Predicting special teams performance in the playoffs (Predictive variables extravaganza, part 2)

In my last post on this topic I walked through a bunch of correlation charts that looked to see what kinds of things are correlated year to year, which we take to be a mark of a repeatable skill.

Conversely, if a certain area of performance is uncorrelated year to year, we take that to indicate that most of the variation between teams in that area are random rather than skill-related.

Now lets look at a more specific question, and ask: what aspects of special teams appear to be repeatable skill and which are more random variation?  The test subject in this case will be to look at performance during the 2013/2014 regular season, and how that translated to performance during the 2014 playoffs.

For example, is this a fair statement?

“These two teams are equal in most respects, but team A had a MUCH better [power play, penalty kill] during the regular season, so I give the edge to team A.”

Is it true?  Let’s take a look.  Here’s the correlation between regular season and playoff success for the PK and the PP (all data from war-on-ice).

Power Play

figure_5 - regular vs playoff pp

Wow.  Basically no correlation at all. How a team did in the regular season on the power play appears to have no relationship to how well they will do on the PP in the playoffs.

Penalty Kill

figure_6 - regular season vs playoff pk

Same thing.  Actually, this time it’s slightly negative!  But so close to zero that its irrelevant.  As with the PP, there is basically no correlation at all. How a team did in the regular season on the penalty kill appears to have no relationship to how well they will do on the PK in the playoffs.

What’s Goin’ On, Man?

Let’s see if we can decompose the situation a bit and get some insight.

First, let me propose to you a simple model of success on the PP and the PK:

  • On the PP, success = shots taken x shooting%.  All else equal, more shots = better PP.  All else equal, better sh% = better PP.
  • On the PK, success = shots given up x save %.  All else equal, fewer shots = better PK.  All else equal, better sv% = better PK.


So let me throw some more charts at you and see which, if any, of those four aspects actually carry over to the playoffs.

figure_1 pp shpct figure_2 - pk svpct figure_3 pp sf60 figure_4 - pk sa60

OK, now we’re getting somewhere.

There remains basically no relationship between how you do in the regular season for PP shooting, PK saving, and PK shot rates against, and how you do those same things in the playoffs.  The first two should not be a surprise – we see that effect everywhere.  Sv% and especially sh% do not sustain year to year.  They vary like crazy.

The shot rates against is a bit of surprise, though, isn’t it?  You’d figure teams that lock things down shot-wise on the regular season PK would be able to it in the post-season.  Doesn’t happen.  Maybe because the competition is so much stiffer and the stakes are higher.

The only sustainable skill?  Shot rates on the powerplay.  If you’re good in the regular season, you’ll likely be good in the playoffs.

What’s It All Mean, Man?

There is a caveat to this, which is that much of this data is pretty small sample stuff.  We’re looking at 16 teams, half of which exited in the first round.  Not a lot of games, so take with a grain of salt.  A better study would take multiple years of data into the mix, but I just wanted a quick look.

Even so, that lack of correlation is pretty extreme.

What I’d say to you is this:

  • If your favourite team is in the playoffs riding tremendous special teams success, they/you probably can’t/shouldn’t count on it continuing.
  • The one thing that does sustain is powerplay shot rates.  So if your team is deadly at generating shots on the PP, that’s likely to continue.  At least it gives a team like that a decent shot (ha ha) at having special teams success continue into the playoffs.

Betting on the rest looks like a crap shoot.  Go with the data, not the myths!


One thought on “Predicting special teams performance in the playoffs (Predictive variables extravaganza, part 2)

Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s