Fancystats: Determining a defenceman’s value through possession and opportunity suppression

Last week, I received an email from a reader who was concerned that the introduction to fancystats article I wrote failed to explain how to judge a defenceman’s impact within the system of analytics. To paraphrase, he said that I only explained how players can be measured as offensive weapons and how I missed out on the value of playing strong stand-up defence.

This is correct. In my introductory article, I didn’t specifically break down how each type of player could be measured. First, I wanted to build an understanding around the basic premise of the advanced stats movement, which is that the ultimate goal of the game is for a team to outscore their opponent, and that possession numbers, or shot attempts, are the most objective and efficient way to measure the extent in which a player is helping their team do just that.

What the individual who emailed me failed to understand, though, is that measuring shot attempts doesn’t just help us understand which player is creating the most offence. These numbers also shine light on which players are able to suppress their opponent’s scoring chances, giving us a metric more effective and objective than the simple eye test for determining a player’s defensive value. Before we get into it, let’s break down why the traditional view of the defensive game isn’t all that useful.

When you think of statistics that measure how good a player’s defensive game is, numbers such as hits, blocked shots and takeaways will likely come to mind.

In a post on his Hockey Prospectus in 2010, analyst Rob Vollman claimed that traditional defensive statistics are flawed, as hits and blocked shots vary from one scorekeeper’s subjective view to another. This relates to what famously regarded sabermetrician Bill James suggested in one of his early Baseball Abstracts. James complained that baseball’s traditional measurement of defensive skill through the use of errors wasn’t useful, because the idea of what should be recorded as an error comes down to who’s keeping score that day. One scorekeeper can be nice, one can be harsh, and at the end of the day, there’s no way to determine whether or not player A’s errors are the same as player B’s errors.

Vollman also claims that there is no quantifiable evidence that can support the idea that any of these traditional defensive metrics actually relate to the prevention of goals scored. Furthermore, he points out there are many other ways in which a player can contribute to the prevention of goals through ways that wouldn’t be quantified as a hit, takeaway, blocked shot, or anything that scorekeepers bother to track.

Another noticeable flaw in the aforementioned metrics is that they all reward a player who’s essentially following the game. When you record a hit or a blocked shot, there’s a very good chance that the other team has the puck, and if you’re doing it all the time, it’s reasonable to infer that you aren’t the one dictating or carrying the play. I mean, it isn’t a terrible thing that a player is hitting or blocking shots by any means, but it’s largely indicative of something you don’t want to be doing, which is following the play.

Kent Wilson, one of the pioneers of the advanced stats movement, summarized it perfectly when he Tweeted: “Blocking shots is like killing rats. Doing it is preferable to not, but if you’re doing it all the time it suggests you have bigger problems.”

So how can we actually measure a player’s defensive contribution? The next traditional place to look would be a player’s plus/minus, which simply measures their one-ice goal differential at even strength. But that’s not really fair, because a player who plays in front of a terrible goaltender is certain to be on the ice for more goals against than somebody who plays in front of a Vezina Trophy winner. It also fails to measure what that player’s individual impact was to a certain play, as he could receive a plus or a minus while not being involved in the play in any capacity.

Rather than looking at the results (the goals), we need to take a step back and look at the process that went into them. The best way to quantify a player’s defensive value is by determining the extent in which they suppress the other team’s ability to score goals. Does that sound familiar? Let’s take what I explained last week and flip it around completely.

Last week, I concluded that keeping track of a player’s shot attempts was the best way to determine their value to the true objective of the game — outscoring the opponent. Things like board battles won, strong outlet passes, and skilled maneuvers in the offensive zone are all represented by the shot attempt. Ultimately, we can infer that when a team generates a shot attempt, they were doing something right, and in that individual sequence, in one way or another, they beat the other team. So if a player owns a Corsi For percentage of 60 per cent at even strength, they’re helping the team generate significantly more opportunities than the opposing team when they’re on the ice, which is the sign of something good.

Now, flipping that around, the Corsi number can also imply that a player is skilled at reducing the other team’s ability to generate chances. For example, you may never see Martin Marincin, a lanky, awkward looking Slovakian defenceman on the Toronto Maple Leafs, throw a hit, or hell, really do anything on the ice, but when you look at his underlying numbers, you’ll notice that very few players in the NHL are allowing fewer shot attempts against per 60 minutes than he is.

This can, and should, be broken down much further. From here, we can isolate Marincin’s numbers to determine whether this elite shutdown play is actually his doing. According to HockeyAnalysis.com, nearly everybody on the Maple Leafs has better possession numbers when they play with Marincin than they do when he isn’t on the ice with them. For example, when Roman Polak and Marincin play together, which they do a lot, they allow 48.08 shot attempts against per 60 minutes. Then, when you split them apart, Marincin allows 49.57 shot attempts against per 60 minutes, while Polak allows 65.23. That’s a huge difference.

But could this be a consequence of situation? Maybe Polak plays easier minutes when paired with Marincin? Thankfully, we can look that up too. When the two of them are together, they make 34.1 per cent of their shift starts (based on face-offs) in the offensive zone. When Polak is playing without Marincin, he makes 29 per cent of his starts in the offensive zone. That number isn’t massive, but it does suggest that when Marincin and Polak are together, they’re playing in more favourable offensive zone situations that could give their possession numbers a positive boost.

Regardless, we’ve determined that when Martin Marincin is on the ice, he does a good job at suppressing the other team’s ability to generate chances. From here, we should sit down and watch him play to figure out why. From my experience watching him in his time with the Edmonton Oilers, I would deduce that he’s excellent at making breakout passes and his long reach is effective in breaking up plays in the neutral zone. At only 23 years of age, we might have the makings of an excellent shut down player who would generally be overlooked for his awkward style and his traditionally soft game.

Of course, this isn’t prefect. But it’s much, much better than using hits, blocked shots, and plus/minus to come to any kind of conclusion about anything, and it’s also much, much more efficient than watching hundreds of hours of game tape to asses all of the players you want to look at.

What I recommend is looking at a large sample size of possession numbers, then once you find the players you’re interested in, using more advanced metrics like zone starts, with and without you stats, and shooting heat charts to get a general idea of a player’s defensive value. Then, after that, you can sit down and watch the players you’ve set aside from the group, good or bad, and look and see whether the numbers actually are indicative of the value the player brings to the ice. By doing so, you might realize that the player you thought was an absolute rock actually isn’t, and that there’s a hidden gem, like Marincin, who isn’t appreciated for whatever reason. That’s what the Maple Leafs’ front office did, and they got a cheap, controllable asset for not a hell of a lot last summer. 

I’ll finish with something from Calgary Flames President of Hockey Operations Brian Burke. He said in an interview last year with Sportsnet 960 that “a computer just registers a blocked shot; it doesn’t show you that this guy dove headfirst to get it.” That is absolutely correct. In no way can fancystats tell you who has the balls to block a shot with their face. That’s something you can only figure out by getting to know the players personally and watching them play. What the numbers can do, though, is show which player is less likely to have to dive face first into a shot based on their overall play and contribution to the game. Then we can go from there.