The fancystats mystery that was Nikita Nikitin

Ask an Oiler fan who the ‘problem’ defensemen last year were, and you’re pretty much guaranteed to hear something “NSF”.  As in: the unlamented Nikitin, the often despised Schultz, and/or the (still) much lamented Ference.

I expounded on the differences between these three in a previous article.

There is no mystery with Ference or Schultz.

Justin Schultz’s deficiencies on defense are widely acknowledged (even by him). But to some extent, he’s gotten away with it because of what he contributes at the other end of the ice. Most Oiler fans can’t stand the ‘lollygagging’, but it’s no coincidence that all three of Schultz’s coaches have made him the defensive ice time leader on the Oilers.  And there is promise, real promise, this year.

Ference is at the other end of the scale. A warrior, a champion, a fitness fanatic, a family man, and a great community guy. Also, not an NHL defenseman anymore. All of these things are true.

Now, the thing is – the fancystats do help make the case for both players. Schultz’s offensive game and Ference’s terrible overall game, they both show in the numbers.

But Nikita Nikitin, now he’s a conundrum wrapped in an enigma. Can’t catch the mystery, can’t catch the drift.

Why a conundrum? Because the eyes (often) say he’s terrible, and the fancystats say he’s (sometimes) OK.

Is Nikita Nikitin preventing a rebound or hiding his face in shame?

For example, here’s a common fancystat view of the three defensemen (all numbers 2014-15 5v5, play by play data sourced from war-on-ice.com, with statistics generated using my own scripts):

CF% DFF% CA60 DFA60
Andrew Ference 45.0% 43.1% 59.7 47.6
Nikita Nikitin 48.9% 48.8% 55.4 39.5
Justin Schultz 50.2% 49.2% 55.5 43.8

* CF% = Corsi, or shot attempts for/against, CA60 = shot attempts against per hour.

Ference’s poor showing is clear – the puck is moving the wrong way (45%) when he’s on the ice, and he’s the worst of the three, actually of ALL regular Oiler defensemen, for the team giving up shot attempts (CA60) when he’s on the ice.

The other two measures you see above are my own, where I’ve weighted every unblocked shot attempt by shot type (e.g. slap shots being more dangerous than backhands) and distance. So the D stands for “danger”, as in Danger-Adjusted Fenwick For, or DFF, and Danger Fenwick Against, or DFA.  As with other shot metrics, the % is for the overall balance, and the 60 represents a rate. (I’ll have a full article explaining this metric, and how the Oiler defense looks through its lens, at some point soon)

And here we see the first mystery.

You can see that Nikitin is head and shoulders above the other two when it comes to preventing dangerous shots (DFA60, lower is better).  My DFA metric has him best at that on last years team.

And it’s not just me … war-on-ice also ranks him highly on preventing high-danger chances, second only to Jeff Petry last season.

Other fancystats views of Nikitin are also quite complimentary. For example, OwnThePuck’s HERO chart shows him mostly as a capable second pairing defenseman:

Nikitin HERO chart 1

And Vollman’s player usage chart suggests that he’s also middle of the pack, not good, but not as horrible as the eyes suggest:

Nikitin vollman

Part of what explains Nikitin’s OK (for the Oilers) possession numbers is his passing ability.  To quote WoodGuy on the topic:

To my recollection and bias when Nikitin gets the puck in the dzone, the next player to have the puck is an Oiler via pass.

When Ference gets the puck in the dzone the next player to have the puck is the opposition via poor pass or “off the glass and out” where the opposition re-load for another sortie.

To really confirm that’s true, we need a different set of fancystats: zone exits, or passing data.  These are being crowdsourced now, but not widely available as yet.

But even the ‘moving the puck’ explanation doesn’t explain why Nikitin’s ‘dangerous chances against’ are lower than the eye suggests they ought to be.

So what’s up with ? How can there be this much of a mismatch?

I figured I might dig deeper into the numbers and see if there’s anything that pops out that might explain the wheres and whys of the dangerous chances. In particular, I looked at the following:

  • what percentage of shot attempts were blocked with NSF on the ice
  • how many unblocked rush shots were given up with NSF on the ice
  • how many unblocked rebounds were given up with NSF on the ice

(The definition of rush and rebound are both from war-on-ice)

Here’s how those numbers look (again, play by play data from war-on-ice but numbers generated by my own scripts, so if there are any errors, they are mine!):

CA FA RushF RebF Block% RushF% RebF%
Andrew Ference 1159 901 553 164 22.3% 61.4% 18.2%
Nikita Nikitin 601 433 284 73 28.0% 65.6% 16.9%
Justin Schultz 1373 1020 725 195 25.7% 71.1% 19.1%

Two numbers stand out a bit – a noticeably larger % of shot attempts are blocked when Nikitin is on the ice. And a smaller percentage of the total shot attempts are rebounds – rebounds representing an exceedingly dangerous shot situation.

Ference is oddly good at reducing rush shots against, but what that means I’m not exactly sure.  Schultz’s number is no surprise – one of the knocks against him is that he pinches too aggressively, setting the team up for rush chances against.  Presumably Ference, who doesn’t jump into the play, ends up not taking that risk.  Nikitin is about average in that department.

Since Nikitin is also about average for shot attempts given up while on the ice, the numbers above do mean that a lower than average number of shot attempts and rebound shot attempts are being direct towards the Oiler net when he’s around.

Ah hah!  So have I cracked the mystery?

Um, no. No, I have not.

Maybe reduced it a bit. Certainly, the ability to clear the slot is one of those things that can go unnoticed in the absence of statistics. We tend to notice the big plays and the big mistakes by eye, but we have the hardest time noticing bad things prevented – whether it’s zone entries denied or potential rebounds cleared.

It may indeed be the case that Nikitin is more efficient at making some of those ‘quiet’ plays. He may be glacially slow, but it doesn’t take speed to clear the slot, and so maybe the play gets smothered more often when he’s around.  But because of those slow feet and susceptibility to the occasional brain cramp, when Nikitin gets beaten, it’s glaringly obvious. We remember the latter but never even notice the first. And that’s what shows in the stats but not by eye.

We’ve not “solved” the mystery of the eye/stats mismatch by any means. Some deeper digging is needed – for example, the rebound% may be the best on the team, but is it a meaningful difference?  And rebound% may not be a repeatable skill at all, and the differences we see across the Oiler defenders are just random variation. Both are certainly possible.

But conversely, maybe it has highlighted a less obvious (good) part of Nikitin’s game. It doesn’t mean Nikitin is good, but it does suggest there might be a reason he’s better than Ference, with whom he’s often lumped.

Uncracked mystery aside … if Nikita Nikitin is ever again an Oiler this year, I will certainly be paying a little bit more attention to his rebound control.

And if he’s never again an Oiler?

These numbers, while suggesting there may be a reason that Nikitin is actually better than the eye suggests, still don’t make me feel like crying in my soup now he’s gone to B’field.  Maybe the fancystats aren’t telling us anything.  Maybe in this case, they’re just plain wrong.


Happy Canadian Thanksgiving!


Advertisements

2 thoughts on “The fancystats mystery that was Nikita Nikitin

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