clock menu more-arrow no yes mobile

Filed under:

The strain of WAR you use can dramatically affect your perception of a player

The two most popular strains of Wins Above Replacement, FanGraphs WAR and Baseball-Reference WAR, have noticeable differences in their formulas. Which players saw the largest swings in value due to those differences?

The Cardinals' Matt Carpenter was worth 1.3 WAR more using FanGraphs' evaluation.
The Cardinals' Matt Carpenter was worth 1.3 WAR more using FanGraphs' evaluation.
Jasen Vinlove-USA TODAY Sports

Wins Above Replacement can be a tricky topic in the baseball community. Some traditionalists dislike the very concept of one metric attempting to encapsulate a player's entire contribution to his team. Even within the community that appreciates WAR, there are questions about the calculation of, and value assigned to, a player's fielding ability, in addition to the various heuristics used to generate the metric.

Making matters a bit more complicated, these different methodologies can disagree on a player-by-player basis. The two most widely-used flavors of WAR come from Baseball-Reference (rWAR) and FanGraphs (fWAR) and have key differences in their calculations.

Most glaringly, FanGraphs uses FIP as the basis of its pitching evaluation, while Baseball-Reference relies on a context-neutralized version of Runs Allowed. fWAR favors Ultimate Zone Rating as its defensive backing from 2002-present, while rWAR uses Total Zone for fielding. Baseball-Reference also maintains a linear-weights-based system to measure base running, and FanGraphs uses Ultimate Base Running and Weighted Stolen Bases (wSB) for the same.

These differences can result in gaps in value ratings for players. Too often people on the internet interchangeably use the two without distinguishing the source of the metric. To illustrate this point, I've combined pitching and hitting WAR leaderboards from both sites and found the players for whom the perception of their 2015 season most depends on the specific WAR metric used in their evaluation.

Originally, these tables were merely going to be a top 10 list for each site, combining batting and pitching. Looking at histograms of the data, the spreads appear to show Baseball-Reference as being slightly more conservative than FanGraphs in its ratings.

Pitcher Spread

Batter Spread

Additionally, I found that there are differences between the marginal spreads of each kind of player. The standard deviations are .4046 WAR for batters and .6891 WAR for pitchers, with an F-Test returning a P-Value of 0.000.

As a result, the below tables have been split into four categories: batters most favored by fWARbatters most favored by rWARpitchers most favored by fWAR, and pitchers most favored by rWAR.

Hitters favored by FanGraphs

rWAR fWAR Margin
Chris Coghlan 1.9 3.3 1.4
Angel Pagan -1.9 -0.5 1.4
Matt Carpenter 3.9 5.2 1.3
Torii Hunter -0.8 0.5 1.3
Aramis Ramirez -0.5 0.8 1.3
Alex Rios -1.1 0.2 1.3
Jayson Werth -1.6 -0.3 1.3

This first group features the most restrained range of any of the four. As anyone familiar with these metrics might guess, fielding evaluations are the largest factor in these swings.

Chris Coghlan went from being considered 0.7 defensive wins below replacement (roughly 7 runs) to 4.2 defensive runs above average when moving from Baseball-Reference to FanGraphs. Angel Pagan was a slightly different case, as FanGraphs was slightly higher on his offense (81 wRC+ to 77 OPS+), and he was considered an above-average base-runner (2.4 BsR), despite his universally disliked defense.

Pitchers favored by Fangraphs

rWAR fWAR Margin
Drew Hutchison -1.7 1.5 3.2
Andrew Cashner -0.9 2.3 3.2
Chris Sale 3.3 6.2 2.9
Jeff Samardzija 0.2 2.7 2.5
Matt Garza -1.7 0.6 2.3
Jeff Locke -0.7 1.6 2.3

Andrew Cashner and Drew Hutchison each benefited over 3 WAR from fWAR over rWAR. Hutchison's 5.57 ERA was the second-worst among starters with at least 150 innings pitched. By his 4.43 FIP he was closer to average but was hit around a lot and prone to home runs.

He had control all season, avoiding walks, but Brooks Baseball reports his four-seam fastball usage jumped to 55.3 percent while flattening out (up to 9.5 inches vertical movement). In this case, he may have lacked fastball command within the zone, and FIP might not account for his lessened stuff.

Cashner struck out more batters than in 2014 (20.5 percent to 18.4 percent), but that came at the expense of other areas - for the fifth straight season, he generated fewer ground balls (down to a career low 47.4 percent), and he also walked more batters than in recent seasons (8.2 percent).

In general, these pitchers have better strikeout and walk rates than their run prevention numbers would suggest.

Hitters favored by Baseball-Reference

rWAR fWAR Margin
Ender Inciarte 5.3 3.3 2.0
Starling Marte 5.4 3.6 1.8
Kevin Kiermaier 7.3 5.5 1.8
Ian Kinsler 6.0 4.2 1.8
Melky Cabrera 1.4 -0.3 1.7

Notably included in this list is Kevin Kiermaier, who is worth an astounding 5.0 WAR on defense alone, according to Baseball-Reference. Fielding statistics are always under question as-is, but that value in particular seems due for regression. That being said, Kiermaier is still a great outfielder by any measure.

Pitchers favored by Baseball-Reference

rWAR fWAR Margin
Zack Greinke 9.3 5.9 3.4
John Lackey 5.7 3.6 2.1
Sonny Gray 5.8 3.8 2.0
Marco Estrada 3.6 1.8 1.8
George Kontos 1.6 -0.2 1.8
Alexi Ogando 0.9 -0.9 1.8

Zack Greinke's 2015 season is the poster child for the ERA / FIP divide. He had an historically low 1.66 ERA for the season and was far-and-away the most valuable pitcher by Baseball-Reference. However, his peripherals were considered very good, but only to the tune of a 2.76 FIP.

Theories about his incessant preparation improving his results abound, and in that case, FIP might not yet be able to quantify that improvement. Otherwise, the numbers available indicate his strong results were influenced by factors beyond his control. George Kontos and Alexi Ogando present two relievers who managed to accumulate such a margin in small 73.1 and 65.1 inning sample sizes, respectively.

At its widest, the margin between the value of a player's season in fWAR and rWAR can be wider than 3 WAR - the value of an entire above-average ballplayer. For context, Greinke's 3.4 WAR season margin was more valuable than Robinson Cano's season. The strain of WAR one prefers to use can have a large effect on the conclusions drawn about any given player's value.

. . .

Spencer Bingol is a Featured Writer at Beyond the Box Score. You can follow him on Twitter at @SpencerBingol.