To get a handle on these questions, I first compared career WAR and career MVP shares for a large group of players. In the first case I used WAR from Baseball Reference. In the second case I used WAR from Fangraphs (and only each player's seven best seasons).
They were everyone who had 5000+ PAs since 1931 (I excluded anyone who played more than half a season before 1931 because that was the first year of the baseball writers MVP award).
Then I included everyone who was in the top 200 in MVP shares (only position players). The lowest career WAR of anyone in that group was 17.3, belonging to Cecil Fielder. So I then also added in everyone who had 17.3+ WAR since 1931. Total players was 810.
Then I regressed MVP shares on WAR. A second order polynomial was a better fit than a straight line regression. Click here to see the scatter plot with trend line. Here is the regression equation
MVPShares = 0.0003*WARSquared + 0.011*WAR - 0.1198
Then I estimated each player's predicted MVP shares and found the difference. Click here to see the entire results. The 20 players with the most negative differences are listed below. Willie Mays had a career WAR of 155.9. So he was predicted by the equation to have 8.89 MVP shares but he only had 5.94. So his differential is a -2.95.
Rank
|
Player
|
Career
WAR
|
Award
Shares
|
Pred
|
Diff
|
1
|
Willie Mays
|
155.9
|
5.94
|
8.89
|
-2.95
|
2
|
Rickey Henderson
|
110.6
|
2.46
|
4.77
|
-2.31
|
3
|
Lou Whitaker
|
74.8
|
0.21
|
2.38
|
-2.17
|
4
|
Wade Boggs
|
91
|
1.2
|
3.36
|
-2.16
|
5
|
Eddie Mathews
|
96.1
|
1.61
|
3.71
|
-2.10
|
6
|
Hank Aaron
|
142.3
|
5.45
|
7.52
|
-2.07
|
7
|
Willie Randolph
|
65.6
|
0.04
|
1.89
|
-1.85
|
8
|
Ozzie Smith
|
76.5
|
0.65
|
2.48
|
-1.83
|
9
|
Bobby Grich
|
71
|
0.43
|
2.17
|
-1.74
|
10
|
Buddy Bell
|
66
|
0.18
|
1.91
|
-1.73
|
11
|
Willie Davis
|
60.8
|
0.1
|
1.66
|
-1.56
|
12
|
Scott Rolen
|
69.9
|
0.57
|
2.11
|
-1.54
|
13
|
Bobby Abreu
|
60.5
|
0.17
|
1.64
|
-1.47
|
14
|
Carl Yastrzemski
|
96
|
2.23
|
3.70
|
-1.47
|
15
|
Graig Nettles
|
67.9
|
0.56
|
2.01
|
-1.45
|
16
|
Kenny Lofton
|
67.9
|
0.58
|
2.01
|
-1.43
|
17
|
Chet Lemon
|
55.2
|
0
|
1.40
|
-1.40
|
18
|
Johnny Damon
|
56.4
|
0.07
|
1.45
|
-1.38
|
19
|
Darrell Evans
|
58.5
|
0.17
|
1.55
|
-1.38
|
20
|
Cal Ripken
|
95.5
|
2.31
|
3.67
|
-1.36
|
This approach is not perfect. Some players might have long careers and so they compile a high career WAR. But if they never have any great seasons, they might not get many MVP votes. Plus, it helps to play on contenders. But Mays had plenty of great seasons and played on many contenders. There is also the possibility that if there are other great players around compiling high WAR seasons, you won't do as well in the voting.
Now here are the players who got more MVP Shares than predicted.
Rank
|
Player
|
Career
WAR
|
Award
Shares
|
Pred
|
Diff
|
791
|
Cecil Fielder
|
17.3
|
1.67
|
0.16
|
1.51
|
792
|
Mike Piazza
|
59.2
|
3.16
|
1.58
|
1.58
|
793
|
Albert Belle
|
39.8
|
2.38
|
0.79
|
1.59
|
794
|
Harmon Killebrew
|
60.4
|
3.23
|
1.64
|
1.59
|
795
|
David Ortiz
|
44
|
2.6
|
0.94
|
1.66
|
796
|
Pedro Guerrero
|
34.4
|
2.3
|
0.61
|
1.69
|
797
|
George Bell
|
20.2
|
1.92
|
0.22
|
1.70
|
798
|
Steve Garvey
|
37.5
|
2.46
|
0.71
|
1.75
|
799
|
Willie Stargell
|
57.3
|
3.3
|
1.49
|
1.81
|
800
|
Roy Campanella
|
34.2
|
2.52
|
0.61
|
1.91
|
801
|
Juan Gonzalez
|
38.5
|
2.76
|
0.75
|
2.01
|
802
|
Jim Rice
|
47.3
|
3.15
|
1.07
|
2.08
|
803
|
Hank Greenberg
|
57.6
|
3.69
|
1.51
|
2.18
|
804
|
Ryan Howard
|
18.9
|
2.49
|
0.19
|
2.30
|
805
|
Yogi Berra
|
59.3
|
3.98
|
1.59
|
2.39
|
806
|
Dave Parker
|
39.9
|
3.19
|
0.80
|
2.39
|
807
|
Frank Thomas
|
73.6
|
4.79
|
2.31
|
2.48
|
808
|
Joe DiMaggio
|
78.3
|
5.45
|
2.58
|
2.87
|
809
|
Miguel Cabrera
|
54.7
|
4.25
|
1.38
|
2.87
|
810
|
Albert Pujols
|
92.9
|
6.9
|
3.49
|
3.41
|
Pujols got 3.49 more shares than predicted, making him the most overrated player by this measure.
Then, using data from Fangraphs, I found all the players who had 4000+ PAs since 1931 and found their WAR from their seven best seasons combined (each player's WAR in 1981 was increased by 50% while it was 40% for 1994-this is due to player strikes). Total players, 931. Click here to see the scatter plot and trend line. Again, a second degree polynomial was better than a straight line regression (if you look closely, the line slopes downward for very low WAR players, which should not make sense-but this is avoided with a sixth degree polynomial whose results are essentially the same, so I used the simpler one here). Here is the equation
MVPShares = 0.0018*WAR7Squared - 0.041*WAR7 + 0.2683
Here are the most underrated players. Boggs was actually number 1 in WAR 3 straight years while his team came if first twice. He was second in WAR 3 times. He reached the post season a total of six times. But the best he ever finished in the MVP voting was fourth. Mays was 117th here. Click here to see the entire results.
Rank
|
Name
|
WAR7
|
Award
Shares
|
Pred
|
Diff
|
1
|
Wade Boggs
|
56.1
|
1.2
|
3.63
|
-2.43
|
2
|
Ron Santo
|
52.9
|
1.23
|
3.14
|
-1.91
|
3
|
Eddie Mathews
|
55.3
|
1.61
|
3.51
|
-1.90
|
4
|
Bobby Grich
|
46.15
|
0.43
|
2.21
|
-1.78
|
5
|
Rickey Henderson
|
59.55
|
2.46
|
4.21
|
-1.75
|
6
|
Chase Utley
|
46.7
|
0.73
|
2.28
|
-1.55
|
7
|
Arky Vaughan
|
50.2
|
1.23
|
2.75
|
-1.52
|
8
|
Scott Rolen
|
44.7
|
0.57
|
2.03
|
-1.46
|
9
|
Bobby Abreu
|
40.9
|
0.17
|
1.60
|
-1.43
|
10
|
Buddy Bell
|
40.65
|
0.18
|
1.58
|
-1.40
|
11
|
Brian Giles
|
40.4
|
0.2
|
1.55
|
-1.35
|
12
|
Andruw Jones
|
47.8
|
1.1
|
2.42
|
-1.32
|
13
|
Robin Ventura
|
39.9
|
0.26
|
1.50
|
-1.24
|
14
|
Chet Lemon
|
36.95
|
0
|
1.21
|
-1.21
|
15
|
Ron Cey
|
39.35
|
0.25
|
1.44
|
-1.19
|
16
|
Darrell Evans
|
38.3
|
0.17
|
1.34
|
-1.17
|
17
|
Jim Edmonds
|
45
|
0.9
|
2.07
|
-1.17
|
18
|
Kenny Lofton
|
42.24
|
0.58
|
1.75
|
-1.17
|
19
|
Carlos Beltran
|
43.8
|
0.76
|
1.93
|
-1.17
|
20
|
Graig Nettles
|
41.6
|
0.56
|
1.68
|
-1.12
|
Now the players who got more MVP shares than predicted
Rank
|
Name
|
WAR7
|
Award
Shares
|
Pred
|
Diff
|
912
|
Barry Bonds
|
76.7
|
9.3
|
7.71
|
1.59
|
913
|
Brooks Robinson
|
45.1
|
3.69
|
2.08
|
1.61
|
914
|
Eddie Murray
|
41.1
|
3.33
|
1.62
|
1.71
|
915
|
Willie Stargell
|
40.7
|
3.3
|
1.58
|
1.72
|
916
|
George Bell
|
20.8
|
1.92
|
0.19
|
1.73
|
917
|
Hank Aaron
|
56.4
|
5.45
|
3.68
|
1.77
|
918
|
Pete Rose
|
43.4
|
3.68
|
1.88
|
1.80
|
919
|
Stan Musial
|
64.7
|
6.96
|
5.15
|
1.81
|
920
|
Jim Rice
|
38.2
|
3.15
|
1.33
|
1.82
|
921
|
David Ortiz
|
31.7
|
2.6
|
0.78
|
1.82
|
922
|
Steve Garvey
|
28.3
|
2.46
|
0.55
|
1.91
|
923
|
Dave Parker
|
36.9
|
3.19
|
1.21
|
1.98
|
924
|
Frank Robinson
|
50.8
|
4.84
|
2.83
|
2.01
|
925
|
Joe DiMaggio
|
54.6
|
5.45
|
3.40
|
2.05
|
926
|
Frank Thomas
|
49.7
|
4.79
|
2.68
|
2.11
|
927
|
Juan Gonzalez
|
27.2
|
2.76
|
0.48
|
2.28
|
928
|
Miguel Cabrera
|
44.1
|
4.25
|
1.96
|
2.29
|
929
|
Ryan Howard
|
20.6
|
2.49
|
0.19
|
2.30
|
930
|
Yogi Berra
|
40
|
3.98
|
1.51
|
2.47
|
931
|
Albert Pujols
|
58.9
|
6.9
|
4.10
|
2.80
|
Did you try a log function instead of quadratic relationship? Seems WAR7 is better through R-squared alone. Is this your assessment?
ReplyDeleteYes, I tried the log function but it wasn't as good. Maybe the WAR7 is better for the reason you give but I am not sure. Thanks for dropping by and commenting
ReplyDelete