Learn Vegas Head Coach Bruce Cassidy's approach to analytics and how it helped his team win the Stanley Cup.

The best coaches are natural analysts. They pore over the video to find an edge ahead of their next game. Information is brought to the coach on line matching, face-offs, power play and penalty kill tendencies, and the coach consumes all of it, trying to retain it all so he can recall a key piece of data while behind the bench. Coaches want anything that will help them win, but it’s not always presented that way.

What we’ve commonly heard from teams and coaches is that “they’ll consider the analytics” or they’ll “look at the analytics to justify a decision.” Both of these positions start from a position of inherent bias in that the team has an idea about something and then goes looking at the data to see if it confirms or refutes that bias. Based on the less than ideal decision-making still being perpetrated in several corners of the league, this is likely what’s occurring behind the scenes.

Instead, teams should use data first to create a foundation of an unbiased viewpoint. What does this look like?

Current Vegas Golden Knights Head Coach Bruce Cassidy (former Boston Bruins bench boss) has been making waves in the analytics community for a while for his post-game press conferences. This story was written in 2019 when Cassidy was really making waves.

First, in talking about the team’s slot chances.

Cassidy first gives his impression of the game, that he doesn’t “recall a lot of slot chances” and then hedges by adding that he isn’t sure if the “analytics will back that up” and then concludes by saying “those are the most dangerous ones.” Right there we learn a lot about Cassidy’s process.

First, he is always a student of the game and learning, as all coaches do, so he provides his assessment of what’s just happened.

Next, he indicates he will check the data after the game to confirm what really happened (did they give up a lot of slot chances or not?). Finally, he provides validation for why he’s talking about this – he knows from analysis done by the team that these chances are the most dangerous so he’s right to be focused on this.

There is very little emphasis put on compete, grit, heart, etc. that you often hear coaches ramble on about, and Cassidy’s post-game presser is much more focused, detailed, and information.

Earlier in the playoffs, he commented on how “the analytics will dictate” how he should feel about his top defensive pairing after a game.

These answers reveal a bit about Cassidy’s process. He is obviously a bright coach and his teams have performed quite well with public expected goals metrics (third highest expected goal differential per 60 minutes according to EvolvingHockey). Coaches are always diagnosing things that happen in real time, making adjustments on the fly, and largely have greater domain knowledge than the average analyst, but Cassidy knows that his eyes are also lying to him on a nightly basis.

Below is a look at how the Bruins have suppressed Dangerous Shot Assists – shots following passes either across the slot or from below the goal line, which are significant predictors of expected goals at the team level. Cassidy is right in that these chances are the most dangerous and the team does an exception job at preventing them. Again, this data comes from Corey Sznajder and is from the last two seasons, Cassidy’s first two full seasons with the Bruins. The sample size averages out to be around 40 games per team.

As you can see, the Bruins have been the best in the league at suppressing shot assists from below the end line and slightly better than average at suppressing cross-slot shot assists. The data is informing Cassidy’s opinion.

It is wonderfully refreshing to hear a successful coach admit that he will check the numbers before “deciding how he feels about his top defensive pair” and compare that intelligent response to other coaches throughout the league. While reviewing video is imperative to improving an area of a team’s game, the data analysis that occurs after a game is done will provide a coach with an unbiased view of what actually happened during the game, despite what he saw or how he felt about it.

Armed with data, Cassidy and his staff and go about their jobs of improving their player’s through video work – this is a much more efficient process of breaking down the previous game and providing teachable moments ahead of the next one.

Coaches that don’t currently use this method of data -> video will find themselves guessing a lot and fall behind their colleagues, both on the ice and off of it. Every aspect of the game should be analyzed and then tested to find out where the greatest value (change in expected goal values) can be had. Then, video is brought in to illustrate positive and negative examples of the events leading to the event in question. Now, the team’s video analysis and coaching is grounded in unbiased, beneficial analysis, and is significantly more efficient and effective.

Hats off to Cassidy and his staff.