Skip to content

Leafs going all-in on advanced stats

Aug 25, 2014, 9:46 PM EST

Dubas

The Toronto Maple Leafs were known as a franchise that was dismissive of advanced statistics. What a difference one summer can make.

New team president Brendan Shanahan shook up the Leafs’ front office by bringing aboard 28-year-old Kyle Dubas to serve as an assistant general manager. Known as a strong proponent for advanced statistics, Dubas proved to be just the start of Toronto’s new movement.

The Maple Leafs have since added extraskater.com founder Darryl Metcalf along with Cam Charron and Rob Pettapiece to help create their analytics department, according to Puck Daddy and Sportsnet.

These moves are punctuated by Shanahan’s recent statement that “we have people in our organization who have maybe been afraid of certain words and certain information.”

It will be interesting to see how that old guard meshes with Dubas’ new department, but it doesn’t have to be a battle. Dubas doesn’t see analytics as a replacement for traditional scouting, he merely believes it’s a tool to help in the evaluation process. At the same time, if everyone’s on board, this could lead to a shift in how the Maple Leafs’ operate.

Toronto was consistently out shot last season and while they initially managed to win in spite of that, the Leafs eventually fell from grace. That descent wasn’t surprising in the advance statistics community, which had largely pegged the Leafs as a team that was over performing. Although there are those that would argue that shot totals are an overrated statistic, it is something that the Maple Leafs appear to want to address.

That desire to change isn’t surprising either. Throwing everything else aside, the Toronto Maple Leafs allowed 3.07 goals per game last season and there can’t be anyone within the organization, regardless of their opinions on advance statistics, that finds that acceptable. For that matter, everyone can agree that the win column matters and there’s no question that’s one statistic the Leafs have been sorely deficient in for years.

Clearly improvements need to be made and if the Maple Leafs’ tactics thus far had led to results, then the hiring of Dubas and the founding of an analytics department might not have attracted as much attention. As it is, the Leafs seem to be a prime candidate for change and while the impact and influence of analytics on the team remains to be seen, their recent hires suggest that they are willing to embrace these relatively new ideas.

  1. jl9830 - Aug 25, 2014 at 10:14 PM

    I love that this is making guys like Steve Simmons’ head explode.

    • withseidelinn - Aug 25, 2014 at 10:32 PM

      Don’t know how TSN lets that guy speak

  2. jinx21fan - Aug 25, 2014 at 10:29 PM

    There’s a bunch of bloggers that watched “MoneyBall” and want to push “advanced stats” into hockey. The big difference, baseball is an individual game played by a team. A batters’ stats are his alone. Which is good, because it’s his abilities alone up their facing the pitcher. His fielding stats’ are his alone.

    Hockey is way to fluent of a sport, with far to many variables on any given play to be able to truly rely on “advanced stats”. Team’s play under system’s implemented by coaches. Do they dump and chase more often? Do they apply pressure in the D-zone or collapse around the net? Do they have confidence in their goalie and play more open? Are they told to look for high percentage scoring chances or throw everything at the net? Even the Leafs new so-called “guru” acknowledges that they are a small “tool” to help augment traditional forms of scouting.

    • blomfeld - Aug 25, 2014 at 11:09 PM

      “Hockey is way ‘to fluent’ of a sport” …

      perhaps friend you meant to say ‘too fluid’ of a sport ? … not trying to be picky, however it’s important that strive to uphold the ‘King’s’ English at all times here at PHT ! :)

      • jinx21fan - Aug 25, 2014 at 11:32 PM

        Yes indeed, I always try to uphold the Queen’s English! Sometimes fail, sometimes pass.

        Fluent may well be more associated with linguistics, but it is very acceptable within the context I use it as well. “Fluid” also would have worked, I guess that’s why it’s listed in the thesaurus when you look up “Fluent” :)

    • dueman - Aug 25, 2014 at 11:18 PM

      With headlines like, “Leafs going all-in on advanced stats,” it doesn’t really matter what the guys that Toronto hired actually said now, does it…

  3. endusersolutions2013 - Aug 25, 2014 at 10:37 PM

    I’ve been a clinical analyst since 1977. I don’t do “voodoo”. I use the information available to help Drs, RNs, Rph’s, microbiologists… to efficiently assess available information and take better care of patients.

    I don’t understand what the controversy is. Staff provides good info for decisionmakers to make good decisions. We don’t replace them.

    • dueman - Aug 25, 2014 at 11:30 PM

      There is nothing wrong with collecting, and using statistics to help come to decisions, but the key word is “help” in that statement. Some stat guys take this to the extreme. Stats are useless without context, and in hockey, a lot of times that context is very hard to define. You can’t Moneyball a hockey team. But you can use some advanced statistics to help out in certain aspects of the game. My issue is with all those guys out there that throw around corsi numbers like they mean everything, when in reality, without proper context, they mean nothing…

      • atwatercrushesokoye - Aug 25, 2014 at 11:41 PM

        If I’m not mistaken corsi is shots on net + shots that missed the net + shots that were blocked, I have no idea how you could possibly throw that around like it actually means something? And I say that hoping that once of these moneypuck people will set me straight and tell me why that statistic is important?

      • rbbbaron - Aug 26, 2014 at 1:37 AM

        “If I’m not mistaken corsi is shots on net + shots that missed the net + shots that were blocked” more or less, but don’t forget that it’s a differential, like the +/- .
        “I have no idea how you could possibly throw that around like it actually means something?” It’s considered to be a possession stat, the idea being that it gives some insight into which players drive or otherwise contribute to puck control, just as the +/- does to some extent with goals scored. It’s not a terribly “advanced” statistic (we should’nt be afraid of arithmetic); it’s simply using more data in effort to capture a more detailed picture of what happens while a player is on the ice.
        For the record, I’ve neither read nor seen Moneyball; I just generally believe that more detailed data and measures can better inform decision-making.

      • beergold - Aug 26, 2014 at 1:09 PM

        So the fact that when Kessel puts up 80 points a season but he is on the ice for 85 plus goals against this statistic means nothing. Seems to me it means he is a career “negative” player that is for certain!

  4. atwatercrushesokoye - Aug 25, 2014 at 11:37 PM

    Maybe at 34 I’m old, but I don’t really understand how a lot of these advanced statistics actually tell you anything, for instance how does knowing how many times a player missed the net prove that they’re a better or worse hockey player? And then add in missing the net plus shots blocked, how does that make player X a better player and give me insight into something I don’t already know?

    Granted there are likely some very good things you can gain from these statistics and teams should analyze all data available to them (some with a grain of salt) but teams have won the Stanley Cup almost every year (thank you Spanish flu and Gary Bettman) for a long time now and there have been some extremely good teams built in that time so I’m not sure that this will change the landscape that much.

    • hockeydon10 - Aug 26, 2014 at 9:30 AM

      I’m no expert.

      A player missing the net or having his shot blocked proves that team had the puck in the offensive zone at the very least. It may also show that they were in a position to get a shot in the right general direction. I’m pretty sure I read that neutral zone dump-ins on net (like for the purposes of a change) don’t count toward these stats.

      This is why people say it’s only part of the information GMs & coaches should use and not put too much weight on it.

      The push-back is that a player with good adv stats on a bad team may really be better than the team’s W/L record indicates, but a player with good stats on a great team may not be that good. Put a guy on a line with Datsyuk, Toews, Bergeron or Kopitar and their adv stats line will look much better than being on a line with a 4th liner called up yesterday for Buffalo.

      • rbbbaron - Aug 26, 2014 at 10:14 AM

        “Put a guy on a line with Datsyuk, Toews, Bergeron or Kopitar and their adv stats line will look much better than being on a line with a 4th liner called up yesterday for Buffalo.”
        That’s where relative/within-group stats come into play, and that’s where the statistical procedures can become said to be “advanced” proper. Theoretically you could infer how much a guy’s stats go up simply as a result of the play of Datsyuk, Toews, etc., and how much of that is his own doing or otherwise an interaction effect. It all depends on how good the data is.

      • jinx21fan - Aug 26, 2014 at 5:18 PM

        So, if the data is good, you can “theoretically infer” part of the equation? Compelling argument for advanced stats.

      • rbbbaron - Aug 26, 2014 at 8:18 PM

        You can always just argue and compare players with anecdotal evidence, because that always yields the best results, right?

      • jinx21fan - Aug 26, 2014 at 11:29 PM

        Well there’s that. There’s also the stats that you don’t have to “theoretically adjust” to conjure up a conclusion. You know, like goals scored little things like that. I know those stats mean nothing to some, but they’ve always seemed a pretty good indication of talent.

        Like here are some pretty straightforward stats:

        Benoit Pouliot(career) GP 371 G 76 A 84 Pts 160

        And, to appease your sarcasm, he tends to take far too many ill-timed penalties (anecdotal evidence provided). So I say 5yrs/20M is an absolute crazy.

        So explain to me why his “real” underlying stats make him a 4M/yr player?

      • rbbbaron - Aug 27, 2014 at 2:50 AM

        Oh I see, so numbers are in fact ok now. But only the simple ones, like points scored? Is that your argument?

        “So explain to me why his “real” underlying stats make him a 4M/yr player?”
        Sorry, I don’t really see how this is relevant to the topic, let alone what I was discussing above. When did I or any other person sympathetic to sports analytics claim to be able to explain any given player’s salary with said analytics? I have a hard time gauging whether this is an intentional red herring or if you just don’t grasp the general idea behind the quantitative research agenda.

      • jinx21fan - Aug 27, 2014 at 4:28 AM

        Well, I’m done with the p*ssing contest. You enjoy researching player X’s defensive zone possession numbers, I’ll enjoy yelling at player X to clear the f**king puck when I think he’s held it too long. We all live in peace and harmony.

  5. blomfeld - Aug 25, 2014 at 11:39 PM

    ABSOLUTE HOGWASH !

    The bottom line is that this BS has been going on since the beginning of time, whether it was some Arab huckster in 320 BC trying to sell three-headed camels who ‘sing’ like Elvis, or some two bit snake oil salesman in Dodge City (circa 1880) trying to sell his ‘magical’ healing potions to anyone foolish enough to listen. MLSE is a corrupt and ‘morally’ bankrupt corporation and their paid ‘liars’ can blab on about analytics until the cows come home for all I care, because in the end it won’t make a difference. This ‘advanced stats’ crap is nothing more than ‘smoke & mirrors’ which is designed to take people’s attention away from what’s really ailing this failed enterprise … a ‘complete & utter’ absence of values. Nothing more and nothing less. Trust me, when Alec Martinez scored the SC winning goal this year against NY, he sure as hell wasn’t reading from some stupid spreadsheet ! Same thing when Captain Brown scored that beauty in OT, or when Carter went ‘top shelf’ in game 3 ! In fact, the only reason I even give a hoot about this organization is because we SC champion LA Kings ‘gifted’ them the planet’s 3rd best goalie in the form of Jonathan Bernier. If it weren’t for him, then these people could drop off a cliff tomorrow and I wouldn’t give it a passing thought.

    F DUBAS !!

    F SHANAHAN !!!

    F ADVANCED STATISTICS IN HOCKEY !!!!

    • blomfeld - Aug 25, 2014 at 11:43 PM

      *** The Planet’s Top Goalies as per the Sports Desk @ Stern Magazine ***

      1) Jonathan Quick
      2) Martin Jones
      3) Jonathan Bernier
      4) Cory Schneider
      5) Henrik Lundqvist

  6. ajpeebles - Aug 25, 2014 at 11:56 PM

    They’re puck possession stats. The theory is if you take more shots (whether they are blocked or miss the net or hit the net) it means you possess the puck more and therefore generate more scoring opps while reducing the opposing teams ability to attack. These can be tweaked by comparing if they’re reg strength or power play as well as comparing the quality of competition.

    • atwatercrushesokoye - Aug 26, 2014 at 12:03 AM

      That’s fair, I guess that assessment can work at times, although is there a way to take into account if a team turns it over in their own zone and you get off a quick shot that hits the net, misses the net or is blocked? Using the statistics just as you described it might look better than it is, as an example: Team A keeps the puck for 10 seconds, turns it over, team B player quickly 1 times the loose puck and misses the net, team A then takes possession again and carries it into the other zone? Using the stats it might look like Team B had puck possession and a chance but in reality it was a wasted effort with no possession.

      *Not challenging what you say in any way, I am just quite curious about advanced stats and was genuinely curious of if they take into account things like that?*

      • rbbbaron - Aug 26, 2014 at 2:42 AM

        Like most stats, the corsi is a bit raw in terms of telling a full story about what happens on the ice. There are situations like the one you describe which can be confounding for people who believe the corsi is purely a “puck possession” thing, but it’s not hard to create such situations for nearly every stat: for example in terms of goals scored think of how the Datsyuk end-to-end goal, by most accounts an amazing individual effort, counts as the same a Holmstrom deflection goal, where sometimes he didn’t even see the shot coming but it grazed off his stick and was therefore awarded to him. Statistically, on paper, these plays are equal and someone who couldn’t observe anything beyond that wouldn’t really be able differentiate between those two plays and by extension the two players in terms of their goal-scoring ability. The moral of the story is that no single stat is the be-all end-all; instead it’s usually more useful to look at a number of different stats over a longer horizon in order to get a better idea about a player when you likely don’t have the time or ability to directly observe all or “enough” of his play.

        In the situation you describe it might also be helpful to ask what led to the turnover– for example was it good defensive coverage, forechecking, etc, from team B that forced team A to turnover the puck? If so, then I’d argue the corsi still has relevance, not so much as a “puck possession” stat per se but rather as a shots-attempted differential– and I’d emphasize the differential part of that, meaning that it can also take into account defensive play.

  7. hockeyflow33 - Aug 26, 2014 at 1:45 AM

    I share the same opinion as most everyone else on here that advanced stats aren’t very helpful for hockey. There really are just too many variables to use quantifiable metrics to measure player’s skill levels.

    What never gets brought up is these guys who develop and push these “advanced stats” have the potential to turn their stats into multi-million dollar companies or consulting jobs that would otherwise never be there.

    • rbbbaron - Aug 26, 2014 at 2:49 AM

      “What never gets brought up is these guys who develop and push these ‘advanced stats’ have the potential to turn their stats into multi-million dollar companies or consulting jobs that would otherwise never be there.”

      More than just “potential”, there surely are sports analytics companies that are worth millions and guys getting hired as consultants (note: I’ll forever hate Toronto for taking Extra Skater away from us– that was a damn good site). But why is that? Are the owners/managers of these franchises just fools who don’t know better? That’s a difficult thing to argue…

  8. Hard to BeLeaf - Aug 26, 2014 at 6:03 AM

    All I know for sure is if things go well, the stats guys with get 100% of the credit. (even though they haven’t done anything yet)… If things go bad, Carlyle and Nonis will get 100% of the blame.

  9. clpalermo94 - Aug 26, 2014 at 6:57 AM

    Advanced stats is the moneyball of the nhl.

  10. ballistictrajectory - Aug 26, 2014 at 9:31 AM

    How does one “oppose” advanced stats?

    GM [hands over ears]: “I’m not listening!!! LA-LA-LA-LA-LA!”
    Coach: “Do you hear that noise? Where’s it coming from? I hear voices speaking numbers.”
    Expensive Free Agent: “I’m pacing myself for the playoffs.”

    Really. Either the numbers mean something or they do not. If the team’s shot blocking is too low the shots against goes up. Unless you have the best of the rest in net your GA goes up too.
    Same happens at the other end. Trend up the shots on net and the GF generally goes up too.

    Broadcast Commentator [enthusiastically]: “When he ties his skates with his own hands he’s 25% more likely to put the puck in the net. If his mother does it he tends to follow the play and not participate as much.”

    I’m interested from the point of measuring relative skills but after that it’s just too much. I’m curious as to how the last 25 Cup winners measure up with advanced stats. The coaches & scouts have been compiling this data for at least that long, although I doubt they’ve retained it. I would be nice to see if the team with the Cup in the air is the team that measures out to be the “most skilled.”

  11. rbbbaron - Aug 26, 2014 at 11:17 AM

    “I’m curious as to how the last 25 Cup winners measure up with advanced stats. The coaches & scouts have been compiling this data for at least that long, although I doubt they’ve retained it.”
    I’d disagree– if the coaches and scouts ever had such detailed data as there is now, then you can be sure they’ve still got it. But I’m pretty sure they themselves haven’t been collecting that data, certainly not going that far back. Rather the league has been collecting more detailed data for some time but only in the last decade or so made some of it available (for example, total giveaways, turnovers and take-aways were only published in the last few years, I can only find faceoff percentage numbers going back to 2000, but then to have these numbers broken down into own/opposing/neutral zone stats? That kind of thing is more recent).
    Transparency is a huge problem right now– few if any sites indicate how/where they got their data from, and given how many sites there are I’d have a hard time believing it’s all proprietary. Much of the resistance to analytics from NHL management has revolved around concerns about the quality of the data; that suggests to me that the teams haven’t been gathering such detailed data themselves (at least not going very far back).

    • wfoddis - Aug 26, 2014 at 1:04 PM

      Here’s the Corsi (Corsi-For %) of the conference finalists for the last 7 seasons (source: hockeyanalysis.com) with the Stanley Cup champion listed first and other finalist marked with an asterisk (*).

      2014 – LAK(1), CHI (2), NYR* (7), MTL (26)
      2013 – CHI (4), LAK (1), BOS* (3), PIT (18)
      2012 – LAK (2), PHX (15), NYR (23), NJD* (12)
      2011 – BOS (14), TBL (8), SJS (1), VAN* (5)
      2010 – CHI (1), SJS (12), PHI * (13), MTL (27)
      2009 – PIT (19), CAR (8), CHI (3), DET* (1)
      2008 – DET (1), DAL (9), PIT*(29), PHI (25)

      Eye-balling the stats, the association between Corsi and conference finalist looks to be reasonable. More notably, 5 out of 7 years the SC champion was top 5 in Corsi. And the years the winners weren’t top 5, the other finalist was top 5 (i.e., Detroit (1st) – 2009; Vancouver (5th) – 2011). Obviously, the correlation won’t be perfect (no statistic is perfectly predictive). I believe Michael Parkatti (boysonthebus.com) did a correlation analysis and found that Corsi was in the .3 to .4 range in predicting the SC champion.

      • Hard to BeLeaf - Aug 26, 2014 at 4:36 PM

        “no statistic is perfectly predictive”
        Goals are. The team that scores more goals will wins 100% of the time. Guaranteed!

        And of course Corsi won’t correlate with Stanley Cup champs… it’s a minor stat. I think it’s funny how little respect +/- gets, but people act like Corsi and Fenwick stats are so “amazing”. They’re just modified +/- stats for shots instead of goals. Nothing to get overly excited over. Shots definitely don’t tell you who will win.

  12. vancouversportsbro - Aug 26, 2014 at 11:32 AM

    I’m a big disbeliever in advanced stats in sports, especially hockey. No amount of stats will take you to the cup.

    Funny how the most made fun of teams in the leagues hired these guys (leafs and oilers)

    • rbbbaron - Aug 26, 2014 at 12:01 PM

      “No amount of stats will take you to the cup.”
      I haven’t done the regression, but I imagine goals for and goals against are pretty good predictors of team success…

      • wfoddis - Aug 26, 2014 at 12:18 PM

        Actually, goals for/against is not as predictive of future performance as shot attempts because of too small a sample size in a season. Michael Parkatti (boysonthebus.com) and others have done this analysis.

      • rbbbaron - Aug 26, 2014 at 1:06 PM

        That’s actually where I was going with it: “where do goals for/against come from?”–> “where do shots for/against come from?” etc., the case for detailed statistics becomes clearer

  13. obamareallysucks - Aug 26, 2014 at 3:22 PM

    leafs suck like obama

    • Hard to BeLeaf - Aug 26, 2014 at 4:38 PM

      Cool… So they’re going to win twice within 4-5 years and everyone will hate it?!!

      I can live with that.

Featured video

Holiday wish lists for NHL teams
Top 10 NHL Player Searches
  1. B. Bishop (2678)
  2. C. Perry (2354)
  3. B. Elliott (2074)
  4. S. Crosby (2070)
  5. S. Weiss (2026)
  1. J. Howard (1827)
  2. S. Varlamov (1722)
  3. J. Schwartz (1637)
  4. N. Kronwall (1502)
  5. S. Downie (1453)