Bill James’ win shares offer alternative to WAR
BY JOHN GROCHOWSKI For Sun-Times Media April 7, 2014 10:08PM
Chicago White Sox first baseman Jose Abreu (79) and Alejandro De Aza (30) celebrate the Sox's 5-3 win over the Minnesota Twins after an Opening Day baseball game Monday, March 31, 2014, in Chicago. (AP Photo/Charles Rex Arbogast)
Updated: April 7, 2014 11:16PM
Wins above replacement has become a commonly used measure for comparing players across all areas of production, largely because WAR is useful and partly because it’s easily available on the Baseball-Reference.com and FanGraphs.com websites.
But there are other measures that take into account all segments of offense along with defense. One is Bill James’ win shares, and they don’t rely on defining a replacement level. First detailed in James’ 2002 book Win Shares, they divide shares in proportion to each player’s contributions to his team’s victories, building from a base of zero.
Players who rank high in WAR almost always rank high in win shares. When they greatly differ, it’s often because of differences in calculations on defense, where advanced metrics are in their relative infancy compared to offensive metrics.
Alejandro De Aza led White Sox position players with 16 win shares in 2013, one more than Alexei Ramirez. In the Baseball-Reference.com version of WAR, De Aza’s 1.7 offensive WAR was second on the team to Ramirez’s 2.2, but De Aza had a minus-2.1 defensive WAR, and his overall WAR of minus-0.3 left him far down a list led by Ramirez at 2.5.
You’ll notice the numbers are much larger in win shares than in WAR. That’s mostly because each team is assigned three win shares for each victory. Each whole number in WAR represents a victory, but each win share represents a third of a victory.
A WAR of about 8 signifies an MVP candidate, but it takes more like 30 win shares to do the same. The Pirates’ Andrew McCutchen had a 7.9 WAR last season and 34 win shares. A middle-of-the-pack regular such as the Cubs’ Anthony Rizzo, who ranked 14th among 25 major-league regulars at first base last season with a 2.6 WAR, had 14 win shares.
In James’ book, he explains that by using three shares per victory, small differences are significant. You can be pretty certain a player with 18 win shares has performed better than one with 16. Using only one share per victory would leave players with performances at different levels virtually indistinguishable because of rounding, while using more shares would diminish the significance of each.
With a comparison to a base of marginal runs scored and allowed, players are apportioned offensive win shares based on their runs created and outs used. Runs created takes into account everything the player contributes offensely, with the formula RC =
(H + BB - CS + HBP - GIDP) x (TB + (.26 x (HBP + BB - IBB))) x (.52 x (SH + SF + SB)), all divided by (AB + BB + HBP + SH + SF).
On defense, the calculation varies for each position, accounting for different
defensive responsibilities. If you want to know more, along with calculations for pitching win shares, you’ll need to go to James’ book.
By their inclusive nature, win shares are a useful way to compare values of all kinds of players — ace middle infielders, defensively challenged outfielders, sluggers, on-base specialists — without need for a replacement-level base.