skip navigation

NWHL DATA ANALYST: Expected Points-Per-Game – How Did We Do?

By Adam Gavriel, 04/11/18, 1:30PM EDT

Share


Photo By Michelle Jay

With another exciting NWHL season in the books, we can finally go back and take a look at our player projection model, and how it stacked up with the actual results from this season.

The 2017-2018 season saw some major standout performances in terms of points-per-game, a lot of which came from the offensive explosion from the Metropolitan Riveters this season. Madison Packer joined the team late in the season after coming back from an injury and took the league by storm with 1.5 points per game. Courtney Burke took home defender of the year with a staggering 1.19 points per game. And don’t forget league MVP, Alexa Gruschow with 1.375 points per game leading the league among skaters who appeared in at least 10 games.

For the Boston Pride, Haley Skarupa recorded a even 1 point-per-game in her five games with the team this year. In Buffalo, Hayley Scamurra matched Skarupa’s 1 point-per-game across 14 games played. And in Connecticut, three-year team veteran Kelly Babstock led the way at 0.643 points-per-game. The league needs to keep a major eye on Babstock next year, who despite a career low shooting percentage of 5.6% still led the team in points, and the league in shots on goal.

Overall, we have a data sample size of 44 players who appeared in the league this year and last, and played at least 5 games this season. Here is how the model shaped up:

In the chart above, the dashed line running diagonal represents an x=y line, meaning any player who fell on this line had their projection match their actual output. Players below the line were over projected by the model, while players above the line were under projected by the model.

We can also display this in raw differences, if we take a look at Actual minus Projected:

This chart does a better job really showing how much the Riveters broke the model this year.

As anticipated, and welcomed, the players really broke the model this year. This is great news for next year’s iteration. We’ll be able to add more data, and the model can ‘learn’ to better project player output. Let’s see how much better we do in 2018-2019. Get your fantasy leagues ready!