cMax Ice Hockey Notes

Welcome to

cMax: Ice Hockey Ratings Notes

The cMax ratings are intended for entertainment purposes only. In particular, they are not intended to compete with the Coaches Polls, which do a terrific job ranking teams.

The cMax methodology has been refined over the course of the past 15 years and applied to over a dozen sports leagues... from 3rd grade recreational girls soccer, to youth basketball, to a number of collegiate and professional sports, including football, basketball, lacrosse, golf, cross country, squash, rowing, and now ice hockey. The rowing results have been posted for several years at A brief description of the methodology used in the rowing model is here, where you will learn that cMax stands for "Chris Maxwell", and that I teach Sports Economics and Game Theory at Boston College... and just do this for the fun of it.

All of the data come from and In developing the model, I have used data going back to the 1999-2000 season.

The statistical methodology makes no ad hoc assumptions, and is driven entirely by game results from the current season (scores, game features (OT, penalties, etc.), venues, dates, and event types (regular season, playoffs, and championships)). The model excludes exhibition games, but does include inter-Divisional games. Because it is a statistical methodology, I am able to generate confidence intervals for the ratings forecasts and do Monte Carlo simulations to estimate the frequencies with which one team might beat another.

The cMax model is a goals/pts model. Differences in cMax ratings tell you predicted goal differentials if two teams were to play a large number of games under many different circumstances, on neutral ice. So, for instance, if Team A has a cMax rating of 9.2 and Team B has a rating of 8.8, the cMax ratings model predicts that A will beat B on average by 0.4 (9.2 - 8.8) goals. I have other models that look only at wins/losses/ties and not scores... but those models require more data for reliable forecasts, and so they (I should say one, the winMax model) will be rolled out later in the season. The best forecasts combine both types of models: goals/pts models and wins/losses models.

The Offense and Defense Ratings have a similar interpretation. Differences in the Offense Rating tell you on average, and playing lots of different teams many times on neutral ice, how many more goals one team will score than the other. So again, with A at 9.4 and B at 9.5, the model says that on average B will score 0.1 more goals than A. Note that you can use these ratings to rank the teams' offenses... as I do. These are the Off Ranks.

And the Defense Rating tells you the same for goals allowed. With A at, say, 9.0 and B at 8.5, the model says that on average B will allow 0.5 (9.0 - 8.5) more goals than A will allow... again, playing lots of different teams many times on neutral ice. And again, the ratings allow you to rank teams' defenses.

Within-Conference ratings tell you how strong each team (as well as its offense and defense) is compared to the other teams in its Conference. Note that these ratings reflect the performance of each team in all of its games, not just those against other teams in its Conference.

You'll also see "Quality Wins/Ties" (WT1-WT3) and "Notable Losses/Ties" (LT1-LT3) listed. How to read these: The WTs are the three best teams that each team has beaten or tied. So if the #1 team has WTs of 4, 7 and 7, they have beaten or tied the #4 team once and the #7 team twice. And the LTs are the three weakest teams that the team has lost to or tied. And so if a team's LTs are 25, 24 and 24, then that team has lost or tied the #25 team once and the #24 team twice. Note that CWT1-CWT3 and CLT1-CLT3 extend this concept to within-Conference games.

This is a work in progress. I will be updating ratings weekly. Feel free to send comments/suggestions: cMax (chrisMaxwell)


Other rankings and ratings:

USCHO Polls   USA Today/USA Hockey Magazine Polls  

DI Men:     Dolphin   KRACH   Lock   Massey   PWR   RPI (USCHO)   Talismanred (Wilson Model)   Talismanred (Wobus Model)


DI Women:     KRACH   Lock   Massey   PWR   RPI (Rutter)   RPI (USCHO)   Rutter

DIII Women:    KRACH   Massey   PWR   RPI (Rutter)   Rutter

Homepages:   Ken Butler   Andy Dolphin   Robin Lock   Ken Massey   Mike Rutter