O’Donnell: On coaching and the analytics movement

Analytics are at a crossroads in today’s NHL, especially within the locker room. Though teams like the Los Angeles Kings have tried to explain the statistics to their players, it hasn’t gone over well, with players such as Drew Doughty calling shot attempts percentage (Corsi) “a bunch of crap.”

It’s not important for players to believe in the statistics; it’s their job to go out there and play hockey. Where the analytics need to be used are at the management level, as general managers and coaches can use them as an advantage.

We already know that there are a number of ways for general managers and executives to use statistics to their advantage. Whether it be roster construction (don’t pay older players with “Cup experience” who are riding high percentages), or through the draft (don’t ignore skilled players because they’re Russian or a bit on the smaller side), we know that there are ways to find exploit undervalued areas in the management side of things.

When it comes to coaching, however, there isn’t as much to work with. Though the tracking of microstats is starting to open up new areas of coaching analytics, we currently can’t say that coaches can simply makes adjustments to their systems; success won’t come from carrying the puck instead of dumping it, especially if the team is chasing a specific Corsi rating (see Goodhart’s law).

This doesn’t mean that coaches can’t benefit from using an analytical mindset to look at problems differently. Let’s take the unique case of the Florida Panthers, for example. Litter Box Cats looked at the team’s difference in first/third and second period performance, and found this:

“The poor starts continued, with the team being outscored 41-59 in the opening frame, with two tilts left to go. Now, we know that goals are random events in hockey (see law #3), and so we’ll look at the team’s score-adjusted shot attempts for percentage (SAT%) at 5 on 5 in the first period, and sure enough, the team only controlled 49.3% of the shot attempts during the first period, which is a mediocre 22nd in the league.

The third period was similar to the first, as the team was outscored by opponents 61-66, and had a sc-adj SAT% of 50.2%; better than the 49.3% in the first, but still in the bottom half of the league overall, as this percentage is good for only 18th in the league.
Move onto the second period, however, and things start to get weird. Here, the Panthers actually performed incredibly well, outscoring opponents 88-79, and posting a sc-adj SAT% of 53.9%, good for 4th in the league. Overall, the change between their performance in the second period and their performance in the other two periods is the greatest in the league.

So, in the first and third periods, the Panthers are barely in the top two thirds of the league.
In the second period? They’re one of the top five teams in the league.”

Now, this wasn’t an isolated trend; the long change in the second period results in more goals, and certain teams thrive with the long change while others struggle. The Cats just happened to be one of those teams that struggled in the first and third periods while dominating in the second period.

The team’s first period struggles resulted in slow starts, and as a result, there was a focus on the team’s mental state headed into the first period; were they prepared for the drop of the puck? How could they get ready for games in a way that they wouldn’t come out flat? Why did it take a period of hockey for them to get going?

Though the mental approach may have a bit of truth to it, looking at the numbers shows that the issue with slow starts might have had more to do with the long change than it did with poor mental character.

With that in mind, what does the team do differently in the second period that results in better play?

I went back and looked at several games from the season, especially at the end of the year, when Jaromir Jagr was playing for the Panthers (their score adjusted shot attempts percentage in the second period went up by 7.6 percent from the time he joined the roster until the end of the season).

The Panthers utilized a very aggressive forecheck during the season, and forced opposing defensemen to make a quick play on the puck. In the first period, when the defensemen were close to their bench, they had no issues simply chipping the puck out of the zone and forcing the Cats to retreat back into their own zone.

When this happens, Florida now has to start a breakout from behind their net, and the opposing team has a chance to set up a neutral zone forecheck. Combine this with an occasional lack of options on the breakout, and it becomes a recipe for turnovers, both in the defensive end and in the neutral zone.

Not only does this result in rush shots against, it also prevents the team from creating any offense.

As a result, the team looks flat, and out matched/hustled as they struggle to handle their opposition’s forecheck. (For another example of a failed “behind the net” breakout, please check here.)

The second period, however, is a different story. The forecheck continued to be aggressive, but now, wary of the long change, opposing defenseman were more reluctant to chip the puck out of their own end; making line changes is harder when you have to go all the way to the opposing blue line. As a result, these defensemen would instead try to skate the puck out of trouble.

What happens here is that the Panthers have had time to set up their forecheck, and are able to force turnovers, regroup in the neutral zone, and start their attack again.

The main difference that results in the better play for Florida in the second period is the fact that they are regrouping in the neutral zone, and not behind their own net. They technically are forcing turnovers in the first period when the pressure the defensemen into chipping the puck out of the zone; their team ends up with the puck again. As Jen LC13 (@RegressedPDO on Twitter) points out in her tracking of zone entries and exits during the Chicago-Nashville playoff series,

“The main bonus to [setting up behind the net] is that it gives more time to the defensemen to read the coverage in the neutral zone. Because the first pass on the breakout is the most important, this conservative option is attractive. If the team has young or inexperienced defensemen working the breakout, this could be a major benefit… The drawbacks of this set up are that it takes a lot of time and allows the players working the neutral zone forecheck extra time to read the play as the breakout starts. It also leaves a lot of time for the forwards in the neutral zone to lose their speed and momentum if they have to wait for very long at all for the pass out of their defensive zone. The highly structured nature of this method also discourages creativity from the forwards and defensemen. Creativity may be risky, but it is also a valuable tool to cause confusion among the players trying to defend the zone entry.

The majority of Chicago’s attacks started from a motion regroup high in the defensive zone. The defensemen rarely got as low as the faceoff dots during these regroups. This allowed Chicago to quickly launch a counterattack, which gave less time for the neutral zone forecheckers to read the play and react. This type of system also cuts down on the loss of movement and speed by the forwards in the neutral zone. The danger of this system is turnovers. More than once during the series, a particularly aggressive forechecker was able to gain possession and get a scoring chance by picking the pocket of a defenseman or forward during a motion regroup.”

At the end of her post, she concludes that each team was more effective when using a motion regroup, instead of going behind the net. This is what the Cats are able to do more of in the second period that leads to their effectiveness in that time frame, and their ineffectiveness in the first and third periods.

Circling back to Jaromir Jagr, one of the reasons why the Panthers were such a good second period SAT% team with him is that the 43-year-old is a master of regrouping quickly in the neutral zone.

When the opposing defensemen can’t get the puck out of the zone cleanly, Jagr is right there to take advantage. He regroups high, makes a quick pass or two, and starts his attack back up.

So, instead of the problem with the slow starts and strong second periods being an issue of preparation, or mental fortitude, it instead can be attributed to a failure of the team to break the puck out of their zone when forced to start out behind their own net. From there, the coach can either work on getting his team to regroup quickly, or on developing a better way to break the puck out of the zone when starting behind the net.

NHL coaches are very smart hockey people, who have been around the game for a long time and have a deep understanding of how the complex pieces of a team come together to form a winning system.

Analytics are another tool for these coaches to use. Even though there’s limited information currently available, coaches can always use information, especially if it forces them to look at problems differently. Sometimes, that slight change in perception can result in a solution being found.

One thought on “O’Donnell: On coaching and the analytics movement

  1. Pingback: Latent Variables Effected Both the Bruins and Kings - TSS

Leave a Reply

Your email address will not be published. Required fields are marked *