logo

Staturday Weekly Column #7: A sneak peak at an analytics project for the NWHL Bubble Season

alt
Ryan Hobart
1 year ago
Firstly, apologies for the last 2 weeks of this weekly column being AWOL, but I’ve been busy with something I’m really excited about.
If you’ve been following this blog lately, you’ll know that the National Women’s Hockey League (NWHL) is currently in a bubble season for 2021 to determine the winner of this season’s Isobel Cup. You’ll also know that the Toronto Six, who this blog are ecstatic to be covering, are playing their inaugural season this year for that Isobel Cup.
Getting wrapped up in all of the excitement around the bubble season, I decided to take on a project to get some analytics for this bubble season. Luckily, I had some volunteers to build out a full-fledged team of our own. The goal of the project is to track some basic counting stats as well as more in depth “microstats”.

What?

The main goal of the project is to get Corsi, or Shot Attempts, for the teams and players. Those are your basic counting stats. We’re tracking each shot attempt, and who was on the ice when it happened. This is valuable because every shot attempt that is taken has some chance, even if quite small, of going in the net if circumstances were ever so slightly different. The historical data (for the NHL and other leagues) shows that teams that have a better ratio of shot attempts (i.e. they’re taking more shot attempts than they’re allowing) are more likely to win in the future. In kind, the players who have a better shot attempt ratio are more likely to contribute on the actual score sheet in the future as well.
We’re also tracking Shot Assist data, meaning that we’re writing down who made the pass that lead to the shot attempt. Some players may be particularly good at setting up shot attempts, and thus deserve more credit than just noting that they were on the ice when the shot attempt happened.
We will tentatively also have zone attempt data from someone else on the team, but unfortunately I haven’t been as involved with that so I don’t have anything to share there yet.
There’s a lot of work involved (I personally spent 9-10 hours tracking the first two Toronto Six games) but it’s getting easier as we go. At the end of it all, though, we should have a nice little database of advanced stats for the NWHL.
We are also going to try to track zone entries and exits attempts. This means that we’ll know who is best at making those plays that help the team’s offense or defense succeed, but don’t end up on the score sheet.

Preliminary Results

So far, we have a little bit of data ready to share regarding the Toronto Six, so I thought I’d put that here for your interest. By the end, we will have a LOT more to share with you, so consider this a teaser. The table below shows 5-on-5 Corsi (or SAT) for and against, as well as the percentage ratio that each player is contributing shot attempts for vs shot attempts against. You also see the columns after showing a similar relationship for shots on goal (SOG).
I won’t extract much from this, and I don’t recommend that you do either, because the sample size is so so small at this point. However, some things that we have the beginning of a trend for are encouraging:
  • Toronto has led in Corsi for both their games, even in the Minnesota game where they lead for most of the match
  • Toronto’s top 3 have been excellent
  • Toronto’s 4th line has been a bit of an issue
  • All of the defenders are holding their own except for Steele
This is very preliminary, but with the next two Six games currently being tracked I can tell you that the trend of dominating the Corsi will continue for Toronto.

Final Thoughts

I know this isn’t all that impressive yet, but I’m personally having a lot of fun with it, despite how difficult it is, and I know that the results will be worth the effort in the end. So, follow me and the rest of the team on Twitter and you’ll get to see snap shots like this of the data. The final data release will happen later, and the means of doing so is to be determined.

Check out these posts...