Ruminations on Advanced Stats, Boxcar Numbers, and Hockey Abstract

Posted by The Coach on August 21, 2013

Advanced stats in hockey is a misnomer. The most commonly used advanced stats, are not advanced at all. Well, they may be advanced if you are a three year old, but if you consider counting and addition advanced, you have bigger problems. Let’s start with Corsi. And yes, it’s Corsi, not CORSI (making it something of a shibbolleth for the stats-inclined hockey fan). If it were called shot attempt differential, people would have no issues with it.

That’s all it is. The more shots you get, the more likely you are to score.

Is it a be-all, end-all? Of course not. It’s really just a good way to figure out where the puck is when player X is on the ice. Someday Corsi is going to be irrelevant, as sooner or later the Stats Inc. SportVu system will be advanced enough for hockey. These cameras are up in over half the NBA’s arena’s now and track player and ball movement to the inch. When this can be applied to hockey, we won’t need a stand-in for puck possession anymore. We’ll know exactly how long a player has the puck on his stick, how often his team has the puck when he’s on the ice, and where on the ice this possession time comes. But until then, we are stuck with Corsi, so deal with it, learn about what it means, and even more import learn how to use it.

To me, that is the fundamental problem with the advanced stats “movement” in hockey. There is a disconnect between what stats are valuable and just how valuable those stats are. At this point there is not a single stat, advanced or other, that has a lot of value on it’s own. Hockey fans have used the boxcar stats forever, but the value of those have been chipped away over the last while. By this point, no one finds plus-minus useful. Goalie wins are not even discussed anymore. Goals against average has very much given way to save percentage as the standard for goaltenders. This is progress. Save percentage, goals, and assists are pretty much all that remain of the boxcar numbers. So how do you judge the players who fall outside of this? Hockey is filled with non-scorers, non-playmakers, and non-goalies. In fact, I would estimate that over half of the NHLPA consists of those players. How do you judge Derek Mackenzie versus Manny Malhotra?

It’s not the easiest question to answer. Your options are to watch every game they play. But even then, your eyes can lie to you. Mackenzie spent a lot of time last season centering Jared Boll and Colton Gillies. Of course he looked bad. When the Jackets added Blake Comeau in place of Gillies, all of a sudden Mackenzie and Boll looked more like actual NHL players. I find this incredibly interesting. The “just watch the game” crowd can be easily duped by this. However, one of the biggest problems facing advanced stats is separating the play of a single player from the impact of his linemates. Hockey is just such a fluid sport. It’s what I love about it, but it is by far the biggest hurdle people face when trying to go from being your average hockey fan to being an expert.

Speaking of experts, Rob Vollman provided a copy of of his Hockey Abstract to the Union Blue. He gets it. I’m not going to go through everything he writes about in the book, but he uses a variety of stats to discuss such questions. These include, but are not limited to, finding out who is the best player, best playmaker, best defensive player, best goalie, best coach, most undervalued player, what are the keys to winning, and what was the most lopsided trade of all time.

It’s a really good read, and I recommend it for anyone. Anyone really into the NHL stats-world is going to like it (if they haven’t already read it). If you are one of these people, I probably don’t need to sell you on this book. You already read Vollman’s work, you’ve encountered a half dozen other reviews, and should just buy a copy already. But for those of you who aren’t into the advanced stats, are on the fence about them, or just don’t quite get what they are all about, well this book holds a tremendous amount of value.

What you will learn is how to use advanced stats. You will understand the important of context. Context is key to finding value in any statistic, whether it be goals, Corsi, or anything. If you lined me up with the Sedin twins, and I started every shift in the offensive zone, I would bet I’d end up with a very high corsi, and enough goals to end up on at least the first few pages of the NHL leaderboard. Would that make me an NHL player? Of course not. You need to find the context of the other stats for them to have any value. That is probably what Vollman does best. He never uses any one statistic to tell us anything, never relies on any metric too heavily. He doesn’t just tell us what the statistics are, and how to use them, he also shows what you can do when you properly use all the statistics at hand to provide context, meaning, and value to all the stats we love (and hate).

As Vollman states in his (brief) conclusion, “Analytics work best as the starting point, not the destination.” This is a key to understand for anyone wishing to broaden their understanding, not just of advanced stats, but of hockey in general. No general fan has enough time to properly see enough of every player on every team to have a solid grasp of their value. A read of Hockey Abstract, a solid understanding of how to apply context, knowledge of whats stats show what, and weekly visits to Behind the Net really help understanding the league as a whole, and where the Blue Jackets and their players fit in the grand scheme of things.


Tags: , ,

Leave a Reply

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

3 × 3 =