The new approach to rate hockey players in $$$ based on advanced stats

The Benchrates model is based on the idea that not all player’s efforts are paid.

There is no consensus, no single methodology, on what stats to take into account when evaluating a player. For example, based on The Collective Bargaining Agreement, the parties may offer evidence of “the overall performance, including National Hockey League official statistics (both offensive and defensive), of the Player in the previous season or seasons” in salary arbitration hearings.

With such an approach, the main player’s stats (points, shot attempts, faceoffs, shots blocked by a player, etc) are taken into account during the negotiations. In contrast, any advanced stats (goals above expected, expected goals for and against, etc) don’t impact the price tag. However, the advanced stats are important for evaluating the player, although there is no way to mix it up with the main stats.

PAID CONTRIBUTION AND UNPAID CONTRIBUTION

As a result, the model divides the Contribution into the Paid Contribution and the Unpaid Contribution and evaluates both parts separately. The Paid Contribution is based on stats represented in NHL official statistics. The Unpaid Contribution is based on the rest of the stats.

All stats have different weights which depend on each other. The setup of the stats weights given the Paid Contribution should match the Cap. The teams most commonly handle their Cap in the following way: ~60% for forwards, ~30% for defensemen, ~10% for goalies. Thus the sum of the Paid Contribution of all forwards in the NHL should match ~60% of the Cap, the sum of the Paid Contribution of all defensemen in the NHL should match ~30% of the Cap, the sum of the Paid Contribution of all goalies in the NHL should match ~10% of the Cap at the end of the Regular Season. We used data from the 2021-2024 seasons to set the weights.

In addition, there is the Power Score — the player's Paid Contribution extrapolated to a full season (per 82 games for the NHL). It helps to evaluate the player’s pace or to compare players who have different numbers of games played. However, the Power Score should be used with caution when the number of games is small cause the sample size is small.

If the skater plays all Regular season’s games, his Paid Contribution will be equal to his Power Score. And if the Cap structure is well-balanced, both numbers will be equal to his Cap Hit. If the Cap Hit is less than the Paid Contribution, the deal will be able to be named team-friendly. If the Cap Hit is between the Paid Contribution and the Contribution, the money terms will be not perfect but will be still affordable cause the player will produce additional value.

SKILLS CONTRIBUTION

The second part of the Benchrates model is evaluating the players’ skillsets cause the Contribution as a result number does not tell us about the player’s advantages and disadvantages in detail. So, there is the other dimension of the Contribution: Individual Offense, On-Ice Offense, Defense, Faceoffs, and Penalty Contributions for skaters and High Danger, Medium Danger, Low Danger, Rebound Control, and Confidence Contributions for goalies.

Check the Contribution Matrix to figure out how the stats are distributed across Contribution groups.

The Defense Contribution is not divided into Individual and On-Ice cause the playing defense is mostly based on the correct positioning and skaters play without the puck possession.

The Faceoffs stats are evaluated separately from Offense and Defense as it is an important skill, but only centers have a large Contribution to it, and adding this number to Offense or Defense Contribution would give them an advantage over wingers and it would be incorrect.

The Penalty stats are also evaluated separately from Offense and Defense, as separate skills.

The Confidence Contribution is a curious number. It's not just goalies who impact the result of the game (a good goalie has bad numbers by playing for not-too-good teams and vice versa) but wins and losses are still important stats and the model evaluates it separately. Also, the Shutout stats were added to this group.

The Contribution is available for all situations and separately for Even Strength, Power Play, and Penalty Kill. Also, there are per 20 minutes, per 60 minutes, and per 82 games filters.

Right now we use only NHL API data for calculating NHL players’ Contributions without any microstats number.

The Paid Contributions and the Power Scores are also calculated for AHL, NL, SHL, KHL, TELH, DEL, and Liiga