Tackling Analytics: Part One
I’ve been playing hockey for literally as long as I can remember. Some of my earliest fuzzy memories are of taking the ice wearing a yellow bike helmet with dogs on it. I joined an organized team at the age of 5 and played in that organization for the next decade. I was always a decent player but never anything special. As a defenseman, I prided myself on my ability to make plays by skating, passing and hitting and I came to appreciate the little things in hockey that often go unnoticed — the chip by a defenseman off of the boards to get the puck by a forechecker or the outstretched stick which just catches a big enough piece of the puck to disrupt the offense’s play. These plays don’t make it onto the scoresheet like a big hit or a shot block do, but they add up to make a huge difference in the outcome of a game.
I watched a lot of defenseman Anton Stralman during his time as a New York Ranger. He was signed by the Rangers a month into the regular season in 2011 after having been invited to New Jersey’s training camp and cut from the final roster. He was known as an offensive defenseman, a slick skater and puckhandler who wasn’t necessarily big and tough enough to be a solid contributor defensively. Yet when I watched him play in New York, I couldn’t help but be impressed. He would use his body and stick to drive a winger wide, then close the gap, pin his competitor along the boards, draw the puck out from his feet, separate from the other player and make a crisp pass to move the puck back up ice. In the span of 3 seconds the Rangers went from defending against a rush to creating a counter-attack of their own. It wasn’t an overly physical play and Stralman didn’t make a sprawling last-second effort to defend against a cross-crease pass, but it was a fantastic play which Stralman would make with frequency and consistency. When combined with his offensive prowess, it was obvious that when Anton was on the ice, the Rangers were the better team.
What ultimately matters is just that: being the better team. Certain statistics and styles of play are lauded for their supposed ability to determine who the best players are, but those stats and playstyles themselves would rarely be evaluated to gauge their correlation with winning. I will never forget an article I once discovered on the Dallas Stars’ SB Nation Site, “Defending Big D”, which contained a quote from Dave Tippett, who at the time had made the playoffs in 8 out of 9 seasons as a head coach for the Stars and Coyotes. The article is available here, but this is the quote that has stuck with me:
“We had a player that was supposed to be a great, shut-down defenseman. He was supposedly the be-all, end-all of defensemen. But when you did a 10-game analysis of him, you found out he was defending all the time because he can’t move the puck.
“Then we had another guy, who supposedly couldn’t defend a lick. Well, he was defending only 20 percent of the time because he’s making good plays out of our end. He may not be the strongest defender, but he’s only doing it 20 percent of the time. So the equation works out better the other way. I ended up trading the other defenseman.”
Tippett’s analysis, in my opinion, is absolutely correct. If that first defenseman lays a hit or blocks a shot, and then is unable to effectively move the puck, the other team gets the puck back. And what do they do once they get possession? They attempt another shot. Now, not all shots are created equal. A one-timer from Sidney Crosby in the slot has a much higher chance of turning into a goal than a backhand shot from Tom Wilson at the half-wall. Even then, every type of shot has some percent chance of going in, and as the greatest player of all time said, “You miss 100% of the shots you don’t take”.
In the past several years, hockey fans and team employees alike have taken a more mathematical approach to evaluating hockey players. There is no doubt that many things in hockey are difficult to quantify but in the past couple of years incredible steps have been taken to isolate aspects of the game and get closer to true evaluation. Many people question why mathematical analysis is necessary when their eyes work perfectly well to tell them what is going on. In one of my Human and Organizational Development (referred to as HOD from this point on) classes, taught by one of my favorite professors, Andy Van Schaack, we learned about the importance of using systematic inquiry to uncover the solutions to problems. During one class session, Van Schaack showed us example after example of things that seemed to make perfect sense, only to be flipped upside down when looked at critically.
One example featured more than 50 wine experts tasting wine from two bottles: an expensive Grand Cru bottle and a cheap Vin de Table bottle. The catch? Both bottles had the same wine in them — a moderately-priced Bordeaux. While I will fully admit that this all means nothing to me, apparently the wine experts couldn’t accurately evaluate them either. When describing the wine from the Grand Cru bottle such descriptors were used as “excellent”, “good”, “complex”, “long” and “round”. When describing the Vin de Table: “unbalanced”, “short”, “flat”, “simple”, “faulty”. Because of the way that the wine looked and was presented to them, these so-called experts in the field were actually completely inaccurate in their assessment of the wines.
The same exact thing happens in hockey. Team officials, media and fans alike will be wowed by a big, bruising defenseman who lays massive hits on opponents and sacrifices his body night after night to prevent shots from reaching his goalie. At the same time, has anyone ever made an effort to discover whether or not those things actually help your team win games? In fact, I would argue that there are many instances where a defenseman blocking a shot is actually a detriment to success. Instead of the goalie, who your team pays millions of dollars to stop shots and is professionally trained in the art, having clear sight of the puck, the defender obstructs his view in the hopes that, despite his lack of equipment and training, he can do the same job the guy behind him is supposed to do.
Statistics like Corsi and Fenwick sound complicated, and some people try to act like they’re really intelligent and superior to everyone else just because they use them. At the end of the day, though, they’re simply an attempt to measure the difference in shots for and against a team or player. Recently there has been further progress made in terms of expected goals models, where each shot is weighted by the type of shot taken (such as wrist or slap), location on the ice, whether the shot was off of a rush or a rebound and other modifiers. These models evaluate the amount of goals a player would give up if they were facing average-level shooters and playing in front of average-level goaltending. Now, when I want to explain why Anton Stralman is such a great defenseman to you without showing you several hours worth of footage, I can instead take you to corsica.hockey and show you that in terms of Expected Goals Against per 60 minutes, Stralman ranked third out of all defensemen who played at least 500 minutes.
These numbers are not meant to be the single solution to every problem related to hockey. If they were, there would be no reason for me to be writing any of this — I’d just put up a spreadsheet of everyone’s Corsi or Expected Goal numbers. But using these numbers as tools can help expand our knowledge and correct hidden biases we may have.
In part two of this feature I’ll be looking at a few of the major objections or complaints I’ve heard regarding analytics in hockey. In doing so, I hope to shed a more positive light on these tools which can really help to move the game forward.