clock menu more-arrow no yes

Filed under:

Please stop misusing WAR

New, comments

We need to stop using projected WAR in isolation to evaluate contracts

David Rand with data from Dave Cameron and Fangraphs

The hot stove is a wonderful time to be alive as the dreadful abyss of the off-season is mitigated by flurries of rumors on players to be moved or signed. In debating the value of different players, many times people, not unreasonably, use a combination of projected WAR and the current $ value of WAR to draw a conclusion. While using data in this manner can be one data point for determining the value of a player or contract, much too often it is used in isolation to draw an absolute conclusion about a player or contract. Arguing that a player or contract is good or bad based solely on projected WAR and the $ value of WAR is a misuse of the data and poor analysis unless it is used in conjunction with other descriptive or observational data.

A good example of how this could potentially happen comes from a Dave Cameron article at Fangraphs on Ian Kennedy. (Please note this in no way is meant to single out and criticize Dave Cameron, who writes a ton of wonderful stuff for Fangraphs. This example just happens to work well in order to make my point)

In that article, Cameron posits

During 2017-2020, when Kennedy’s contract calls for him to be paid $62.5 million, he’s projected to be worth $26.6 million, a $36 million deficit relative to his contract. That’s a pretty big liability.

Cameron uses the projected WAR for Ian Kennedy along with a $ value of WAR for the next four years to come to the conclusion that Ian Kennedy will only be worth $26.6 M over the remaining years while his contract pays him $62.5 M. Thus, after awarding Kennedy $6 M of value for the remaining part of the season to bring the total projected value to $32.6 M, Cameron concludes,

that puts his overall value at about $30 million in dead money over the remainder of his contract.

Cameron then proceeds to use this number as the basis for drawing several other conclusions in his article.

To put that in prospect terms, $30 million is roughly equivalent to the value teams are paying to acquire middle-of-the-top-100 types.

...but with a $30 million anchor attached, the talent return would certainly be less than each of those three commanded.

If the Royals are willing to concede that Kennedy’s roughly a $30 million liability...

Herein lies the problem and the misuse of projected WAR and $ value of WAR; in reality, that liability as identified is not truly $30 M, but could potentially be a vastly different number when taking into account error of measurement of just WAR itself. Below you can see the table from the article that makes the calculation of Kennedy’s value.

Fangraphs Ian Kennedy Contract Estimate
Dave Cameron - Fangraphs

The main problem with using WAR in this manner is using fractional parts of WAR under the assumption that there is a statistically significant difference between a WAR of say 0.5 and 1.0. The error of measurement around a single season WAR calculation can be assumed to be about 1.0 WAR based on how the statistic is defined on Fangraph’s own website.

WAR is not meant to be a perfectly precise indicator of a player’s contribution, but rather an estimate of their value to date. Given the imperfections of some of the available data and the assumptions made to calculate other components, WAR works best as an approximation. A 6 WAR player might be worth between 5.0 and 7.0 WAR, but it is pretty safe to say they are at least an All-Star level player and potentially an MVP.

The application of a +/- 1.0 WAR to the original table would be appropriate in order to establish a range in value that the remainder of the Kennedy contract could potentially be worth. If it is assumed that Ian Kennedy compiles WAR values that are identical to the outlined projections and the potential error of measurement of $ value of WAR is completely ignored, Kennedy’s contract would still have the potential range in value of $72.5 million from a low end of -10.2 M to a high end of 62.3 M. Keep in mind, this is simply applying the error of measurement in WAR to the calculation and changes nothing else.

Ian Kennedy Contract Value with Error
David Rand with data from Dave Cameron and Fangraphs

When taking into account the error of measurement of WAR, Ian Kennedy’s projected 3.0 WAR over the next 4 years could be any number between -1.0 and +7.0. When the $ value of WAR is placed at nearly $10 M, it can be seen how the perceived value of a player or contract could have a vast range even when agreeing upon the projected value of the player and the current $ value of WAR.

Again, this is not meant to single out and/or criticize Dave Cameron, who writes a ton of great stuff for Fangraphs. The purpose is to point out that while advanced statistics give access to valuable information, it is important to understand the limitations and error involved in the calculations so they can be weighed appropriately with other observational and descriptive data to conduct analysis. I see WAR misused in this manner quite often, whether it’s to evaluate a contract or argue the difference of 0.25 or 0.5 in projected WAR. If we acknowledge the limitations of the data associated with advanced statistics, as well as the limitations of all other types of baseball data collection, it will lead to more accurate analysis.