Advertisement

Peter Chiarelli, Kris Russell, and the value of 'micro-stats' (Trending Topics)

613445688
613445688

The hockey analytics community is well-known for its general revulsion toward Kris Russell, and for good reason.

This is a group of people who worked for years to prove the positive link between something as simple as raw shot attempt numbers at 5-on-5 to goals for and against, which ultimately correlates strongly to winning.

Russell, meanwhile, has generally been seen as shot-attempt poison: Put him in your system and you lose, because he gets outshot very, very badly. As such, people like me strongly advised teams against signing Russell in general — and certainly at the rumored asking price in the $30 million range seen prior to the free agency period — because giving him any kind of big-money or prominent role would likely end in tears.

[Sign up to play Yahoo Fantasy Hockey for free | Mock Draft | Latest news]

So the league waited literally an entire offseason as Russell went unsigned. We got occasional reminders that he was out there, and that a handful of teams generally wanted him but needed some other money stuff to shake out first. These assurances were the limit of what we knew about Russell, whose largely unsuccessful time in Dallas probably dinged the hell out of what he might have otherwise reasonably commanded.

But they turned out to be true: Once the Oilers offloaded Nail Yakupov to free up some wiggle room, they immediately spent that money signing Russell to a one-year deal at $3.1 million against the cap. In terms of money, he became the Oilers’ fifth-highest paid defenseman, and if Russell is your No. 5 defenseman, well, you could do worse. The price point isn’t ideal but the real issue his teams have generally run into is that they use him as more of a middle-pairing offense-driver, and sometimes more than that. It’s a role he’s ill-suited for. But if you’re keeping his minutes tightly managed and you’re using him as, say, a power play specialist, there might be some value there. You just have to go into it with the understanding that the things he does against even middling competition lead to his team generally being outshot. Which leads to more goals. Which leads to more losses.

Now, the issue for the Oilers is that they have enough defensemen already who need their minutes managed to be effective, and this is an additional cook after having about four in the kitchen already. At some point, Todd McLellan isn’t going to be able to do the geometry, and it’s going to lead to some risky decisions he’s basically being forced to make.

With all this in mind, it’s important to look at things from GM Peter Chiarelli’s perspective. If people can generally agree that Russell isn’t that good — and most people in Alberta, whether they’ve seen him several times a year just from watching Flames/Oilers games, or have a longer view of him, would probably be included here — then why bring him aboard? Here’s Chiarelli on Oilers Now talking about just that very subject:

“Relying on analytics for d-men is tough. We looked at all that stuff. I saw all that stuff on him. But we need someone who can retrieve pucks and can transport the puck through the neutral zone. Just a side note, analytics in regards to Kris. He was second in the league as far as clean entries into the offensive zone and the neutral zone, whether it’s a pass or skating. Second in the league. That’s not in conventional analytics that are just kind of derivatives of Corsi. That’s on our own program that we have, and there’s a few other teams that have it. Second in the league.”

The idea here, then, is that Russell fills a need: The Oilers struggle to get the puck out of their own zone (not sure Russell helps with that), through the neutral zone, and into the attacking zone. Russell allegedly does this better than any current options Chiarelli had on hand. Fair enough. If you see a hole and you feel like you can fill it, why wouldn’t you?

Of course, this is to some extent what someone who acquired a number of corsi-deficient defensemen in recent months would say about using shot-based metrics to measure defensive effectiveness. But that consideration notwithstanding, this is a general problem the hockey world as a whole should have about dealing with the micro-stats that come from companies like SportLogiq, or some of the stuff you saw during the World Cup.

The question you should ask is, “So what?”

Kris Russell was second in the league in clean zone entries. How is that being measured? What does “clean” mean here? Over how many games? Against whom? And more to the point: How does clean entries translate to goalscoring, which translates to winning?

It should be noted that Russell more often than not outperforms his teammates in terms of goals-for percentage, but a lot of very smart people have done enough research that they feel confident in saying that this may still be something of a fluke. Russell’s PDO fluctuates wildly from one season to the next, with no real rhyme or reason to it. Sometimes his on-ice shooting percentage is through the roof. Other times his save percentage is much higher than the league average. In general, he doesn’t seem to be able to really “drive” one or the other.

If his ability to create clean zone entries led to, say, more shots on goal, wouldn’t that show up in his possession numbers? And if it led to demonstrable shifts in shot quality either for or against, wouldn’t that show up in his relative shooting or save percentages year after year? The past two seasons, Russell’s on-ice shooting percentage has been significantly higher than that of his teammates. But the year before that, it was only slightly higher, and the year before that it was somewhat lower. Likewise, his on-ice save percentage last year was lower after being marginally higher for two years, which followed a season in which it was substantially higher. Again, if it were under his control, either directly or indirectly, you’d expect reliable differences between what he does and what his teammates do.

And by the way, it’s tough to say the Oilers should be even taking the data that shows Russell as being a top defender at zone entries at face value. A number of independent writers who tracked zone entries for the Flames on a regular basis in the past few years not only show Russell as not being close to the top of the league, but not even being best at it among the Flames. Some it is here, here, and here. Plus, here’s Corey Sznajder, who has worked for an NHL team in the past and tracks pretty much everything he can across the entire league, throwing in his two cents on the Russell zone entry stats Chiarelli cited:

https://twitter.com/ShutdownLine/status/786240328848244736

That raises questions about not only the efficacy of the data cited, but also how accurate it is to begin with. And if NHL teams are making decisions based on data that might not be that accurate (disclaimer: it’s also possible all these other people might just be tracking zone entries differently), that’s a big worry. In much the same way SAP was criticized for selling the NHL “snake oil” on advanced stats, who’s to say that there aren’t other companies out there just flat-out getting this kind of thing wrong, and NHL teams buying it anyway?

Is Russell a decent bet as a bottom-pairing D? Sure. But given how Chiarelli is talking about him, and how he was used in their season opener against Calgary (alongside Andrej Sekera on the second pairing), that won’t be his role. You can therefore bet a reasonable amount of money on the Oilers getting outshot with Russell on the ice once again. Make all the clean zone entries in the world, and it doesn’t matter, because Russell is still spending more time in his own end than the opponent’s.

And if you can’t draw a straight line between clean zone entries and outshooting your opponent, well, why don’t I just give Chiarelli himself the last word?

“So I understand the analytics group, and you can spin, you can derive analytics any way you want at the end of the day. There’s been a lot of moves that are based on analytics that don’t work out.”

No kidding.

Ryan Lambert is a Puck Daddy columnist. His email is here and his Twitter is here.

All stats via Corsica unless otherwise stated.