How repeatable is performance in the Offensive, Defensive and Neutral Zones?

A few years back, Eric Tulsky (and others at Broad Street Hockey) pioneered the start of neutral zone tracking, or rather the tracking by individuals of every entry each team makes from the neutral zone into the offensive zone during a hockey game.  The idea of this tracking was simple:  Neutral Zone play is obviously important to winning a hockey game, but NHL-tracked statistics contain practically no way to measure neutral zone success overall.   Zone Entry tracking remedied that, by giving us both individual and on-ice measures of neutral zone performance.

An overall measure of neutral zone performance that we can find with zone entry tracking is called “Neutral Zone Fenwick.”  By using the average amount of Fenwick events resulting from each type of zone entry (Carry-in or Dump-in), we can create an estimate of what we’d expect a player’s Fenwick % to be with them on the ice based on the team’s neutral zone play with them on the ice.  In essence, this is a measure of a player’s neutral zone performance, helpfully done in a format that we’re already pretty familiar with – like normal Fenwick%, 50%=break even, above 50% = good, below = bad.

In addition, by being able to quantify overall neutral zone performance, we can then quantify overall offensive zone performance with similar metrics.

OZ100 for example, takes a look at the amount of shots by a player’s team with them on the ice and compares that # to the amount of shots we’d expect from the neutral zone play.  An OZ100 of 100% = the player did exactly as we’d expect in the offensive zone from his neutral zone play – in other words, his offensive zone play was basically average.  An OZ above 100 means that the player put up more shots than we’d have expected, so their offensive zone performance was positive.  Each # above 100 represents a % better than we’d have expected – so an OZ100 of 105 means the player’s offensive zone performance was 5% better than expected, while a 95 means it was 5% WORSE.

DZ100 works the same way.  A DZ100 of 105 means that there were 5% LESS shots than we’d have expected opponents to take based upon the NZ entries opponents’ made while 95 means there were 5% MORE.  (In the future this metric won’t be reversed like this, but it suffices for now)

When Eric wrote his initial post on NHL Numbers, he noticed the following about his data, based on basically these metrics, coming from the 2011-2012 Flyers:

 The net result here: our surprising results in the offensive and defensive zones appear to be based on a not-particularly-reproducible metric. It may still be true that some players have skills that help the team get more shots per zone entry, but at the end of a year we can’t reliably tell which players those are — we know who did well in the offensive zone this year, but don’t have strong reason to believe they’ll do well in the offensive zone again next year.

The neutral zone is a different story. The split-half reliability there is 0.44, high enough that we can be 97% sure that this is a real correlation and not just random results. Given a half-season of neutral zone data, we can make a decent guess at what will happen in the other half-season.

Moreover, the neutral zone results look like they are not just statistically significant, but meaningful in practice as well. It’s pretty reasonable to see Jagr and Claude Giroux leading the forwards and Timonen and Matt Carle leading the defensemen.

Last year, however, a few more teams had people tracking their own neutral zone play.  And when I looked at the split-half reliability of 3 of those teams – the Isles, Oilers, and Hurricanes – I didn’t quite find the same thing as Eric did.  In fact, for the Isles, I found that OZone and DZone performances were MORE repeatable than Neutral Zone Play!  For the Oilers, Neutral Zone Performance was the most reliable, but OZone wasn’t that far behind.  And for the Canes, none of Neutral, Offensive, or Defensive zone performance was very repeatable at all.

Of course last season was a shortened season, and so we had much less data than Eric did in 2011-2012.  So how about this season?  Well again, I have 3 teams’ worth of data to look at, and the results don’t suggest that Eric’s initial hypothesis is correct, and in fact the results are a bit all over the map!

Lets start with the Flyers, the same team Eric’s results were about 2 years ago.

Below are graphs for the Flyers’ play in each of the three zones this past season.  Each graph shows individual Flyers’ performance in Odd games on the x-axis and on Even games on the y-axis.  If a metric – in this case NZ Fenwick (measuring neutral zone performance), OZ100 (measuring Offensive zone performance) and DZ100 (measuring defensive zone performance) – is reliable over a sample (in this case, over 41 games), you’ll see a decent correlation between performance in odd and even games.

Here’s what we see for the Flyers:

Flyers Correlations 1 100s

This is pretty much what we’d expect to see from Eric’s older data – there’s a strong Neutral Zone split-half correlation – far stronger than even Eric saw in the older data, in fact.  By contrast there’s basically no offensive zone repeatability and there’s very little if any defensive zone repeatability.

So the Flyers check out with Eric’ theory.  How about the Isles?

Isles Correlations 100

Here we start to see some things that don’t fit in Eric’s theory.  Again, Neutral Zone performance is the most repeatable of the three zone skills, although the R^2 is now a little weaker.  And again, Offensive Zone repeatability is very low.  But here, defensive zone performance appears nearly as repeatable as neutral zone performance.  And this is the second year in a row that defensive zone performance was repeatable at this level, unlike with the Flyers.  Of course, offensive zone performance was repeatable last year for the Isles.

Okay, how about a third team?  Here are the Sharks.

Sharks X

Here we don’t see reasonable correlations for play in ANY of the three zones – although there is a very small correlation for the neutral zone.  You don’t see anything like the other two teams in the Sharks.

More interestingly, the Sharks as a team showcased a performance – where they put up a 3rd in the league 54.6% Fenwick close – which was pretty much driven entirely by offensive and defensive zone play.  The Sharks were a NEGATIVE neutral zone team this year – 49.4% NZ Fenwick.  Instead of winning the neutral zone, the Sharks succeeded by massively outperforming against opponents on entries based off of carry-ins and dump-ins, something which Eric’s initial Flyers work suggested wasn’t possible to sustain.  And while individual Sharks didn’t seem to maintain their positions between odd and even games, as a team, the Sharks continued to be a superior team in the offensive and defensive zones in both odd and even games, despite not being a positive neutral zone team.  And given that the Sharks have been a positive possession team for a few years, this doesn’t seem likely to be a fluke.

Conclusion:

Eric’s initial thesis as to the importance of neutral zone play and the lack of repeatability in offensive and defensive performance certainly does fit the Flyers.  But it doesn’t fit anywhere near as well other teams that have been tracked.  This suggests to me there’s a heavy coaching influence going on here (although the Flyers did shift coaches this year).  Neutral Zone Play is certainly incredibly important.  But teams have seemed to succeed via offensive and defensive zone success WITHOUT neutral zone success, which suggests it may not be AS important as Eric’s initial work suggested.

 

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