On whether descriptive, non-predictive stats really have use

As many of you know, I love reading about baseball and soccer analytics. I think that in baseball’s case, having the “first mover” element means that later-adapting sports like hockey can look to copy many of their concepts and ideas, even if the sports are quite different. Soccer, meanwhile, is quite a similar game to hockey – just a slower version – and therefore many of the more specific practices translate quite well.

Soccer is at an interesting point though because they, unlike baseball, are uncovering new statistics (courtesy of companies like Opta) while also just now figuring out which of those statistics are meaningful. You can get an idea of how far behind analytics in soccer is by the fact that its main predictive statistic: Total Shots Ratio – which is essentially the soccer version of corsi – is actually based on corsi.

Anyway, I was just reading this article by Mike L. Goodman (seriously, what would we do without Grantland these days?) and I was drawn to one particular paragraph, which discussed a concept that I think hasn’t completely been fleshed out – and may not be for some time – when it comes to descriptive vs. predictive statistics, and their merits. Here’s the passage:

“Arguing against the box score (and counting numbers) argues against using statistics for descriptive purposes. While traditional baseball stats are not particularly predictive of what will happen, they are very descriptive of what has happened. Try describing how the Baltimore Orioles played over the last week without using statistics, or try explaining how good Clayton Kershaw is. Stats that are increasingly becoming discredited in baseball don’t fail to describe how good or bad performance has been; they fail to explain the whys of that performance and, consequently, whether or not they would continue.” (Bolding my own)

Goodman claims that statistics that are descriptive but not predictive still have value in answering questions like “how well has team X played over the past week” or “how good is X player”. But is that really the case?

Let’s take an example that hits close to home for hockey fans. PDO. It doesn’t stand for anything but I like to think of it as “Percentage-Driven Output”. It is shooting percentage + save percentage, and if that number is significantly higher than 100, you’ve probably been lucky, and vice versa. Why? Because there has been no proof that in most cases maintaining that is sustainable. Goodman, based on this article, would say that a number like goal differential, which incorporates both fairly sustainable (shot differentials) and not very sustainable (shooting success) elements, is useful in determining which teams are good or have been good, even if they don’t predict which teams will be good in the near future.

But does that really make sense? On the one hand, we want to recognize past accomplishments, whether or not they are repeatable. Think Justin Williams and his game 7 prowess. But on the other hand, if it’s understood that success (or failure) isn’t sustainable, then is it really fair to use that success in descriptions as “good” or “bad”?

Let’s look at this another way. Goodman says that conventional stats like RBI or goal-differential are “descriptive of what has happened.” On the surface, this is true. But if they don’t give  an accurate description of why things have happened, then what what good does it do? In the end, it just adds to one of the bigger roadblocks to analytics acceptance that there is: belief that we understand the past, when we don’t.

Consider these quotes from two renowned authors and thinkers in the fields of psychology and statistics.

“The core of the illusion is that we believe we understand the past, which implies that the future also should be knowable, but in fact we understand the past less than we believe we do.” – Daniel Kahneman

“The illusion that we understand the past fosters overconfidence in our ability to predict the future.” – Nassim Taleb

It may seem harmless to say, “Albert Pujols had 4 RBI last week, so he was clearly at his best,” it is in fact misrepresenting not only the future, not only the player’s true talent, but also the past.

I wrote at the beginning of this post that I thought this was a concept that both hadn’t been completely fleshed out, and that I thought might not be for some time. That is because although I think this argument is persuasive, I’m not sure it’s the only one. As I wrote in the link I posted earlier on, I think unsustainable streaks of magic have their worth, and deserve to be celebrated. It’s unclear to me, though – and I believe still is deep down in the minds of even the most adamant analysts – how these two views can be reconciled.




Controlled Situation Corsi: Patrice Bergeron, David Krejci, and the Set Play

Screen Shot 2014-09-09 at 7.01.06 PM

I wrote last week about the Bruins’ struggles following offensive zone faceoff losses. Obviously, Shawn Thornton played a part in that, but a fourth line can only get so much ice time, and obviously there was a bigger issue. When I looked over the numbers from the past few years, it became clear that David Krejci and his linemates, who received a large portion of the offensive zone starts, were hurting the team the most in those situations. Considering Krejci’s new 6 year, $43.5 million contract, and the ongoing debate over whether he or Patrice Bergeron is truly Boston’s number one center, I decided to look briefly (update: not so briefly) at the breakdown of their numbers and where such struggles might be coming from. First of all, here are the numbers.

Screen Shot 2014-09-09 at 5.45.55 PMThe chart was built to hold formulas, rather than to be clear to readers, so allow me to explain. These cells represent Bergeron’s and Krejci’s corsi for (or shot attempt differential) percentages, respectively, in the 21 seconds following defensive zone faceoff wins, the 37 seconds following defensive zone faceoff losses, the 29 seconds following neutral zone faceoff wins, the same for losses, and then the 37 seconds following offensive zone faceoff wins, and the 21 seconds following offensive zone faceoff losses. Why those specific numbers of seconds? It is after those amounts of time that Tyler Dellow found that the effects of faceoff locations and results were nullified. The final column shows open play corsi, which is essentially zone start and faceoff result-adjusted corsi, except with a slightly more limited sample size.

We could easily just examine these totals from the past four seasons, but considering we are trying to evaluate for the present and future, I wasn’t satisfied with weighting these seasons equally, so I borrowed the Marcel forecasting system, courtesy of at various points Tom Tango, Eric Tulsky, and most recently garik16, weighting the most recent seasons more highly than those further in the past. Here were my results.

Screen Shot 2014-09-09 at 5.46.27 PMAs you can see, the impact isn’t huge, but I wanted to be sure that we weren’t focussing on problems from years ago that have since been resolved. Anyway, it’s clear from these numbers – and the knowledge that Krejci doesn’t play against tougher competition or with significantly worse linemates – that Bergeron is the better possession player. The areas in which Krejci struggles compared to Bergeron are following neutral zone faceoff wins and following offensive zone faceoff losses. Considering, however, that the league average CF% following neutral zone faceoff wins is just over 60%, the takeaway from those results is clearly that Bergeron is a god among men, rather than that Krejci is horrible in those situations. Following offensive zone losses, however, is another matter. The average CF% in the 21 seconds following offensive zone faceoff losses in 2013-14 was 56.71%, so it’s clear that that is an area in which Krejci’s line needs to improve.

I dug into the video to try and find out why.

And I didn’t find much. Whether it was Bergeron or Krejci, the B’s would be aggressive and rightfully so, sending the two wingers to pursue the puck. Sometimes they would get to the puck in time, and sometimes they wouldn’t. I didn’t see anything that indicated a particular flaw in tactics or technique. Until I did. Maybe.

Screen Shot 2014-09-08 at 7.59.39 PMOn certain occasions, Milan Lucic, the LW, would line up directly behind Krejci, at the top of the circle. The hope from such a play would be that Krejci would win the faceoff straight back, and that Lucic would have an immediate chance for a shot on goal from the high slot. Unfortunately, when Krejci lost the faceoff, this would happen.

The two forecheckers would be far too late arriving. A zone exit would be practically a certainty. This play was used fairly regularly last season from what I saw. There were two conclusions I drew from examining this admittedly small sample of plays (I watched offensive-zone faceoff losses from every 10th Bruins game last season).

1. Krejci and Lucic weren’t in sync on these set plays.

Obviously, after watching just under a dozen of these, it is hard to tell whether I simply got a poor sample or if there was a legitimate problem, but from what I saw, there were a few too many faceoff wins that ended up like this.

Besides bad luck, I notice that Lucic sets up a little bit too far to the left. Considering Krejci’s right handedness, and therefore his stick positioning when he makes contact, it is inevitable that even with a fairly clean faceoff win, the puck will trend right, at best winding up in Lucic’s skates. It would take the cleanest of faceoff wins, consistently, for Lucic’s positioning there to work, and very few players can win faceoffs that cleanly that consistently.

2. Bergeron is a faceoff god and Krejci is a merely above-average.

Bergeron can certainly win clean faceoffs, but his mastery is such that even when he doesn’t do so, he can still recover the puck and put it in a place where his teammates can make something happen. Notice here how Bergeron gets the puck out of the scrum to Marchand (stationed further inside), who makes a heads up play, recognizing he doesn’t have the puck and shuffling it over to Krug, for a solid chance.

So is the play worth it?

Bergeron’s mastery of faceoffs is critical to the decision-making surrounding this play because losing the faceoff is a free zone-exit. If I were the Bruins, I’d be examining every one of these situations from every game (rather than every 10th game) to determine for both Krejci and Bergeron a) the CF% and scoring chance % in the seconds directly following a faceoff win with the play implemented, b) the same with the play not implemented, c) the CF% and scoring chance % with the play following a faceoff loss, and d) the same without the play. You get the idea. The reason is that by using Bergeron and Krejci’s faceoff percentages – 58.6 and 51.2, respectively in 2013-14 – one can determine the expected value of each offensive zone faceoff with and without the play, and thus come to a conclusion as to its effectiveness.

Inevitably, considering the nature of the play, that expected value is bound to be far higher for Bergeron than for Krejci, even if they were both equally adept at pulling it off after faceoff wins. For now, it is clear that Krejci’s numbers following faceoff losses suffer as a result of this play, and one can only guess as to whether that gap is made up following wins. My guess would be not.

On Martin Brodeur, “GOAT”, and Longevity

There was a report published today that the Montreal Canadiens have, or had, or might have had interest in Martin Brodeur as a backup goalie to Carey Price. I’m shady on the details not because I’m journalistically ignorant, but because I don’t care. Brodeur is, frankly, no longer an NHL calibre goalie, the Canadiens are set (or, I guess, already over capacity) in net, and to be quite honest none of this has anything to do with why I’m writing this piece.

I’m writing on Martin Brodeur because Joe Posnanski answered a mailbag question about Derek Jeter today that reminded me of a topic I had been meaning to write about for some time: Longevity.

There are many who claim that Brodeur is the greatest goaltender of all time because of his 688 wins or 124 shutouts, and then there are those who retort that wins are a highly flawed, team dependent stat, and that save percentage is far more important than shutouts. These people point to guys like Patrick Roy or Dominik Hasek as those who put up sparkling save percentages during shorter careers, and end up right at the top of era-adjusted save percentage lists, rather than 10 or 20 rows down.

Screen Shot 2014-09-09 at 11.28.36 AM

For the longest time, I was completely pro-Roy and anti-Brodeur when it came to the GOAT debate – which was logical having grown up in Montreal, and also having cheered feverishly for the 2001 Colorado Avalanche as the bandwagon of my youth. But my attitude on the matter has started to shift. I still believe that at his best, Dominik Hasek was the greatest goalie this league has ever seen. This graph, courtesy of Chris Boyle, shows Hasek’s save percentage at each age, relative to league average. The numbers speak for themselves. Hasek only made his NHL debut at age 26, but he was outstanding right up until his age 38 season, when most goalies are already golfing six days a week.

Screen Shot 2014-09-09 at 11.53.06 AMI also still believe that Patrick Roy, all things considered, was the greatest of all time. To me, he had the right mix of peak performance and consistency, individual and team accomplishments, to fit the bill. But I’ve come around on Brodeur; I’ve elevated his status in my mind from “overrated reputation-riding team free-loader” to “overrated, but still pretty remarkable.”

After all, this is a guy who (has) played at least 30 games in 19 different seasons, and 29 games in a 20th. His save percentage was above .900 for all of them, and while in this era .900 isn’t exactly the benchmark of the elite, considering that nobody else has ever done that, it’s something.

Screen Shot 2014-09-09 at 11.34.17 AMSo what’s the bottom line? Just like how Derek Jeter will ultimately be remembered for being a very good baseball player for a longer period of time than just about anybody else, Brodeur will be able to say that he was an average or better goalie for longer than anybody else (except maybe Roy, who quit while he was ahead), and who at times was very very good.

Staying healthy, keeping a starting job, winning trophies, winning Stanley Cups, none of those things are easy. Marty managed to accomplish them all over a long period of time, longer than the Dominator. And that’s worth something. So no, Martin Brodeur wasn’t the greatest goalie of all time, and he certainly wouldn’t be near the top of my list of starters for a one-game playoff, but he survived for a long time in a league where so many don’t, and his reward was the most games played, wins, and shutouts in the game’s history. And that’s nothing to scoff at. 

On P.K. Subban and the NHL’s Marketing Gaffe

I was chatting with a friend yesterday and the discussion turned to hockey. Now this was somebody who has worked in a professional sports front office in analytics, has met with a number of commentators, coaches, managers, and players in the sports world. It’s safe to say ESPN is glued onto his TV screen for major live sports. The guy knows sports, and he happens to be of a minority group. “The Calgary Flames,” he told me, “were always a team I liked because of Jarome Iginla. He’s great, and he’s black!” “Oh,” somebody else chimed in, “well what about P.K. Subban of the Montreal Canadiens?” “Wait, there’s another black star in the NHL now?” was the response.

I was shocked and yet somehow not surprised.

This was a player who was not only potentially the best defenseman in the National Hockey League, not only African-American (or Canadian, if you’d prefer), but also possibly the most marketable player in league history. Subban has style, he has spunk, he’s modest (at least on camera) but still bold and confident. He’s generous and seems wise beyond his years, but still has a youthful charm and unpredictability that draws cameras – and certainly fans who have heard of him – to his Ice Bucket Challenges and charity signings. 

Yet here was somebody who should be learning to love hockey through the star who most closely resembled himself. And he had never even heard of him.

The way I see it, the league has two excuses for not shoving its stars down our – or rather Americans’ – throats on a more frequent bases.

1. It’s a team sport, so the teams should be emphasized more than the players. My response? Just because Sidney Crosby can’t win a Stanley Cup on his own – NO, HE CAN’T, CERTAIN MSM – doesn’t mean he can’t be promoted and idolized. He’s not LeBron James, but to Americans he doesn’t have to be. He just needs to be Sidney Crosby, rather than…you know…”who?” 

2. Hockey’s stars, unlike those of other sports, have dull personalities and spout cliches and are therefore harder to market. My response? This is true. Crosby and Steven Stamkos are fairly dull. Patrick Kane looks like a choir boy. Drew Doughty should be a member of 30 Seconds to Mars. Alex Ovechkin speaks another language. It’s not easy, and I’m sympathetic to that. But Subban is like a gift from the hockey gods. It doesn’t matter whether he does or doesn’t have character issues away from the cameras. In public, he is the most marketable hockey player in the world (if he were American, it would be of all time), and it’s time that the NHL took advantage of that.

So how to do that, well I have a couple of ideas?

1. Rig the NHL cover vote to get him on the cover. Believe it or not, that game isn’t really catered to hard core hockey fans who actually want realistic action. Casual fans love it, and keep in touch with the game by playing it. But Patrice Bergeron? Meh. He’s one of the most underrated players of this generation, but he’s quiet, French-Canadian, white, and unimposing. He simply doesn’t do much for the NHL brand. Subban would.

2. Commercials featuring Subban on major networks. Let’s face it, commercials for hockey on NBC Sportsnet don’t do much, because the people who watch NBC Sportsnet are already watching hockey. But imagine a well-made commerical featuring P.K. Subban – his camera antics, his post-season heroics, the chip on his shoulder – on ESPN, or if that wasn’t possible, FOX, or CBS, or ABC. The point is to acquire new hockey fans, not to pander to current ones. “Hey hockey fans, don’t you dare forget about those playoffs coming up, history will be made and all!” is great, but mostly meaningless. Showing the greatness of hockey to outsiders, particularly on sports networks, pulls in fans.

I wrote about one race problem the league may still have yesterday. This might be another minor one. It’s unacceptable that the league wouldn’t use P.K. Subban (and guys like Evander Kane, and maybe even more importantly, the American Seth Jones) to try and attract a young generation of African American sports fans. It’s becoming cool to like hockey again, after all, so now is the perfect time to strike. Don’t waste god’s recent gift to hockey; embrace it.

On Akim Aliu, Racism, and Character in Hockey

I’ve been reading Gare Joyce’s “Future Greats and Heartbreaks”, a book about a journalist learning the art of scouting, and there have been a number of neat tidbits about the 2006 and 2007 NHL drafts and the junior hockey that was being played at that time. But there was one story that I didn’t find as much neat as I did sad. There’s no secret that there’s a race problem in hockey. It’s no longer hard racism so much as it is soft. White Canadian coaches prefer compliant, dull, white Canadian players, and so when a guy like P.K. Subban –  with a lively personality and something of an ego – comes along, it sticks out like a sore thumb, and words like “enigmatic”, “selfish”, “undisciplined”, and “cocky” get thrown around. 

As Joyce travels around the Canadian Hockey League venues, he encounters a draft-eligible prospect by the name of Akim Aliu. As a hockey fan, I remember Aliu as somebody who was drafted in the second round by the Chicago Blackhawks, who briefly played for the Calgary, and just recently played a season with the Hamilton Bulldogs. But it turns out there was a lot more there I had missed. 

In his draft season, Aliu found trouble when he got in an altercation with fellow Windsor Spitfire Steve Downie. Reportedly, when Aliu failed to participate in a team hazing exercise, Downie cross-checked him in the mouth in practice. Aliu went to the dressing room but soon returned and began trading fists with the future NHL pest.

“Nobody wanted to hear my side of it,” he says. “When they told me I had to go in [the hotbox on the bus] I wouldn’t do it. I wouldn’t do that to anybody else, they shouldn’t do it to me – that’s just how I see it. It’s not part of the game or the team. We’re here to play hockey…

At the practice Downie cross-checked me in the mouth. I went to the dressing room and saw that three of my teeth were chipped and broken. That’s when I lost it. I came back out on the ice and I went after him.”

Downie, despite an even more alarming history of violence, played on Canada’s world junior team that year. Aliu, who was similarly ranked according to Central Scouting (although a year later) never even got an invite to an U-18 camp. Hockey Canada officials at the time told Joyce that there was “no interest in Aliu whatsoever.”

Here’s more from Joyce:

“Alieu’s bearing isn’t what I expected. He’s soft-spoken. He doesn’t make eye contact easily. He’s a little socially awkward. When we go to a restaurant downtown, he doesn’t want to order anything. When he finally does, he sees a garden salad on the menu and asks the waitress if “that’s the one with tomatoes…”

Aliu seems like a complicated case on the ice and from what’s common knowledge about his personal history – the hazing incident and fight with Downie in Windsor, his suspensions, his performance at the Top Prospects Game. his backstory, though, is stuff you couldn’t write as fiction.

Aliu’s father Tai is Nigerian. I assumed as much when I saw that Akim was born in Okene, Nigeria. What I didn’t know was that Akim’s mother is Ukrainian and that he grew up in Kiev….’My family didn’t come to Canada until I was twelve,’ he says, ‘I didn’t speak any English. We spoke Russian at home.’

I ask Akim if he cosniders himself Canadian, Russian, Ukrainian, Nigerian, or some mix. He leans to the Ukrainian-Russian side. He says he has some good memories of Kiev. ‘I like the Russian way of doing things in school,’ he says. ‘There are lots of things they do better.’

He sounds disappointed about his family leaving Kiev – at least until he talks about his father. ‘The police would harass him,’ Akim says. ‘They’d pick him up on the street and take him in just because he was black. That’s why my parents wanted to leave. They didn’t want that for my older brother and me.'” (bolding my own)

Aliu also had problems adapting to North American society. He and his family lived in a single bedroom on welfare; Akim wanted to play hockey but his family couldn’t afford equipment so he had to borrow. 

“When we first came here from Kiev I had a fight every day. I didn’t understand what people were saying so I thought they were saying something about me.”

Aliu, however, from the time he started playing hockey at the age of 12, was a clear phenom. His Toronto midget team with John Tavares and Sam Gagner lost two games all year, and in many of the games Aliu was the most impactful player. And going into his draft year it was clear he understood the challenges he faced as a late-comer to the game.

“In junior, everyone is learning. I got a late start. I have to learn more [and] faster than everybody else.”

So what does this all mean? After all, Aliu basically flamed out despite a cup of tea with the Flames. Well scouts always talk about character when it comes to players, and coaches about guys that fit in the room and are coachable. My attitude had always been that if a guy is talented enough, coaches should simply learn to put up with potential character issues for the sake of having a good team. But this Aliu story has changed my perspective to an extent. Maybe the responsibility on coaches should be even greater, especially at the junior level. For a guy who doesn’t know hockey culture, who doesn’t even know North American culture, maybe the responsibility should have been placed on Aliu’s coaches and mentors to teach him how to be a good player, and a good person. Not to say they didn’t try, but for Hockey Canada to flat out turn down a potential first-round pick without so much as a sniff? That makes me question whether they truly have their players’ best interests at heart, and of course how deep their discrimination really lies. 

Of course, in 2014, the name Akim Aliu could easily be replaced as Josh Ho-Sang, a guy who was a prodigy playing with Connor McDavid in midget and has since been portrayed as lacking character and humility. Hockey Canada has ignored the talented youngster in every way possible, and rather than an indication of confidence, any statement about his own abilities has been taken as a further black mark. 

Maybe it’s not incumbent upon young black players to refine their characters to suit hockey’s norms. Maybe, like in so many other ways, it’s time hockey change its attitude, and its developmental system, to include the most talent possible.

On Neilson Numbers and Alternatives to Corsi

Roger Neilson

I received an email the other day with a proposal for player tracking, which for the sake of privacy, I won’t go into here. But the proposed metric was similar in some respects to the somewhat infamous Neilson Numbers (which you can read up on here). I thought I’d publish some of my thoughts on these types of metrics here, since at first glance they seem much more useful than something like corsi.

But first, what are Neilson Numbers. They are a version of +/- that attributes a “plus” to a player who is seen to directly enable a scoring chance for, and a “minus” to a player who makes a mistake causing a scoring chance against. They were created by a legendary coach named Roger Neilson in the 1970s, and have been adapted and used by a number of front offices and coaching staffs around the league; it is believed they remain in use today.

In theory, Neilson numbers are great. They are obviously better than conventional +/- both because they don’t award marks to players who have nothing to do with a play and because they are based on scoring chances rather than goals, which is good because shooting success is largely – but obviously not totally – luck-based.

So what’s wrong with them? Well first of all, anything to do with scoring chances is incredibly subjective. The most objective way to track them is to count any shots from the home plate area that spans from the goal posts to the faceoff dots to the top of the circles, but that really just tracks “close in shots” rather than scoring chances.

Neilson used the home play area, but also incorporated screened and tipped point shots, which add a subjective layer. Even if one person could avoid bias and maintain an objective view of such a chance constantly, once anybody tried to assemble a team to track these numbers league-wide, one would encounter serious scorer issues.

The second major issue with these numbers is that they don’t take into account the probabilistic nature of hockey, not just when it comes to shooting, but also driving play up the ice. Basically, Neilson Numbers attribute scoring chance causality to the player or players who directly cause the chance, whether it be with a seeing-eye pass, or a deke, or being in the right place at the right time. But the factors that cause chances go far beyond that. Players that are able to make defensive stops, to transition to offense quickly, to generally lead his team to good results without necessarily making the flashy play, are going to be severely undervalued using these metrics.

Corsi (or on-ice shot attempt differentials, if anybody is new to fancystats) certainly has its flaws. Often players will get a “plus” or “minus” for a shot attempt they had nothing to do with, but the advantage is that because the sample is so large, this variance after long enough will get filtered out. Sure, if you flip a coin 10 times, you might get 8 heads. But if you flip it 1000 times, you probably won’t get 800 heads.

As I wrote in the email:

“While something like corsi has its flaws, the main advantages to it are that it is mostly objective in terms of tracking, and that it focuses on results over a large sample. In essence, it says ‘I don’t really care how you make your team score more goals when you’re on the ice than the other team, I just care THAT you do.’ Obviously, this is once you take contextual factors into account.”

That last point is really important. Process is obviously important, but in the end if one can prove with statistical significance that player makes his team better, then the question of why is of second-hand importance. Guys like Anton Stralman would probably register pretty awful Neilson numbers, but their ability to push play in the right direction is critical to team success, and isn’t recognized by direct-contribution metrics like that of Roger Neilson.


Controlled Situation Corsi: Shawn Thornton and the Boston Bruins


More arguments over the role of character in team sports have been waged than Simpsons episodes watched this past week. Well, maybe not. But every day it seems, a main stream media member churns out a piece on the value that a fourth liner brings to an organization, one supposedly well beyond their pay grade and ice time. Inevitably, bloggers and analysts shoot back about the lack of correlation between toughness and victory, or the impossibility of measuring something like leadership in the grand scheme of the game.

The truth of the matter is that neither side of the argument, in its purest form, is wrong. I’ve played enough team sports in my life to know that the presence of guys with positive, influential, and commanding voices can lead to better teamwork, to pushing that little bit harder, and to look on the bright side when times are tough. On the flip side, that difference is nearly impossible to measure, and therefore sacrificing talent for character — from a management perspective — or painting a role player as a star contributor — from a media point of view — isn’t justifiable.

I can understand the instinct to support good people because I’ve experienced it. Having the good fortune to work for the Capitals the past two seasons, I’ve interacted with a number of players, and some stand out as the real nice guys of the sport (at some point I’ll write about my awkward confrontation with the awesome Adam Burish).

But maybe the nicest of the bunch was a 6’2”, 217 lb enforcer named Shawn Thornton. As somebody who prides myself on objective, thorough analysis no matter the team, I’ve had to resist blurring the line between supporting a good man and having that skew my writing. A Thornton or Burish being a great individual does not make them inherently good hockey players. As Houston Astros GM Jeff Luhnow once said, “there is no correlation between being a nice guy or a good person and being a good baseball player.” The same applies to hockey.

Thornton has been a catalyst for debate over the past season, both because he is one of the last of the dying enforcer breed, and because he is generally seen to be an “important fourth liner” playing on the “best fourth line in the league”.

Joe Haggerty recently wrote this article about the outgoing Bruin Thornton, now of the Florida Panthers, and it sounded like a media member who will sorely miss a departing friend who clearly had an impact on the mental state of his teammates. There’s nothing wrong with that, either morally or factually, but in order to examine whether Thornton was also important to the team’s success on the ice, one needs to examine the question with a higher level of scrutiny.

Screen Shot 2014-08-25 at 3.20.11 PM

A few months ago, before being poached away by the Oilers, blogger Tyler Dellow delved into the world of controlled situation corsi percentage. He found that after a certain number of seconds, the effect of a faceoff’s zone location and result is neutralized, leading to what he termed “open play corsi”, or rather corsi devoid of faceoff bias. This summer, figuring that Dellow might not be in the public eye much longer, I sought to replicate his work with the idea of expanding on it.

I haven’t tested for the repeatability of his findings, or their correlation to winning. That is work that I hope to do in the near future, but for now, this is a taste.

Screen Shot 2014-08-25 at 3.20.28 PM

In analyzing Thornton, I decided to examine his performance following offensive zone faceoff losses for a few reasons. First off, enforcers are often given these minutes because coaches feel they need to be “sheltered”. The upside to sheltering fourth liners is that it’s less likely the players immediately get trapped in their own zone and/or scored upon. The opportunity cost, however, is that the team’s stars can’t take on these assignments and begin their shifts with direct opportunities to score. Second, despite the Bruins being a top possession team for the past number of years, they have traditionally ranked fairly low following offensive zone faceoff losses. This, of course, is only the fourth line’s fault to a small extent, but the gap was notable enough to investigate.

The final reason I chose offensive zone faceoff losses (or more strictly, the shot attempt differential in the subsequent 21 seconds) is because they require wingers like Thornton to exert forecheck pressure. Centers and defensemen generally are less important at keeping the puck in the zone than their wing counterparts in these situations. Here are Thornton’s numbers from the past two full seasons (considering the small sample sizes I felt using 2012-13 would be even less reliable) compared to some of the other prominent fourth liners from top teams. I’m not so sure “best fourth line in the league” was really an accurate claim.

Screen Shot 2014-08-25 at 3.19.54 PM

Dellow’s research with regards to the Oilers uncovered tactical decision-making that led to suboptimal outcomes. I watched Thornton’s first 10 and last 10 on-ice offensive zone faceoff losses from the 2013-14 season to see whether I felt that his poor results were tactical or technical in nature. In the end, the answer seemed pretty clear cut. At least in this small sample, it seemed as though Thornton’s lack of speed and inability to win board battles despite his size was a hindrance to his line in offensive zone situations. Take a look for yourself, I paused the video at moments where I felt that Thornton was at least partially responsible for the puck exiting the opposition’s zone.

Acquiring character players can give a team a psychological edge, but assigning causality based on that with the type of primitive analytics we have at our disposal means risking massive mistakes. Thornton is a great guy, and by all accounts a fantastic teammate and leader, but if a restructured fourth line means improved play in sheltered minutes, than Thornton might not be missed on the ice as much as he will be off it.

On the Green Bay Packers and public sports ownership

Packers ownershipJack Han, formerly of the Montreal Canadiens, now of Habs EOTP, wrote an interesting article yesterday on whether the Canadiens should allow their fans to vote on the team’s captaincy, or, on a larger scale, whether sports are trending towards a non-profit model.

It was only this summer that I found out that such a model was present anywhere, let alone  in the NFL – for a Bears fan, the idea that the Packers are innovative frustrates me to no end – but I spent several hours researching the idea, and was fascinated by what I found. For those that don’t know, allow me to summarize.

The Green Bay Packers are the only publicly owned North American professional sports franchise. That means that, similar to mass companies like Coca Cola, Apple, or Google, the public can buy stock in the team, and therefore literally own a piece of the club they love. Of course there are serious differences. Unlike with the companies I referenced, there are a quite limited number of stocks released only about five times per century, and the company pays out no dividends of any kind. That means that people can’t profit from these stocks. Their value is completely symbolic, except for the fact that fans can attend a once annual “owners’ meeting” at Lambeau Field, where they get to vote on organizational issues – although those are more in the range of “whether to invest in upgrades to the stadium” rather than “how much money to give quarterback Aaron Rodgers on an extension”.

The Packers are a non-profit; they invest any money they make back into the product. When they need more money, such as when stadium renovations are required, they sell more stocks. It’s basically a fool-proof model, after all, there will always be sports fans willing to invest in the team they support. And the system has serious advantages. It is written in the team’s bylaws that if it were ever sold, all of the profit from the sale would be donated to charity (read: it will never be sold). Ticket prices are below average despite constant sellouts. The team won’t move because it won’t be sold and because it makes money (the team recorded record profits last year). It truly is a dream scenario.

So why hasn’t it caught on? Simple. Public ownership is outlawed in just about every professional sports league, the NFL included. How do the Packers get away with it? They outdate the NFL. The Packers were grandfathered into the league as the only team that could be publicly owned. And it’s not an exaggeration to say that this system is the only reason they still exist as a franchise at all.

Consider this, Green Bay is a city – town? – of 300,000 people. That’s good for 262nd biggest in the United States. The next smallest with an NFL team is Buffalo, of 1.2 million people (good for 70th biggest). According to Forbes, the Packers are the 12th most valuable NFL franchise, and a number of studies have found them to be the average American’s second favorite team. With the gleam of warmer climates and massive populations beckoning, there’s no way the Packers would have survived under private ownership, let alone become one of NFL’s healthiest and most recognizable brands. Team, fans, and city all push in the same direction, because they’re literally the same people. And god knows we’ve been told enough times what teamwork can do in the sports industry.

So what does this all mean for the future of ownership? Well as I mentioned, it’s strictly outlawed for any franchise in North America to replicate the Packers’ approach – it’s any other private owner’s worst nightmare – but CBAs don’t last forever. What the Packers have accomplished – and soccer clubs across Europe have replicated – works, and just as we discussed with analytics earlier, things that work tend to catch on eventually. A league of publically-owned sports franchises would be a healthy and stable league, without much of the greed and arrogance that makes up professional sports today. It would certainly be a refreshing experiment.

So should the Canadiens allow their fans to vote for the next captain? Probably not. When you have private ownership, every marketing stunt tends to stink of a marketing stunt. The fans would select P.K. Subban, the Canadiens might not agree, and the narrative of Subban and the fans vs. management would just be further developed. It’s a nice thought, but like public ownership, it’s probably nothing but wishful thinking for the near future. As for the long term, we can always dream.

On the Chicago Cubs, Draft Inefficiencies, and the Last Mover’s Advantage

061312_epsteinThere’s a scene in Moneyball where Red Sox owner John Henry offers his vacant GM position to Billy Beane. “Anybody who’s not tearing their team down right now and rebuilding it using your model,” he says. “They’re dinosaurs.” The Oakland A’s were the first movers of modern sports analytics. They took a risk, and while there were stumbles along the way, they benefitted as a result. In hockey, it took a decade longer for any kind of true analytic implementation, and we’re still not quite in “tear down and rebuild using your model” territory. So why has it taken so long? I think the answer lies in the hockey world’s view of baseball. NHL executives are drawn to the differences between the two sports rather than their similarities. Yes, baseball is a stop-start game whereas hockey is fluid. And yes, baseball involves more one-on-one matchups and less team play. But beyond that, the games – and the strategies that result in building the best possible teams – are actually quite similar.

Why does this matter? Well I think that until recently, NHL teams have been afraid to commit to analytics as an organizational approach – and to admit to conducting analytical work to the media at all – through fear of being the first mover. Executives were afraid that if they were to, say, hire a blogger to their front office, or evaluate players largely based on statistical trends and evaluations, they’d be out of a job if results didn’t go their way. But while such an executive might be a first mover in hockey, he wouldn’t be in the grand scheme of things. Analytical approaches work, and baseball has proven that. So what the first NHL GM to truly buy into analytics had was the advantage of being a last mover in the sports industry – avoiding early mistakes and being able to build on other models – while avoiding the disadvantages that generally go along with arriving late to the party because hockey teams specifically (the main competitors) were all just as far behind. The first GM to seriously adopt analytics in hockey was poised to have the best of both worlds, but because most saw baseball as just a stop-and-start game of catch, it took years for anybody to take the plunge.

I make this comparison between hockey and baseball because baseball writer and sabermetrician Rany Jazayerli wrote a piece at Grantland a couple of days ago about the Chicago Cubs’ rebuilding efforts. The Cubs, now run by former Red Sox GM Theo Epstein – aka the guy who got the job Beane turned down – are using analytics to exploit a rather basic market inefficiency. They found that drafting and developing pitchers was a far riskier proposition than doing so with hitters, and delivered less value. So they just stopped doing it.

Not completely, of course. They didn’t go full Sham Sharron and hit the beach while letting algorithms decide their picks, but with the luxury of four consecutive top-10 selections, and ignoring the conventional baseball wisdom that hard-throwing young pitchers are critical to success and therefore the ideal top draft picks, they took hitter after hitter after hitter after hitter. And while a number of the pitching prospects they were supposed to take have faded whether as the result of injury or lost control, the Cubs’ hitters are thriving and look poised to take the major leagues by storm.

But what about pitching? Well rather than bearing the risk of developing young pitchers, the Cubs have waited and pounced later in the process. Earlier this year, for example, they acquired a pitcher by the name of Jacob Turner on waivers from the Florida Marlins. Turner’s stock had fallen considerably (since being the main prospect in a package the Marlins acquired from the Tigers) as a result of his earned run average jumping from 3.74 to 5.94. But statistical analysis in baseball has revealed that a pitcher only really has control over walks, strikeouts, and home runs, and that therefore fielding-independent pitching (FIP) is a better judge of pitching talent. Turner’s FIP had, in the same short time period, improved from 4.43 to 4.00. He had suffered bad luck as a result of his opposition’s hit location and the fielding behind him, and the Cubs were able to grab a pitcher poised for a bounce-back to complement their stable of young hitters. Rinse and repeat.

In hockey, hit location and fielding are essentially congruent with shot location and goaltending. In our sport, analysts have found that defensemen have very little control over either team’s save percentage while they’re on the ice, and that therefore the best method we have for evaluating them at this point is using Corsi% – in unison of course with contextual factors like offensive zone start percentage and quality of teammates. Corsi is essentially our FIP, and while baseball and hockey may be vastly different sports, the approach the Cubs have taken to rebuilding is one that could be replicated in hockey quite easily. Here is a chart of young defensemen who have either been linked to trades or could potentially have been available in the near past.


Screen Shot 2014-08-12 at 11.14.38 PM

Some of these guys were available based on low PDOs that could have largely been the result of variance, some simply didn’t fit the mold of defensemen coaches were looking for, and some may not have been on the block at all, but for the right package might have been had. And this is only a very brief list of the most obvious and enticing candidates.

One of the largest market inefficiencies in pro hockey is an under-appreciation of young, puck-moving defensemen. Some of these guys could have been acquired, and teams that had spent draft picks primarily on forwards – hockey’s draft equivalent of baseball’s hitters – likely would have had the resources to get them, rather than wasting picks on guys like Alex Plante, Keaton Ellerby, Dylan McIlrath and Tyler Cuma.

To bring in yet another sport, Sacramento Kings owner Vivek Ranadive recently stated that ” In the 21st century, math will trump science…you don’t need to know the why, you simply need to find the pattern”. While I would disagree that this is always the case – we can be fooled by flawed analysis if we don’t understand the science behind it – with strict scrutiny he is essentially correct. It doesn’t matter why forwards are easier to forecast than defensemen and goalies, and it doesn’t even matter if the strategy is generalized to the extent that teams might miss out on certain star defensemen the scouts dub sure things. On a macro scale, drafting mostly forwards and then exploiting trade inefficiencies on young defensemen and goalies is a strategy that would bring value. The math is there to prove it.

An NHL team that decided to go in this direction would have a massive competitive advantage – although with the recent hires it’s shrinking every day – they would have first mover’s advantage in an industry in which the method is already tried and on the verge of being proven true.

If hockey executives spent a little bit less time downplaying the relevance of baseball’s findings and little bit more time learning from them, they could change the game for the better. They could innovate. And in doing so, they could build a winner. Analytics is about information, adaptation, and innovation. It’s time for the dinosaurs of hockey to head John Henry’s warning. Tear down and rebuild. Learn from the tried and true.

Introduction to AP Hockey

IMG_0488In June of 2010 I joined Twitter, and I had no idea at that point what an impact it would have on my ability to analyze and process the game of hockey. In the four years since, I’ve read thousands of articles, and was fortunate to be welcomed into Habs Eyes On The Prize, able to share my writing, research, and analysis with its loyal and diverse readership. Don’t worry, I’m not stepping down from anything, and I hope to continue to contribute analytical Habs-related content to EOTP, but I decided near the end of last season that I wanted to start my own blog, giving myself ultimate editorial freedom and the ability to jump from team-to-team, sport-to-sport, topic-to-topic and serve what I hope will become a readership of my own. Now that we are nearing the beginning of another NHL season, I thought it was the right time to act on that decision.

So what will this website present, exactly? First and foremost, when I read something I find interesting, rather than adding my thoughts in a series of tweets, I will post here. But beyond that, I hope to contribute analytical work focusing more on the league as a whole, some film study breaking down systems and optimal tactical decision-making, as well as random musings on anything from the analytical revolutions in other sports to the best of Bruce Springsteen. Imagine this as a medley between Tom Tango, Joe Posnanski, and Mc79 Hockey. Obviously feedback (both positive and neutral) is welcome, and anybody with an idea as to how to improve the site’s layout is encouraged to help out.

I don’t mean this to be a long post, but I’ll leave you with this. When you’ve read as much on analytics as I have in the last couple of years, articles tend to blend together. But some stick with you. Some seep into your brain and influence who you are. Some inspire you. At some point I may do a post on the articles that have had this effect on me, but for now, I’ll leave you with one quote that shapes the way I look at any sport, and would repeat to any general manager if I had the chance.

“If we weren’t already doing it this way, is this the way we would start?”

It is in our nature to hate change; we like the way things are, and we treasure the way they have always been. There are evolutionary reasons why those aspects of our nature made sense at one point, and in some cases still do. But when it comes to an area like sports, we must fight our nature with strict scrutiny, because our nature isn’t optimal. We can, and must, do so much better. I hope you all enjoy the site.