Safe is Death has come to Vancouver

By his own admission, and in stark contrast with his predecessor, Vancouver Canucks head coach John Tortorella has an old school view of how to approach coaching an NHL team. The firey ‘Torts’ doesn’t rigorously manipulate matchups, he’s often been heard to lambast the common habit of "over-coaching," and he even derides the notion of playing a "system," preferring to characterize his gameplan as a "mindset". Tortorella has also talked at length about not really understanding analytics.

Yet for all of that luddite bluster I’m struck by the extent to which the Canucks have, in the first 30 games of the John Tortorella era, played a style of hockey that adheres closely to some of the conventional wisdom drawn directly from shot based metrics. 

Read past the jump for more.

Before we wade much further into the muck, I’d like to share share this graph that Senators writer Travis Yost created. Travis’ graph demonstrates the change NHL teams have made between this season and last season measured in "Score Close Team Fenwick%." For the uninitiated (welcome, we’re not that smug I promise!) Team Fenwick% is an expanded version of +/- that counts up all goals, shots and missed shots both for and against, and expresses the ratio as a percentage. Also worth defining: a "score close game state" includes all situations in which a game is within a goal in the first 40 minutes or tied in the final 20.

A quick note on the score close game state. Because teams change the way they play when they’ve got a lead (teams go into a defensive shell, dump the puck in and change constantly, etc.), there’s a sensible preference among hockey analysts working with shot based metrics to slice up the data by game state so as to correct for the "score effects." Basically we’re trying to figure out how a team plays when they and their opponent are both going all out to establish a lead.

To use a cross sports analogy: a running back getting 15 rushing yards in garbage time against a prevent defense is neat, but if you want to separate the Marshawn Lynch’s of the world from the Jamaal Charles’ – look at their yards per carry in short-yardage situations when they’re running headlong into defenses with 7 or 8 guys in the box…

Anyway, and please excuse my digression, I captioned Travis’ graph on Twitter and it’s worth looking at here, I think:

Under John Tortorella (ella, ella) the Canucks have, through 30 games, been an improved puck possession team. Not all of that is necessarily coaching – it helps to have a healthy Ryan Kesler driving play again on a secondary line, for example – but it’s certainly good to see. Not so good to see? As much as the Canucks have improved their puck possession game in score close situations at five-on-five this season, six other Western Conference teams have improved more…

Vancouver’s marginal improvement at controlling the flow of play at even-strength with the result in doubt is cool and all, but the more dramatic change ushered in by Tortorella – and the change more likely to be a function of coaching rather than personnel or noise – is the club’s play when leading. 

You almost certainly know this by the eye-test already, but "Safe is Death" has arrived in Vancouver. And goodness is it a relief.

I put together a table to help us look at the quantifiable extent to which the Canucks have begun to play more aggressively with the lead under Tortorella (data from Obviously there’s a significant sample size issue here as the 2013-14 portion of the table only accounts for 30 games. But 30 games gives us enough data to point out a suggestive trend in my view, even if we be very reluctant to draw any very firm conclusions for the moment…

Season Team Fenwick% Tied Fenwick% Close Fenwick% Up 1 Fenwick% Up 2 Fenwick% Down 1
2009-10 49.5% 49.3% 46.2% 42.3% 55.6%
2010-11 53.5% 53.4% 49.1% 43.9% 60.5%
2011-12 52.7% 51.4% 47.3% 45.3% 56.4%
2013 52% 51.7% 45.4% 43.8% 59%
2013-14 51.1% 53.5% 49.5% 51.1% 59.5%

As you can see, the way the Canucks are playing when they’re up this season represents a pretty stark change from years past (so far). When up by one goal the Tortorell Canucks are controlling the flow of play at a top-six rate, and they’re also one of only  five teams to be above 50% of Fenwick events when up by two. The other four are San Jose, St. Louis, Chicago and Los Angeles; who make for decent company. I’m wary of suggesting that this will continue particularly since Vancouver hasn’t led all that often this season, but these numbers are still worth spotlighting in my view.

Let’s dial it back momentarily because there are two facts we should note before heaping scorn on Vigneault’s conservatism. The first is that with Vancouver there was nothing unique about the way the Canucks would sit back with the league, in a "macro-sense" as Cam Charron once put it: all coaches seem to employ a similar strategy.

The other thing we should note is that Vigneault’s Rangers are among the short list of teams posting a higher Fenwick% when up a goal than the Canucks are this season. They’re down from their "Fenwick% Up 1" number during the final season of Tortorella’s tenure, but "Safe is Death" appears to be alive and well in New York in the early going this season too…

Beyond my desire to quantify an observable difference in this season’s version of the Vancouver Canucks, I find this point interesting because it speaks to an "analytic bent" in Tortorella’s coaching and overall outlook. It’s human nature to bunker and play conservative hockey when you’re leading, but the smarter play (and it has been compellingly demonstrated with data) is for teams to continue to try and outscore their opponents even in possession of a narrow lead. The prevent defense only prevents winning, after all.

Making that sort of counter-intuitive leap requires confidence and some intellectual backbone. Even if Tortorella and the Canucks coaching staff aren’t looking over spreadsheets to figure this stuff out, it’s worth noting that this team is behaving the way we’d expect a team more openly informed by the data to.

  • JCDavies

    I’m blown away by how similar the underlying numbers are this year to 2010-2011. Canucks had a better powerplay, better shooting percentage, and better goaltending that year, but still, that’s a very positive sign.

    • JCDavies

      The FenTied% has been declining ever since 2010-2011.

      What the Canucks are doing this season (getting down by a goal so often and then throwing everything at the net) is not even remotely close to the SCF team or the 2012 Kings…

      • As I said, there are lots of factors differentiating this team from that one, but the underlying numbers are generally positive and somewhat surprising, except for the one you selectively chose.

        Also, go back and refresh your memory about the Kings from the first half of the 2011-2012 season. They spent a lot of time down a goal, and were merely a good possession team until they acquired Sutter and Carter, after which they became a beast.

  • I definitely agree that with @NM00 that VANs tied performance is a cause for concern in the early going. But for now I’d suggest that score close remains a modestly more reliable forecaster (just bc of sampling issues). This will be worth revisiting in another 10-15 games.

  • Brent

    So who do we trade and who do we acquire to increase the shooting percentage and hopefully also increase the fentied percentage. A lot of people are talking about the Canucks needing a trade to get a goal scorer, but there are a lot of no-trade clauses in the lineup. Hopefully we don’t loose some of the potential youth.

      • You’d also deal the Sedins and Kesler for a rock so, yeah. You’re far from an intelligent voice or someone with any knowledge in this area. This has been confirmed over and over by many.

        Anyway, moving on from Douchebag00. We can’t make many significant moves for this year. If we miss the playoffs it could cost Gillis his job. It may also encourage some players to partially waive their no trades. Edler is still a commodity we can deal and get something in return. Burr’s contract is way too high to deal him. I’d also look at dealing Hamhuis or Bieksa in the off-season (depending on the returns). A lot of big IFs. I don’t think there will be many moves for a long time but who knows.

  • @NM00 I get that, but they’ve still spent almost twice as much time in score close situations as score tied. Score effects matter here, but so does sampling.

    Point is the #s and splits are extreme, odd, and infantile enough that I’m going to give them 10-15 more games to shake out before putting the most negative possible spin on it.

  • We’re actually 4th in the league in fenwick for at tied, but 26th in fenwick against. Is this something we see in Torterella teams or is it just growing pains as the team learns a new system?

    Looking at individual numbers (even smaller sample) the problem seems to be at the top of the lineup. Sedins and Hamhuis have been subpar at tied (6-12% down on last year). Can’t help that our 3 best defensive wingers have played the same number of minutes at tied was our 3 worst d forwards (Hansen, Booth, Burrows; Welsh, Richardson, Sestito).

    Basically, we don’t have the forward depth to support ANY injury in the top 9 (even without injuries, we’ve only got 8 top 9 forwards). One injury and we’re a two line team. We really shouldn’t be expecting our fen tied % to improve with Burrows and Edler out.

    People seem to forget that part of the magic of the Kings was an unprecedented run of health.