Wednesday, January 29, 2014

On statistics and hating probability

By BRETT CYRGALIS
            In the throes of a hockey season, and hours before one of the NHL’s showcase games outside at Yankee Stadium, I’m brought back to an essential question:
            Do you root for the probable outcome?
            In the age of advanced statistics, ones that go well beyond the realm of sports and into things that actually matter, like science and politics, there is a deep divide that separates two distinct types of people – those wanting the numbers to be proven true, and those that want the inexplicable.

             This was prompted by the smallest bit of reporting in an interesting piece by NickCostonika of Yahoo!, writing about the coach of Anaheim Ducks, Bruce Boudreau.  If you’re here because of golf, let me fill you in.
            The Ducks are fucking unbelievably good.  Here’s Costonika:

The Ducks are the top team in the NHL with 83 points through 54 games – 21-3-0 in their past 24 games, 21-1-2 overall at home. They are third in offense (3.33 goals per game) and fifth in defense (2.33 goals against per game).
This is a confident group. The Ducks are 20-2-5 in one-goal games. They are 23-4-4 when scoring first, but they’re also 16-6-1 when their opponent scores first. No other NHL team is even .500.

            But then here’s the next graf, using a bit of new-age statistics without actually mentioning it by name.
           
Are they really this good – better-than-everybody-else-in-an-era-of-parity good? Maybe not. Their shooting and save percentages are 102.3 combined, tied for second-highest in the NHL, which might mean they have been a little lucky and could come back to earth at some point.

            That bit about combining shooting and save percentage adds up to a stat called “PDO”.  It’s really an interesting view at performance, especially when it comes to bringing plus-minus back to earth.  (Plus-minus being how many goals are scored for a player’s team or against his team while he is on the ice, tallied in a rating like the “plus-20” of the Devils’ 41-year-old Jaromir Jagr, or the “minus-11” of the Islanders impending free agent defenseman Andrew Macdonald.)
            What has happened with PDO is that stat-centric pundits like to use it as a tool to declare how lucky a team has been.  That’s kind of the way Costonika used it, just to temper all of the good about the Ducks.
            And is their pace unsustainable?  The difference in answering that question now compared to 20 years ago is the fact that people looking at PDO will say it’s close to a certainty that the Ducks will come back to the mean.  I had a conversation with an interesting Twitter follower, @garik16, who said any PDOs “above 1010 or below 990 are beyond repeatable performance.”
            Beyond repeatable?  If a team has a really good set of shooters, and is getting really good goaltending, what makes it impossible that they keep it up?
            The argument is that history – in this case, going back about 20 years in which these types of stats have been kept – say it’s true.  Teams beyond these thresholds always come back to the middle.  But how different is that then what someone would have said 50 years ago, looking at nothing more than their wins and goals-for and goals-against, saying they are on pace for record-breaking numbers, and it’s unlikely for this to be an all-time great team?
            The point is, that old-time observer left himself a little more open to the possibility that what he was watching was the unfolding of history.  He allowed himself to enjoy what was before him, enjoy the greatness while it lasted.  And if it went on for the whole season, well then, hell, wasn’t that a blast?  Do you think Grantland Rice kept waiting for the ’27 Yankees to come back to Earth because they relied on the home run too much?  (Rice was a myth-maker extraordinaire, but you get the point.)
            The purpose here is not to rail against modern stats.  Personally, I think they’re great.  They create a whole new perspective on how to judge performance.  Yet to judge a hockey player on goals and assists and plus-minus is just as incomplete as to judge him on Corsi or Fenwick or individual PDO.
            These are predictors of the future, and mostly good ones.  Will the Ducks come back to reality?  Odds are they will.
            But how about the chance that they’re just the best damn team the league has seen in decades?  Can’t we enjoy that idea anymore?
            I, for one, am always rooting for stats to be proven wrong.


            Email: brett.cyrgalis@gmail.com; Twitter: @BrettCyrgalis

No comments:

Post a Comment