June 26, 2019

WAR and Win Shares and Charlie Blackmon

March 23, 2019 by · Leave a Comment 

It appears that WAR, in attempting to assign a single number to represent a player’s season, has a problem with fielding.   And the problem is not so much with the player’s defensive WAR score – but rather with trying to COMBINE this score with the offensive WAR.   A very good or very poor fielding score appears to seriously distort the player’s total WAR score.

As an example, Matt Chapman had a great fielding season at third base in 2018 and his situation illustrates quite well why WAR does a relatively good job of evaluating his offensive value (oWAR) and his defensive value (dWAR) – but appears to stumble when trying to COMBINE the values to determine his OVERALL WAR value for the season.

  1. Baseball Reference (bWAR) has Chapman’s 2018 WAR = 8.2. This gives Chapman the #3 best season of any position player for 2018.  But Chapman’s offensive WAR is ranked only #20.  That’s right.  The claim here is that a third baseman’s defense is so good that he has the #3 best overall season even though his offense ranks as #20.   Is that believable?
  2. Fangraphs (fWAR) has Chapman’s 2018 WAR = 6.5. This result is significantly below that of B-R (26%) and raises all sorts of red flags regarding the calculation of the WAR score for both systems.   The main cause of the discrepancy appears to be the EXAGGERATED value given to Chapman’s fielding by bWAR.  On Fangraphs, Chapman is credited with the #7 best season in 2018.

If WAR was a consistent entity, there would be no way that the WAR scores could be so different for the same player.   In bWAR, the defensive portion appears to represent 43% of Chapman’s total 2018 value – while in fWAR, the defense appears to represent about 30% of the total value.  Again, quite a difference – raising questions about what is going on with WAR.

It appears to be obvious that Matt Chapman’s defensive work is being OVERVALUED when combined with his offense in both of these WAR calculations.    And remember this is a THIRD BASEMAN.   Just how much value can a third baseman’s defense have to his team?

It is this COMBINING of the offense and defense that appears to be the “problem” with calculating a single WAR score for a season for some players.

This implies a serious question regarding WAR.   We may be able to “trust” the separate scores (oWAR and dWAR) for a player to some extent – but NOT always the combined total (particularly where a player is considered to be a very good or very poor fielder).

And, of course, the bigger question is:  How much faith can be put in the  WAR results when they differ so much depending on who calculates them?  As a mathematician, I am programmed to distrust any “formulas” that produce such varied results.  WAR is simply NOT a mathematically sound metric.   But, of course, we already knew that since B-R in its explanation of WAR states: “But WAR is necessarily an approximation and will never be as precise or accurate as one would like.”  

It would appear that if one wishes to see which players really had the best season, Win Shares (WS) would seem to be the more mathematically sound (and realistic) system for comparing total season value.

For 2018, according to WS, Matt Chapman is ranked #22 offensively – quite similar to the B-R oWAR placement mentioned above.   When his defense is combined, he moves up to #18 in overall ranking.   His defense counts for 21% of his total – a seemingly more realistic assessment when compared to the 43% and 30% awarded by WAR (above).

Would not logic (and baseball intuition) dictate that there has to be a “cap” on the value of a player’s defense – depending on the position played?   This appears to be a missing piece in the WAR calculations.

Some might argue that Chapman is not being given enough credit by WS for his defense – and that may be true.   But, in any case, there has to be an assumption (a judgment) as to how much value can be awarded for the defense of a third baseman.   And that judgment is apparently at the heart of WAR’s “problem” with fielding value.  WS appears to have made the more appropriate judgment – especially when you consider how different are the two WAR calculations in some cases.

The Strange Case of Charlie Blackmon

Perhaps the most dramatic example of how questionable the bWAR system can be (because of the “fielding problem”) is that of Charlie Blackmon in 2018.

For the 2018 season, for example, Jose Altuve is ranked #22 by WS, #21 by fWAR and #20 by bWAR.   Similarly, Francisco Lindor is ranked #6 by WS and #5 by both bWAR and fWAR.

Of course, these results are not surprising since we would expect the results of the systems to be reasonably close to one another.   And they are much of the time.

Here are the top 15 ranked position players in WS for 2018:

  1. Mookie Betts
  2. Mike Trout
  3. Alex Bregman
  4. Christian Yelich
  5. D. Martinez
  6. Francisco Lindor
  7. Matt Carpenter
  8. Jose Ramirez
  9. Mitch Haniger
  10. Nolan Arenado
  11. Trevor Story
  12. Manny Machado
  13. Freddie Freeman
  14. Javier Baez
  15. Paul Goldschmidt

Twelve (12) of these players are also in the top fifteen (15) when ranked by bWAR.   And the other three are relatively close:

#11 Trevor Story is #16 in bWAR,

#15 Paul Goldschmidt is #19 in bWAR and

#7 Matt Carpenter is #21 in bWAR.

It is when fielding becomes an issue that WAR sometimes seems unable to combine the two numbers (offense and defense) realistically.   And this is perhaps best illustrated by Charlie Blackmon’s 2018 rankings.

WS awarded Blackmon 24.6 win shares for the season (a very good season) and has him ranked as the #23 best position player.

fWAR did not like Blackmon’s fielding very much and this detracted from his offensive achievements (to some extent) and his fWAR = 2.8.  He was ranked #66 for the season.   (This simply seems to illustrate that in any form of WAR good or bad fielding is somewhat exaggerated.)

But it is bWAR’s ranking of Charlie Blackmon’s 2018 season that is really mind-boggling.

According to B-R, Blackmon’s oWAR for 2018 was 3.5 (the #53 best offense for a position player).   However, his dWAR was –3.1 which resulted in a total bWAR score of 0.8 – and A RANKING OF #231.

Yes, you read that correctly.   Win shares says Charlie Blackmon was ranked #23.   fWAR says his fielding was bad and his WAR was 2.8 and he was ranked #66.   And bWAR says his fielding was a disaster and his WAR was 0.8 and ranks him as #231.   Both versions of WAR appear to be giving too much value to his fielding but bWAR really seems to have lost all perspective.    (And how can the two WAR scores differ by as much as 0.8 and 2.8?)

One last question.  How does a player go from being the #53 offensive player in 2018 according to oWAR to being ranked as the #231overall player – no matter how poor his fielding?   Does anyone really think that makes any sense?

Thank you for your time.

 

Mike Hoban, Ph.D.

Professor Emeritus (mathematics) – City U of NY

Author of DEFINING GREATNESS: A Hall of Fame Handbook (2012)

profhoban@gmail.com

Speak Your Mind

Tell us what you're thinking...
and oh, if you want a pic to show with your comment, go get a gravatar!