Game 17 @ CAR: Is the Metropolitan Division Underrated?

Before this blog’s third inception, I decided that one of the changes I make would be to keep a schedule with topics assigned to certain games, so that I’d have some sort of inspiration lined up and avoid that slippery slope of skipping articles until I gave up on the site altogether. I even put together a backlog of emergency ideas, just in case I had neither an established idea for a game nor a reasonable amount of time to research one. And knowing what I’d write about in advance gave me plenty of time to think up bad puns to title my posts (one of the purer joys of blogging). It seemed brilliant at the time.

With that in mind, the rhetorical question for today’s post was going to be “How Bad is the Metropolitan Division?”, and the answer was going to be Pretty Fucking Bad. We’ve all seen the numbers – a woeful 28-39-11 record against other divisions and 5 of its teams in the league’s bottom 9, and they were even worse two weeks ago. Yes, I told myself, these are the sort of outlier statistics that’d make really interesting talking points! And the Hurricanes are in the Metro, which makes this the perfect game to write about them because what else do I even know about this team?

And then the day came around and as it turns out, a whole truckload of other better-written articles have already covered it.

Fortunately I remembered this is an advanced statistics blog, and that a lot of these underlying numbers are meant to help us look past a team’s record and into the qualities that help drive wins and scoring. I didn’t think anybody had checked if the Metropolitan Division was as bad as they appeared, and was curious to see if the data supported it or not. There’s no doubt that the West is crushing the East this year as it has been for ages, but is the Metro really that much worse than the Atlantic?

PDO is considered by many analysts to be the most useful metric to quantify luck in hockey. As with most advanced stats, it isn’t actually an advanced concept at all – it amounts to team shooting percentage plus save percentage at even strength, both of which are considered to be largely driven by random chance. For this reason it could be construed as luck with goaltending effects (presumably a team starting Henrik Lundqvist can expect a better save percentage than one featuring Karri Ramo), though it’s been argued by people far smarter than myself that the talent of goaltenders doesn’t show over a single season.

I personally feel that PDO is most useful when applied to players, as we can then compare his PDO to that of the team when he’s off the ice and remove a lot of variables (goaltender quality, shot counts by arena, and other team effects). If the team sees better bounces when a player is on the ice, it stands to reason that his numbers will be inflated and his play will better pass the eye test, and vice versa. It’s a great way to evaluate players for fantasy hockey as far as buying low and selling high goes, because – and here’s the rub – it will very reliably regress to 1000 over time.

This should make sense intuitively – if you flip a hundred coins and the first ten come up 90% heads, your remaining flips should still be fifty-fifty. They might remain at close to all heads by pure fluke, but a better bet is that your overall record will take a sharp turn back towards the expected value. It’s a massive oversimplification for hockey, but given enough data points the same arguments apply – just replace the odds of flipping heads with those of scoring or saving a goal. A player shooting at well above his normal rate is expected to decline, as probability dictates that his future shooting percentage will most likely fall at his career average.

For teams, this is best shown in some excellent research by Tyler Dellow. Dellow takes the top-20 and bottom-20 PDO teams from the first quarters of each of the 2003-2008 seasons, and then charts their percentages over the remainder of that season. The result is startling – regardless of how well the teams did to begin the season, their PDO at large displayed zero sustainability over 82 games. When people talk about regression to the mean in hockey, they almost invariably refer to a team or player’s PDO and its likelihood to come to league average.

So what does that mean for the Metropolitan Division? If we check out their even strength statistics, we can see that their PDO numbers are pretty ugly. Not one of them is over the mean of 1000, and all eight are in the bottom half of the league. On average, their PDO is 98.4 – a decent chunk below any of the other three divisions, particularly the Central and Atlantic. This speaks to some poor luck going against them, especially if we consider that their possession numbers are fairly close to 50% (though we should bear in mind that this is an extremely crude analysis as many of these Corsi events come in intradivisional games and so are counted twice).

pdo

Why has the Atlantic seen more success? To put it succinctly, goaltending. Their records have been posted on the backs of established stars like Tuukka Rask and Carey Price, or phenomenal tandems such as the Maple Leafs’ James Reimer and Jonathan Bernier, or even Ben Bishop posting a .930 ESSV% despite only being able to move in diagonals. The division has three teams with top-5 save percentages, with Ottawa at 8th and Detroit 12th. Their PDO tells a similar tale – sad-sack Buffalo and Florida notwithstanding, all contending Atlantic teams are above 1000, including 4 in the top 7. This has proven tremendously helpful in dealing with possession numbers that have lagged behind the rest of the league.

The Metropolitan Division, on the other hand, is drowning in poor shooting percentages with 5 teams in the bottom 7. These numbers are highly nebulous and all wisdom indicates that there is no reason they should remain that way for long. Given PDO regression, there’s no reason the Metro shouldn’t surge to at least level with the Atlantic (especially after the brutal injury to Steven Stamkos earlier today). New Jersey in particular should see a major turnaround, as they have extremely strong possession stats and two solid goaltenders in Cory Schneider and Martin Brodeur. The Rangers have also suffered from some uncharacteristically poor goaltending and can expect to be in the playoff mix as their shooting numbers improve and Henrik Lundqvist finds his form.

For the Avalanche, this doesn’t reflect well – as of this writing our shooting and save percentages are both overblown for a PDO of 104.5, tops in the league. As I’ve mentioned before, our possession numbers are still quite good, with Corsi tied now at 8th in the league (though Corsi close has dropped to 18th, as the Avalanche led by one for much of the last two games and never trailed), but it stands to reason that we’ll see some one-goal games go the other way soon, or even an extended losing streak like the Sharks are currently under. Once some of these bounces stop going our way, the results may not be pretty.

Some say you should stay up all night to get lucky, but I wouldn’t lose too much sleep over it – as the Metro could attest, sometimes shit just happens.

Advertisements

About Arthur

I like the Colorado Avalanche and tolerate the Toronto Raptors.
This entry was posted in 2013-14 Season, Colorado Avalanche and tagged , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s