2016-17 NHL standings predictions using analytics

Watch as the Montreal Canadiens and their fans welcome each and every player in their home opener on Tuesday.

Last season we attempted to project the season’s final standings by averaging the ratings of three competing methods of assessing team performance – WAR-on-Ice.com’s Wins Above Replacement, Tom Awad of Hockey Prospectus’ GVT-based team projections known as VUKOTA, and the Weighted Shot Attempt metric xGF%. The logic underpinning the idea of doing it this way was that a flaw in any of the three methodologies would be countered by the other two metrics.

So how did the projections for the 2015-16 season perform? The combined model was accurate for 11 of 30 teams in the NHL that fell within their projected point range, and another four teams ended up within two points of the range projected.


Eight clubs qualified as massive misses, where all three models diverged wildly from the final results, with point totals deviating by 10 or more from what was projected. Washington, Florida, Arizona and Buffalo were surprises on the positive side of the ledger while Columbus, Winnipeg, Tampa Bay and San Jose failed to meet expectations.

In hindsight the largest failing of the projections was that the variance in the three projections was applied too narrowly to the standings points. The spread in standings points predicted by the models should have been twice what was quoted, and applying this modification would have accurately captured 18 of the 30 NHL clubs (60%) with another two falling within two points of their projected range. The model deviation ranges should theoretically capture 68 per cent of outcomes, so this would be more in line with expected.

War-On-Ice.com is gone now that its founders Andrew Thomas, Alexandra Mandrycky (both to the Wild) and Sam Ventura (Penguins) have been hired by NHL clubs, so to replace its publicly available GAR data I am using a similar approach they outlined to a variety of player metrics using data obtained from Emmanuel Perry’s Corsica.Hockey. The xGF% data quoted also originates from Perry’s website, while VUKOTA rankings are provided courtesy Hockey Prospectus.

Applying the aforementioned adjustment to this year’s projections we obtain the following for this year’s standings in the Eastern Conference:


Washington has the highest average ranking in the NHL between the three models, which should come as little surprise after they won the Presidents’ Trophy last season. Interestingly, the main reason they outperformed all three model projections last year were the breakout performances of Andre Burakovsky and Evgeny Kuznetsov. Any team that has a player who more than doubles their production from the season prior, as Kuznetsov did in going from 37 to 77 points, is going to surprise some people.

A few Eastern Conference teams to keep a close eye on this year are Boston, Toronto, the NY Islanders and Detroit.

The Bruins got a massive swing in their WAR rankings thanks to the exchange of Dennis Seidenberg for John-Michael Liles on their blue line and Anton Khudobin replacing Jonas Gustavsson as their backup, while also bolstering their forward group with David Backes. If Liles can remain healthy while Chara staves off his decline in performance, a bounce back season from Tuukka Rask could push the Bruins back into the conversation for elite teams in the NHL.

Toronto’s melding of Mike Babcock’s coaching with an analytically driven push towards youthful elite skill is the aspect of the projection that is relatively easy to forecast. Where things get challenging to prognosticate is the potential improvement on the defensive end. Blue line upgrades from rookie and KHL all-star Nikita Zaitsev, trade deadline acquisition Connor Carrick – who led the AHL in post-season scoring last year – and yes, a top-four season from Martin Marincin, could push this team towards competing for a playoff spot. Frederik Andersen and Jhonas Enroth will also need to provide stability in goal.

The Islanders and Red Wings have fallen remarkably in these rankings from a year ago. That’s to be expected when you see top line talent such as Pavel Datsyuk and Kyle Okposo depart. The Islanders lost a number of top-end players from their forward group and are counting on youth (Ryan Strome and Mathew Barzal) to help fill the void. However you slice it though, seeing 37-year-old Jason Chimera alongside John Tavares on the top line isn’t a recipe for success.

Detroit has attempted to stem the bleeding by poaching Frans Nielsen from the Islanders, but seem intent on shooting themselves in the foot by playing sub-replacement level players in their lineup and losing youthful replacements on the waiver wire (Teemu Pulkkinen). Their 25-season playoff streak appears to be at great risk of ending.

Here are the projected standings for the Western Conference:


Everyone’s favourite whipping boy entering this season, the Vancouver Canucks, fare horribly by all three rankings. Their top five skaters match up well with the top five of any team in the NHL. Chris Tanev is looking to cement himself as a premiere shutdown defender in the NHL and the addition of Loui Eriksson gives a serious shot in the arm to the aging Sedins.

Where things get problematic is on defence. With only three NHL caliber blue liners according to Wins Above Replacement, it would be surprising if the team didn’t hemorrhage scoring chances and goals against. The Canucks front office continues to assert that this is a playoff caliber team, but No. 1 ranked draft prospect Nolan Patrick probably makes more sense as a target if the Canucks know what is best for them.

Los Angeles and Nashville are interesting stories entering the year for similar reasons. Both have issues in goal. L.A. has apparently lost Jonathan Quick to a groin injury that could see him miss months of time, and given that his backup is Jeff Zatkoff, they need a stop-gap. It also doesn’t help that the Kings have the weakest forward depth of any contender in the conference, with their forward WAR actually ranking 12th out of the 14 teams shown.

Nashville would rank far higher here if they could figure out a way to replace Pekka Rinne. The team allows very few scoring chances and has an excellent set of defenders led by newcomer P.K. Subban, but Rinne falls well below expected SV% given the quality of scoring chances he faces. He has essentially performed at replacement level the past three years.

Anaheim looks to be the analytical outlier of this season. Many people argue they are a top contender for the Stanley Cup, but multiple analytical assessments view them as a non-playoff team.

There are a few reasons why I wouldn’t anticipate this turning out similar to Washington a year ago (where the punditry proved correct). First, the team exchanged starter Andersen for Jonathan Bernier, who while a decent NHL goalkeeper, is a significant downgrade in expected performance.

Secondly, they replaced Bruce Boudreau with Randy Carlyle. Carlyle may have changed his stripes since he led Toronto to some of the worst defensive results in NHL history, and he is blessed with a vastly superior set of defenders in Anaheim, but it seems unlikely that he will be able to improve the lack of offensive depth after the team lost Patrick Maroon, David Perron, Jamie McGinn, and Brandon Pirri in the past year. Filling those holes with Jared Boll, Mason Raymond and Antoine Vermette likely won’t get them over the hump.

Now the fun part is watching the rest of the season and seeing where things end up. On with the show!

When submitting content, please abide by our submission guidelines, and avoid posting profanity, personal attacks or harassment. Should you violate our submissions guidelines, we reserve the right to remove your comments and block your account. Sportsnet reserves the right to close a story’s comment section at any time.