• Context and the Canucks

    by  • April 24, 2011 • Post, Uncategorized • 14 Comments

    With the Canucks having suffered a pair of ugly losses to the Blackhawks to turn a 3-0 lead into a 3-2 lead and the Canadiens having pissed away a 2-0 lead earned on the road in Boston, times look dark for the teams carrying Canada’s banner in the playoffs. Or so you’d think if you buy into the idea that the narrators of sport push upon us, that there’s a sort of cut and thrust to a playoff series, with momentum swings driving success and failure, subject to any history between the teams that can be used to tell a more interesting story.

    In reading the writing about the Canucks-Blackhawks series, I’ve noticed an awful lot of discussion about past Vancouver teams falling apart against Vancouver. This is the sort of stuff that drives me nuts because it’s utterly unfalsifiable. While understanding the history can help us understand what members of the team might be thinking about this, it’s not really evidence that helps us understand what might happen going forward. I’m not really a fan of this sort of writing – it’s context that helps us put this specific moment in the context of the broader history of the Vancouver Canucks, as opposed to context that helps us understand what 7-2 and 5-0 losses mean in the context of this series. Media in general, and sports media in particular, are good at producing the former but they have an irritating tendency to try and pass it off as the latter.

    I dug into some data from playoff series played between 1987 and 2010 to test this idea. I chose 1987 as my baseline because that’s the year that the NHL went to having all playoff series be best of seven. I’ve excluded series that weren’t played in the 2-2-1-1-1 format from my dataset – basically, that leaves out the Oilers-Bruins series in 1988, a Caps-Pens series in 2000 which appears to have been set up as 1-2-1-1-1-1 for reasons that I can’t uncover and some series played during the 1990s when the NHL flirted with 2-3-2 for some series.

    VANCHI2

    Let’s start by looking at things from a great distance. What’s the record of the higher seed in specific games? It’s in the table at left. The higher seed plays games 1, 2, 5 and 7 at home, of course. Unsurprisingly, it has a better record in those games, with a .613 winning percentage in games played at home and a .510 winning percentage in games played on the road. I’m actually sort of surprised at the size of the home ice advantage, to be honest – you would think that the travel factor would be somewhat neutralized in the playoffs, as most games are against teams that have had similar travel to you.

    There’s some fluctuation there but nothing particularly out of the ordinary – a .624 winning percentage in G2 isn’t that surprising in the context of a .613 winning percentage overall. Some math in Excel tells me that we’d expect .624 or better to happen with a true talent of .613 in a 338 game sample more than 35% of the time. It’s well within what could be random. Same thing goes for the record of .500 in G6 – we’d expect a winning percentage of .500 or lower over 87 games for a team with a true talent of .510 more than 41% of the time. It’s all pretty tight.

    VANCHI1

    What if we look at the record of the higher seed in terms of the score in the series? That’s in the table at left. I have, for the sake of clarity, colour coded the deviations of the higher seed from the expected winning percentage based on whether it was home or away. Blue means a deviation of less than +/- .010 from the expected winning percentage, orange is between .010 and .020 and yellow is in excess of .020.

    It’s the boxes coded with yellow that are most interesting, I think. Higher seeds have much higher winning percentages than expected when down 1-0, down 2-0, up 3-0 and up 3-2. They have much lower winning percentages than expected when down 3-0 and down 3-2.

    It’s tougher to pull what that means from it. If I was a sportswriter, my explanation would probably be something along the lines of teams performing with their back to the wall. While there might be something to that, I wouldn’t arrive at that as a conclusion on the evidence presented. It might also be that coaches of the higher seed tend to coach more desperately when they’re down 1-0 or 2-0. This might, for example, take the form of shortening the bench earlier in the game or just generally directing more minutes to their stars. In other words, it might be desperation, but desperation driving the manner in which the team is coached, rather than motivating the higher seed on the brink to do more or the lower seed having been lulled into a sense of “We got what we came here to get” after a win. This is something that could be checked – if anyone’s interested, I have some ideas.

    The deviation at 3-0 and 0-3 is probably pretty easy to explain as well. I would suspect that, in those cases, .510 is probably not the expected winning percentage for the higher seed. If they’re up 3-0, they’re probably significantly stronger than the average higher seed or facing a significantly weaker lower seed. The converse is probably true when you talk about higher seeds being down 0-3 – odds are that they’re either a weak higher seed or facing a strong lower seed.

    The difference in winning percentages at 3-2 and 2-3 is more difficult to explain and, to be honest, I can’t think of a plausible reason for why it might be the case, as opposed to simply being some randomness. Changes in coaching style are all I can come up with – maybe coaches tend to empty the tank to win it in game six when they’re up 3-2? It’s not particularly compelling.

    Implicit in the argument that the Canucks are in trouble now is the idea that the Canucks that a 3-0 lead in the series that has now turned into a 3-2 lead in the series is worse than a 3-2 lead earned in some other way. If the series hadn’t unfolded, from the Canucks’ perspective, as WWWLL (with the two Ls being ugly), nobody would be talking about this. Chicago’s a pretty good team despite having committed roster seppuku immediately after winning the Stanley Cup. Their goal differential was north of +30 this year. They’re a good, but flawed, team. It’s the manner in which this has unfolded that has people talking about the Canucks collapsing.

    Does going from 3-0 to 3-2 tell us something this series that simply knowing that it’s 3-2 doesn’t? 119 higher seeds have gone into a sixth game of a series, leading the series 3-2. Their record in those games six is 67-52, good for a .563 winning percentage. Of the 52 teams that lost game six, their record in the following game seven is 30-22. That means that higher seeds leading a series 3-2 heading into game six have a record of 97-22 in the period I’ve looked at, an .815 winning percentage.

    vanchi3

    We can break this down a bit though, in order to look for some evidence that teams feel their collars tighten as a lead disappears heading into that game six on the road or gain confidence as they storm back from an early deficit in the series. Unsurprisingly, it’s difficult to find much of anything in the numbers. Teams that went up 3-0 have done well in game six on the road, despite having lost two in a row. Going back a step further, teams that opened 2-0 have a very strong record as well, despite having lost two of three heading into game six. You’d be hard pressed, on the evidence, to make any sort of an argument that the recent history supports any sort of an argument that the Canucks are in any sort of a different position than the typical higher seed that leads 3-2 simply because they went WWWLL.

    That leaves only one thing to consider: is there evidence to support an argument that the Canucks are in a worse position than the typical higher seed because the roof has completely fallen in over the past two games, with the Canucks losing by a combined score of 12-2? As it so happens, there does not appear to be an example, in the past 23 years, of a higher seed losing consecutive games by a total of ten goals. This series is unique.

    There are plenty of examples of higher seeds losing back to back games by at least six goals though – 55 in total. Higher seeds haven’t done well in those series, winning just 9 of the 45 series in which this has occurred, a 20% rate. Curiously, they’ve done better when they’ve taken the beatings early on in the series: 3/7 teams that lost the first two games by a combined 6+ goals won the series, 1/12 for games two and three, 3/15 for games three and four, 0/8 for games four and five (troubling if you’re a Canucks fan), 2/10 for games five and six and 0/3 for games six and seven. The samples on that stuff are all tiny, of course. It’s tough to know what to make of it though – even if you push it late in the series before taking a whipping, things haven’t usually ended well.

    One final piece of data: I mentioned above that higher seeds leading 3-2 win the series 81.5% of the time. According to the implied odds from Betfair, the Canucks have about a 75% chance of winning the series.

    About

    14 Responses to Context and the Canucks

    1. Saj
      April 24, 2011 at

      Great job with this. It really annoys me how the media throw around the “momentum” word without giving any evidence that it’s relevant

    2. April 25, 2011 at

      I really enjoyed reading this. Great work.

    3. paulklos
      April 25, 2011 at

      I would beg to differ.

      Sure the numbers are the numbers and I bet if you had the data you could show Persia beats small Greek polis 99.9% percent of the day and that Marathon happens…

      The reality is Vancouver did not loose just games four and five they got hammered (and its not like the 3 wins bot blow-outs), and its not like the media is dredging up some 1970′s era Hawk’s run of victory over the nuck’s, but recent history that involves many of the current team(s) members.

      People remain people and not math models. On balance I agree if I was writing this 15 hours ago I would agree that Vancouver still had the better record and 2 chances to Chicago’s 1 to win so still likely advances on math – but once the Hawks won a second decisive win and going home I would not make that bet.

    4. paulklos
      April 25, 2011 at

      Still on my first cup of joe, sorry for the spelling errors above.

    5. Simon Lamarche
      April 25, 2011 at

      @paulklos

      If the Hawks win, people are gonna talk about momentum and how the Canucks collapsed. We’re probably gonna hear about how Luongo is a choker.

      If the Canucks win, people are gonna talk about how, when cornered, the Canucks rallied. And about how clutch Luongo is.

      We love to see patterns in what happens in the real world and we are wired to find the easiest explanation possible.

      If you look at the goal differential for those 2 teams for the last 6 games, I agree that the Hawks look like the better team and maybe there is such a thing as momentum.

      {i}”…with the Canucks losing by a combined score of 12-2? As it so happens, there does not appear to be an example, in the past 23 years, of a higher seed losing consecutive games by a total of ten goals. This series is unique.”{/i}

      I think this post’s main purpose is to show that whichever team wins, there might be a better explanation (better coaching, better team, more luck) than momentum.

      Or maybe it is momentum. We just shouldn’t stop our analysis there. The first people that saw thuder thought the Gods were angry. If everyone agreed without thinking “maybe there’s something else”…

    6. Simon Lamarche
      April 25, 2011 at

      I thought the italics were worth a try…

      [i]Sorry for polluting your blog with trials, but does this work?[/i]

    7. Robert Cleave
      April 25, 2011 at

      Use the style guide right above the comment box if in doubt, Simon.

      The lesser than/greater than symbols enclosing the action you want should work.

      If you have it right, you’ll get italics.

    8. Simon Lamarche
      April 25, 2011 at

      wow… this is very embarassing

      I completely missed that…

    9. paulklos
      April 25, 2011 at

      Hey my background is econometrics so I not saying toss the stats aside or the reality is not that in the very long run the numbers are right most of the time (but didn’t Kaynes say something about that ‘The long run is a misleading guide to current affairs. In the long run we are all dead’)

      In this case however I would have be willing and wanted to modify my bet before game 6 no matter what the long term numbers say because people are still people. Chicago did not just barely grind out 2 victories that hammered their opponents. They chased Luongo twice and its not like he was 1A in some rotation system. The history of losing to the Hawks was not old but recent, etc. The media pressure cooker in a Canadian city after all the President’s trophy season hype is real.

      Thus sure the perspective is still Vancouver should likely win but at this point I doubt it, and I doubted before game 6.

    10. mkyeg
      April 25, 2011 at

      We aren’t a particularly rational species, we look for patterns and we find them much too easily. It’s good that we look, but we’re very afraid, easily scared, terrified of death, and often we are very stirred without quite knowing why. Some fairly banal examples, I suppose, are landscape and music in combination, or alone; love in combination with either of these; or perhaps looking at the vault of heaven, as Hamlet would have put it—at the “fretted gold” of the sky at night. (Christopher Hitchens)

      And he might well have added, thinking that how two teams arrived at a 3-2 position in the playoffs matters when predicting who will win the next game.

    11. DD
      April 25, 2011 at

      That Caps-Pens series I think actually ended up being a 1-2-2-1-1 format. IIRC, Pittsburgh`s arena had booked a number of events that would have made it impossible to have Game 4 in Pittsburgh so they had to adjust the schedule.

    12. Matt D
      April 25, 2011 at

      Thanks for this.
      ‘Momentum’ is my biggest pet peeve among sports journalists. It’s a description masquerading as an explanation.

    13. April 26, 2011 at
    14. Lee
      April 29, 2011 at

      The question that comes to mind immediately on this is ‘what’s the point?’ Yes, we can reduce everything in life to a math model to prove that the actual human element is negligible over large data sets. It’s certainly a provable hypothesis and one that can make us feel better about ourselves intellectually, but is it actually a worthwhile criticism of sportswriters?

      These folks are paid to write compelling copy and get the masses stirred up over the human elements of these stories. They’re not paid to be Mathematicians or expected to have every cliche they spew be subjected to the full rigor of quantitative analyses.

      Canucks v Hawks Game Seven was great theater because of the human factor. If that means buying into the ‘illusion’ of momentum, well that seems a small sacrifice to make for the entertainment value it provides. Statistics and math can certainly illuminate and enrich the spectator sports experience. It can also make it a fairly joyless one as well.

      Although, I will enjoy using some of the data you’ve provided here to enrich my betting experience over the remainder of the playoffs, so thanks for that!

    Leave a Reply

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