clock menu more-arrow no yes

Filed under:

McGill and Beyond: Analytics in women's hockey

New, comments

What can we learn from crunching numbers in CIS women's hockey? Statistically, it's a whole other game.

If you buy something from an SB Nation link, Vox Media may earn a commission. See our ethics statement.

Richard Whittaker

A week ago, FiveThirtyEight's Allison McCann published an article deploring the lack of rich data available in elite women's sports. The argument holds especially true in hockey.

(Before you go any further, check out the McGill Martlets' crowdfunding campaign - the team is headed to Calgary next week for the CIS National Championships and is looking for $10,000 to help cover some of the expenses incurred.)

Since an image is worth a thousand words, let's take a quick peek at the most detailed game stats provided by a few different leagues:

NHL Full Play-by-Play

nhlpbp

Until player tracking becomes fully implemented, NHL's current RTSS system (essentially two older gentlemen sitting in front of a large keyboard up in the press box, manually recording everything happening in a game) remains the most comprehensive way to quantify the game.

QMJHL Play-by-play

qpbp

The QMJHL's play-by-play offers a good level of detail concerning individual production and shot location in a slicker graphical package than the NHL's real time stat page, but does not readily provide the full list of players on the ice for any given event, which makes calculating things such as Corsi/SAT%, QoC/QoT and WoWY more challenging.

RSEQ/CIS Game sheet

rseq

Things get much dicier once you move into women's hockey. The CWHL does record real-time game events, but the most detailed raw numbers a CIS team can get, without counting things internally, is a game sheet that looks strangely like the one used in my beer league. This is not ideal.

To overcome this lack of data as well as to give the team a competitive edge, I have been working with Coach Peter Smith and his staff at the McGill Martlet Hockey program to implement a real-time stats collection system. By reprogramming the team's existing video analysis software and exporting the raw data generated into a spreadsheet, we are now able to track shot attempts, zone entries/exits and player deployment as the game is playing played out. Instead of needing four to six hours to watch, re-watch and code each game manually, we are able to create a fairly comprehensive, easy to understand analytics report less than 30 minutes after the final whistle.

Without going into any more details about the data we have gathered and the on-ice improvements we have seen, I do want to briefly go over some differences between high level men's and women's hockey which are supported by the numbers.

1) Controlled zone entries

Though there is no hitting in women's hockey, it would be a big mistake to assume that the top-level game is not a physically demanding one. There is plenty of shoulder-to-shoulder contact during puck battles, and the act of "angling" the puck carrier into the boards can produce some pretty big collisions. Still, the emphasis on puck possession and entering the offensive zone with control is a lot higher in women's hockey than in men's hockey. In a typical McGill Martlets game, we carry or pass the puck across the offensive blue line over 65% of the time. At the NHL level, that number is closer to 50%, as evidenced by Eric Tulsky's past research.

2) Shot quality

Because hockey is played on the same ice surface regardless of gender, and because of the difference in shot power between men and women, it can be said that the rink plays "bigger" in the women's game. A P.K. Subban slap shot from the edge of the blueline, coming in at 90mph, could occasionally burn through the reaction times of the shot blocker and his goaltender, and almost always looks threatening to the naked eye.

2013-14 goals by shot location (Alex Ovechkin) - via Super Shot Search

ov

2013-14 goals by shot location (Wayne Simmonds)

sim

However, even the most accurate and powerful shooters in the CIS will have a difficult time in putting enough velocity on the puck to beat an elite female goaltender cleanly from 50 feet away or more. Reducing distance to net before shot release is far more critical in the CIS and the CWHL than in the NHL. If you want to score goals in a women's league, you are better off playing like Simmonds than Ovechkin.

3) Variance

While the vast majority of NHLers post Corsi/SAT possession numbers fall somewhere between 40% (John Scott) and 60% (Patrice Bergeron), the spread between the best and worst in the CIS is somewhat bigger. Some of the league's top players are routinely on the ice for eight to nine shot attempts for, per attempt against.

Occasionally, you will also see something completely crazy happen with PDO (on-ice shot% + on-ice save %). In a recent game, one of our lines was on the ice for 13 shots attempts for and five against. Unfortunately, they failed to score on any of their eight shots on net and saw their opponents miss the net twice and score three times. These three players therefore finished the night with a -3 rating and a PDO of 0 despite controlling 72% of play at 5-on-5.

If you found this article interesting, please make a donation to the McGill Martlet program - the team is headed to Calgary next week for the CIS National Championships and are looking for $10,000 to help cover some of the expenses incurred. Unlike the best male players of the same age, our girls are not on million-dollar contracts and all-expense-paid road trips, so every little bit helps! Thank you in advance!

Note: the title of this article was inspired by a terrific documentary on the hockey career of one our players. Watch it here: Leslie Oles, McGill & Beyond