What advanced analytics tell us about the UND hockey team
A look at expected goals reveals a lot about the Fighting Hawks.
GRAND FORKS — The UND hockey team started the season with aspirations of winning a conference title for a fourth-straight year — something that's never been done in program history — and heading to the NCAA Frozen Four in Tampa, where it lifted its eighth national title in 2016.
But the first half of the season has not gone as planned.
The Fighting Hawks have an 6-8-3 record as they head to Western Michigan this weekend for their final pre-Christmas games.
The biggest issue seems apparent: UND gives up way too many goals. But a deeper dive into advanced analytics can better put the first half of the season into context.
One of the more commonly used advanced statistics is expected goals (xG).
Here's how it works: A computer program tracks every shot attempt during a game and gives it a value — a percentage of the chance it has to go in the net. It analyzes the location of the attempted shot, as well as the type of shot — whether it's a slap shot, wrist shot, backhand or deflection. At the end of the game, it adds up the value of all the attempts and tells you what it expects the score to be.
The most frequent way to measure the tenor of a hockey game is by shots on goal. But some teams fire a lot of shots from the outside and they are not quality scoring chances. Expected goals is a metric used to analyze the quality and quantity of chances and predict the final score.
Here's the stunning part for UND — based on xG, the computer model thinks the Fighting Hawks should be 13-4 right now. It projected both games against Minnesota, one against Arizona State and one against Denver as the losses. In the other 13 games, the statistic shows UND has generated more quality chances than its opponents.
For other teams in the National Collegiate Hockey Conference, xG is much more accurate in predicting records. It predicts Denver, Minnesota Duluth and Miami's winning percentages perfectly. It's very close with Omaha (xG says 10-6, reality is 8-6-2) and Colorado College (xG says 8-8, reality is 6-9-1).
It's off on Western Michigan (xG says 13-5, reality is 10-8), St. Cloud State (xG says 8-8, reality is 13-3) and UND.
What does that mean for UND?
It means either UND is not finishing its chances at expected rates or the goaltenders are giving up more goals than expected.
Let's dive into that.
UND's expected goals this season: 59.5. In reality, it has scored 60. It predicts UND's offensive output to almost exact precision.
UND's expected goals against this season: 49.71. It has allowed 58. That means the goaltenders have allowed 8.3 more goals than expected — which is the most in the NCHC. Western Michigan is second at 5.8. Denver is third at 0.65.
The goals allowed above expectation explains why xG thinks both the Fighting Hawks and Broncos should be better. That stat is also reflected in save percentage. UND and Western Michigan have two of the five lowest in the country. St. Cloud State's No. 1-ranked team save percentage explains why the Huskies are overperforming.
But it's not only about the goaltenders for UND.
They've had odd seasons and faced strange workloads.
They're not seeing many shots. UND ranks sixth in the nation — tied with top-ranked Denver — in fewest shots on goal allowed per game (24.5). Yet despite the low shot totals, xG still predicts UND should be way down at 39th nationally in goals against (2.92).
That wide disparity between shots allowed and expected goals-against tells the story that UND may not give up many shots, but when it does, they're quite often high danger ones.
Many of UND's other analytics are quite good.
The Fighting Hawks have had more offensive zone puck possession than their opponents in 15 of 17 games. The only outliers were Friday at Minnesota and Saturday against Denver.
Even when losing to St. Cloud State last weekend, UND's possession metrics were surprisingly excellent.
On Friday, UND had more puck possession at even strength (56% to 44%), more time in the offensive zone (13:39 to 10:32) and a higher xG (4.3 to 3.6). Those are astounding numbers for a game the Fighting Hawks lost 7-2.
On Saturday, UND had even wider disparities in those categories — puck possession at even strength (59% to 41%), offensive zone possession (14:50 to 7:46) and xG (3.3 to 3).
These numbers indicate UND has a lot of underlying positives. It possesses the puck more than its opponents, it has the puck in the offensive zone more than its opponents and its generating better scoring chances than its opponents in the vast majority of games.
If it finds a way to bring down high-danger opportunities for opponents and get improved goaltending, the record should turn around, too.