• Chris Ramondelli

Can WAR (Wins Above Replacement) predict Salaries

INTRODUCTION TO WAR (Wins Above Replacement)

WAR, Wins Above Replacement is a way to assign a total value to a player’s, basically how much a player contributed to his team. More specifically it measures a player’s value according to the fraction of wins they contribute above what a ‘replacement-level’ player would contribute.

Several WAR methods have been created in the past to evaluate NHL skaters. These weren’t all called “WAR”, but each attempted to evaluate the entire value of a skater. WAR-on-ice’s write-up has a more complete history so please reference the link below, but here are a few of the more well-known examples. A note: except for Emmanuel Perry’s, none of these models are up to date:

  • The first model/system that attempted to evaluate hockey players in a way similar to WAR was Alan Ryder’s Player Contribution method from August, 2003.

  • Michael Schuckers and James Curro created a player evaluation model in 2012 (updated in 2013) called ThoR (Total Hockey Rating). While this system is not current, it appears the data is still available.

  • The now prior team at (No longer up) (Andrew C. Thomas, Sam Ventura, and Alexandra Mandrycky) created their WAR model in the fall of 2014 and hosted it on their site. The entire series explaining the model is still available online.

  • Dawson Sprigings developed a WAR model that was released in the summer of 2016 and was in production for the entire ’16-17 season. The 5-part series was hosted on Hockey-Graphs but is no longer available.

  • Emmanuel Perry created his own version of WAR in the summer of 2017 and posted an introduction to the concept of WAR here. His in-depth explainer of the model can be found here. This model is available on

  • Gordon Arsenoff presented his WAR model at the 2018 RITSAC conference. His slides can be found here. It doesn’t appear that this model is currently available publicly.

Like baseball, there is no single process in hockey by which players exert an influence on the creation of wins. Hockey differs from baseball in the complexity of events during the game. Any event on the ice surface during play can affect any play on the ice – from a faceoff win, to a check, to a broken-up pass attempt. The most structured contest between two hockey teams is a chaotic system with things happening all over the ice. Baseball is fundamentally turn-based and more easily broken down into components. This makes is simpler to measure a single players impact on a game. Assume that a position player has the following responsibilities:

1. Batting

2. Base running

3. Fielding

You could then obtain that player’s WAR by finding how many runs they contributed relative to a replacement-level player in each category, then converting runs to wins. Both remaining steps have their own challenges and they are largely shared between sports.

The challenge in calculating WAR for hockey is the overlap between many actions happening simultaneously. Consider that a hockey-WAR should include a faceoffs. That is, a player’s success at winning faceoffs, or losing faceoff draws, should count as a sub-component of their Wins Above Replacement. Now, you want to include a second component: the player’s partial impact on shot blocking/stopping. You can find that Patrice Bergeron, 60.1 FO% and -8.72 Rel CA/60, is worth 7 goals above replacement.

What went wrong with the analysis?

By doing this method, faceoffs are being counted more than once. Part of the faceoff won or lost is also part of the ensuing rate of shots allowed, lost draws have more shots allowed etc. If you insisted on including faceoff-WAR in a proposed model, all remaining factors would have to be adjusted accordingly.

The WAR numbers in this analysis come from the

Website for the 2014/2015 Season thru the 2018/2019 season. A total of 3913 data points are used.

Is WAR correlated with Salaries?

The hypothesis based off of Wins Above Replacement is that the higher a players WAR then the more they should get paid, more wins equals more player value. Using all the salary data from the previous 5 seasons, the following chart is a scatter plot that graphs WAR vs players Salaries.

From this chart we see that the correlation between WAR and Salary is positive, but not very high at .3446. Overall this does not support the hypothesis that a higher WAR will be a very good predictor of a higher salary.

Next, intervals of salary values are used to see if WAR and Salary have higher correlations in certain salary intervals.

The following intervals were used, and correlations calculated:

From this table, the correlation between WAR and Salary constantly decreases as the lower limit of salary AAV increases. The highest correlation between WAR and Salary actually occurs when the Salary is less than 1 Million AAV.

We see that WAR is ultimately not a good stand-alone predictor of salaries for NHL players over the past 5 seasons.

Using the data available from and more in depth analysis will follow to find which statistics are more highly correlated to salary data and can be used in an overall model to predict salaries.

Who had the Highest WARs over the past 5 seasons?

Over the past 5 seasons there were 21 instances where a players WAR was 4 or higher:

Only 2 of these players had a contract over 7 Million AAV during the season (Giroux and Crosby).

4 of the 21 instances included players on their entry level contracts (Stone, Pesce, Parayko, McDavid). McDavid and Stone have both gotten big paydays since their entry level deals have expired.



©2019 by The Hockey Fanatic: Analytics. Created by Christopher Ramondelli