adjustments for all kinds of factors including impact of the ball park, it is possible to compare players from different eras.
Measuring the player's value as his contribution to a team's wins is the most basic insight. The audaciousness comes in wanting to fold in all possible ways in which a player can help and hurt a team. The analytical ability in pulling this is off is evident in how easily the system passes the smell test. Results make sense and players generally rank the way you expect. There are very few surprises and almost no nonsensical results. Can the system be improved? Of course. Since the publication of the book (which includes Win Shares for players from 1861-2001) James and others have been tweaking it.
So what exactly do these Win Shares look like and how do you interpret them? They are whole numbers that start with zero and go up. Decimals are rounded out since they don't add any precision. In a regular season they tend to spread between zero and about 30-40 on the high end. Over the more than hundred years of baseball's existence there have been a few seasons where some players reached extraordinary heights not just because of their unparalleled skill, but also due to some unique circumstances prevailing at that time. I think averaging their performance over the most productive part of their career can give a good perspective on how good they were. Specific examples will follow in a bit.
Players separated by one or two Win Shares are essentially equal in value. Generally speaking, Win Shares below 10 indicate a mediocre season, 10-20 indicate a good season, 20-30 indicate an All-Star season and more than 30 indicate an MVP season. James provides Win Shares for all players from 1861 to 2001 in his book. The wonderful sabermetric website The HardBall Times has tweaked his formula a bit and provides Win Shares for all players from 2004 onwards. In Part II of the post we'll use both sources and see what insightful baseball nuggets we can come up with.