A Word About WAR

Analysis

WAR is a statistic that some of you may not be familiar with, but one that we will likely use a fair bit on this site, so we wanted to write a few words about it for those of you who may be new to the concept. Anyone looking for a more detailed explanation can find it here. Before we get started we should note that there are two slightly different ways of calculating WAR that are generally accepted in the baseball community- one used by Fangraphs, and one used by Baseball Reference. Generally speaking, Fangraphs’ WAR tends to be slightly higher than Baseball Reference’s. We will use the Baseball Reference number on this site to be consistent.

WAR stands for Wins Above Replacement. The idea is to measure how much additional value a given player provided over a ‘replacement player’. An example of a replacement player would be anyone available to a team that is not currently playing- think their best minor leaguer or a waiver wire pick up. The concept of WAR treats such players as commodities and implies that a ‘replacement-level’ team could be built and would win a certain number of games (52, actually). If you were to add a player with a WAR of 5.0 to that team they would be expected to win about 57 games. Obviously WAR isn’t perfect and things like luck play a significant role in baseball, but it’s the most all-encompassing measure of a player’s value that I’m aware of.

So, how’s it calculated? Well, it’s complicated and obviously there are different measures for pitchers than position players, but the building block is runs. On average ten runs produced or saved is worth one win (though that number has fluctuated somewhat as the average number of runs scored per game has changed over the years). Baseball is a game with a wealth of historical data and WAR takes advantage of this by calculating the likelihood that a particular scenario will lead to a run being scored. Basically, WAR simply keeps track of how much the plays each player made increased or decreased the likelihood that runs would be scored.

Position Players

Position players can impact runs in one of five ways:

Batting: Each batting outcome (walk, strikeout, other out, single, double, etc.) is assigned a value based on the number of runs that outcome produces on average.

Baserunning: Players are given credit for the increase in expected run production of a stolen base or taking an extra base on a hit and punished for being caught stealing or getting out while trying to take an extra base.

GIDP: Players expected run production is increased or decreased based on their ability to avoid grounding into a double play.

Fielding: A zone rating is used to determine a player’s expected run production in the field. There are a lot of components that go into this, but the most notable ones are range and arm strength. The field is carved into zones and players are given credit for being able to get to a ball in a specific zone or not as well as their ability to throw runners out or prevent them from taking extra bases. The velocity of the batted ball is factored in.

Positional adjustment: Since WAR is being used to compare players that play different positions it is necessary to adjust based on the position played as a catcher can have a larger impact on run expectancy than a DH (for example).

Pitchers

The impact pitchers have on runs is more obvious. The starting point used for pitchers is runs allowed per inning. This number is then adjusted to account for:

1. Level of opposition.

2. Team defence (pitchers shouldn’t be held responsible for having a weak defence behind them).

3. Park factor (as it’s easier to score runs in some parks than others).

4 Whether the pitcher is a starter or reliever.

Now that we have a general sense of how WAR is calculated the obvious question is: how accurate is it? Here’s the data for 2012:

Team

Expected Wins According to WAR

Actual Wins

Difference

Pythagorean Wins

Difference

COL

72.4

64

-8.4

69

-3.4

NYY

99.9

95

-4.9

95

-4.9

CHW

89.8

85

-4.8

88

-1.8

MIN

70.8

66

-4.8

68

-2.8

SEA

79.8

75

-4.8

77

-2.8

KCR

76.7

72

-4.7

74

-2.7

HOU

59.5

55

-4.5

59

-0.5

LAA

92.5

89

-3.5

88

-4.5

STL

91.2

88

-3.2

93

1.8

TBR

93.1

90

-3.1

95

1.9

CHC

63.7

61

-2.7

65

1.3

ARI

83

81

-2

86

3

BOS

70.9

69

-1.9

74

3.1

TOR

74.8

73

-1.8

74

-0.8

TEX

94

93

-1

91

-3

DET

88.9

88

-0.9

87

-1.9

OAK

94.4

94

-0.4

92

-2.4

MIL

81.4

83

1.6

85

3.6

SDP

74.4

76

1.6

75

0.6

MIA

67.3

69

1.7

68

0.7

NYM

71.9

74

2.1

75

3.1

LAD

83.8

86

2.2

86

2.2

CLE

65.4

68

2.6

64

-1.4

PHI

77.8

81

3.2

81

3.2

CIN

92.7

97

4.3

91

-1.7

WSN

92.8

98

5.2

96

3.2

PIT

73

79

6

78

5

ATL

86.9

94

7.1

92

5.1

SFG

86.4

94

7.6

88

1.6

BAL

85.1

93

7.9

82

-3.1

You can see that WAR gives us an expected win total that is within 5 wins for 24 of the 30 MLB teams and the highest discrepancy was 8.4 wins. However, Pythagorean win totals use total runs for and against to predict how many games a team should have won if some luck were taken out of the equation (this is obviously useful since WAR is also built based on runs). If we attempt to reduce the impact of luck by looking at teams’ Pythagorean win totals we see that WAR appears even more effective with 25 of 30 teams falling within 3.5 wins and no team having a discrepancy of more than 5.1. Not bad when you’re talking about teams that play a 162 game schedule.

As you can see WAR is not a perfect measure of player or team value, but it does a pretty good job of summarizing all of the ways in which a player can provide value for their team. As such, we feel it’s a useful statistic and will be using it regularly in our analysis of the Jays.

Find Blue Jays Way online:
@BlueJaysWay
@BJW_Kieran
@BJW_Matt
Blue Jays Way on Facebook

Leave A Comment