Where Are Teams Actually Driving “Good” Possession?

Lately I have been developing a keen interest in advanced hockey stats and how they can be used in practical situations. I have developed this blog to share my opinions and findings  and develop a place for everything to be together. For the most part, discussions here will be focused on using the stats already available and trying to make sense of them. There won’t be much of the hard mathematical modelling that some analysts use because, frankly, I don’t have the capabilities to do such analysis, seeing that I am an undergraduate Sport Management student at the current time. However, I do plan on developing this aspect and possibly adding it into some posts here.

Now to the numbers. What I have been looking at lately is the score-adjusted possession numbers that are being circulated out there. Usually I’m looking at the numbers on puckon.net, and now that hockeystats.ca has incorporated it, these numbers can be viewed on a game-by-game basis.

I really like how puckon.net determines and presents their numbers, so for this discussion I am using numbers from their site. The formula they use is based off of Eric Tulsky’s method, but tweaked just a little bit. This tweaking just using the current league averages, rather than a fixed number, along with the specific team’s TOI according to the situation instead of the league’s average TOI.

It has been shown previously (i.e. in Tulsky’s analysis) that score-adjusted measures have greater predictive validity than “tied” measures, and even better than the “close” measures that have been widely used. What I have been looking at lately is the score-adjusted numbers at puckon.net, and really trying to find how they are being derived/generated.

So what I have done is taken each teams CF% for each score situation (down 2+, down 1, tied, up 1, up 2+) and compared it to the league average for that score situation. It’s one thing to look at a team’s possession numbers when they are down on the scoreboard, but it really means nothing unless you compare it to how the league performs at that score state. For example, Montreal gets 55.8% of Corsi’s when down by 2 or more. At a glance, this seems fairly good because they are getting more shot attempts that their opponent. However, it doesn’t really tell you anything because you can’t compare it to anything.

The following graph shows each teams performance when down by 2 or more and down by 1 relative to the league average at that score state:

cfdown2down1

For the most part, you can see that each team is either above average or below average in both score states. This shows what teams control possession when down compared to those who don’t. As expected, teams like Chicago, Nashville, and Detroit control large amounts of possession when trailing compared to the league averages (57.6% when down by 2+, 55.1% when down by 1).

Looking at teams who are below average when trailing, you’ll find the likes of Buffalo, Colorado, and Calgary, which you can expect. However, you can see that Pittsburgh finds themselves below average when trailing. For me, this is a good finding as the Penguins are typically seen as a “good” possession team, and are known to have the offensive firepower to perform when down. What you can see here is that this might not be the case, and Pittsburgh may not be driving the possession numbers that are expected of them.

The next two graphs here are relative CF% for each team when leading by 1 or 2+, along with Tied:

cfup2up1 cftied

When all three charts are presented, you must look at each team over each score state and compare when they are driving good possession numbers. Teams like the Blackhawks and the Lightning (top-4 teams in the league) are driving positive possession numbers relative to the league average at all score situations. What I like is my Minnesota Wild driving good possession numbers, compared to previous years, being positive relative to the league in all score states as well.

I want to talk about Toronto here for a second. In each situation, they are consistently below average relative to the league, yet are holding a playoff spot in the East. They are being outplayed and out-possessed on a consistent basis but are finding a way to get points. This is definitely not because of the possession numbers, but are a result of their PDO (which is a whole other discussion itself). With a high PDO to begin the year, many are projecting the Leafs to regress, especially with the stretch of opponents they are matching up against in upcoming weeks. This type of case is why possession numbers are not always correct in the short term, but this discussion at the end of the season would provide more of an inference regarding team possession numbers.

For one last chart, I am presenting the league-wide Score-Adjusted Corsi percentages, based on the formula used by puckon.net:

sacorsi

This just provides you with a picture of what teams are driving possession better than the league, weighted by their TOI per situation. A team that has possession numbers even with the league average at each score state would have a score-adjusted CF% of 50.0. This is a good baseline to measure a team’s performance at each state of the game throughout the season.

One thing to note with all of these charts: the situational relative corsi-for numbers get weighted in the score-adjusted formula based on how much a team spends in that specific score state. So a team like Buffalo, that has spent 360.8 minutes down by 2 or more at EV, will have their CF% in this situation weighted more than that of a team like Chicago, who has only spent 87.1 minutes of their EV ice time down by 2 or more.

Future discussions in this area should look into this factor. But for now, TOI data for each situation is available at puckon.net, which can be used to compare to the team numbers in the charts above. To look at these numbers the best, one should look at what situations teams are spending the most ice time in, look at their CF% in those situations and compare it to league averages, along with divisional and conference opponents.

Another possible future path for this discussion can look at division/conference-specific numbers and comparing teams to their divisional rivals. This can be telling because these teams play each other the most throughout the season.

Overall, these charts and numbers can tell a lot about teams and how they play based on the state of the game. If a team has an above-average CF% when trailing, but a below-average CF% when leading, with relatively similar TOI numbers for each situation, it can show that they may work harder when trailing and tend to sit back and “protect” when they get the lead. Tied score states can tell which teams dominate possession when the both teams are “equal” in terms of score effects (which is pretty much common sense).

Advertisements
Where Are Teams Actually Driving “Good” Possession?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s