2016-2017 Season Predictions

October 3, 2016, Micah Blake McCurdy, @IneffectiveMath

I simulated the 2016-2017 NHL seasons a million times to estimate what is likely to happen. I used a simulation harness linked to my predictive model for single games, Cordelia. Curious readers will find lots of detail in those links, but the short version is that I consider offence and defence, shooting and goaltending results, special teams, penalties drawn and taken, rest, and home ice advantage. Injuries, suspensions, and all off-season moves are included.

The method I used is strongly similar to the one I used last year, with some key improvements; some of the explanation is copied from last year's preview.

Eastern Conference Point Totals

TeamMean pointsStandard Deviation
Penguins 100.08.3
Capitals 98.18.3
Lightning 97.88.4
Canadiens 96.48.4
Panthers 95.78.4
Islanders 95.38.4
Flyers 94.68.4
Red Wings 94.28.4
Bruins 94.08.4
Rangers 92.38.4
Hurricanes 90.48.4
Senators 90.28.4
Maple Leafs 87.88.4
Devils 87.38.4
Blue Jackets87.18.4
Sabres 83.18.4

Each team's bar is centred on the average point total obtained in the simulations for each team, sorted with the highest averages to the right, with the divisions indicated by colour. The changing colour intensities indicate "stanines", that is, each coloured square shows half of one standard deviation.

Thus, for instance, there is a 20% chance that Detroit finishes with between 92.1 and 96.3 points; the two adjacent boxes each represent a chance 17%, the next two 12% each, the next two 7% each, and the extreme pair 4% each. The darknesses of the boxes are chosen to reflect these probabilities.

Thus, we expect around six of the teams to finish in the darkest box, around ten to fall into the adjacent boxes, around seven to fall into the next pair of boxes, around four into the next pair, and one or two teams to fall into the faintest boxes. These one or two teams, which I make no attempt to predict ahead of time, will be much discussed. The proximate cause of their success or failure will doubtless be a superhuman goaltending performance, like Carey Price in 2014-2015, or a horrific cavalcade of injuries, like Columbus in 2014-2015, or an exuberance of last-minute goals and hot shooting, like Calgary in 2014-2015. Part of why I make predictive models is that I enjoy knowing just how unlikely are the various unlikely things that happen every year.

Playoff cutoffs: 93.4 in the Metropolitan, 94.1 in the Atlantic.

Western Conference Point Totals

TeamMean pointsStandard Deviation
Kings 101.08.4
Predators 98.78.3
Sharks 97.78.4
Stars 95.28.4
Wild 94.08.4
Blackhawks93.98.3
Blues 93.58.4
Ducks 93.08.4
Jets 92.98.4
Flames 91.18.4
Oilers 87.98.4
Coyotes 86.28.4
Canucks 85.68.4
Avalanche 84.28.4

The West is tighter than the East, especially in the middle where the churn for the bubble playoffs spots is. The central teams have a decisive upper hand, though the pacific is by and large weak enough to make the Central easier to make the playoffs from than the Metropolitan.

Playoff cutoffs: 92.0 in the Pacific, 93.2 in the Central.

Divisional Breakdowns

Atlantic Division

TeamPlayoff Chance
Lightning 69%
Canadiens 67%
Panthers 64%
Red Wings 57%
Bruins 56%
Senators 38%
Maple Leafs28%
Sabres 13%

Every team's most likely finishing position is marked---all three of Florida, Montreal, and Tampa Bay are more likely to finish first in the Atlantic than in any other particular position, which is another way of seeing just how tight the division is. Buffalo will need to see considerably larger improvements from their young players than is likely if they are to make the playoffs; the same to a lesser degree is true of Toronto.

Metropolitan Division

TeamPlayoff Chance
Penguins 80%
Capitals 74%
Islanders 62%
Flyers 58%
Rangers 47%
Hurricanes 38%
Devils 25%
Blue Jackets24%

The Metropolitan division is a little more stratified, without so much bunching of teams. The Taylor Hall trade is a big improvement for New Jersey but they need a lot more pieces. The two middle teams (NYR and Philadelphia) rely more than the others on their goaltending, so there may yet be confusion there. Pittsburgh look considerably stronger than most defending cup champions, they've lost no substantial pieces.

Central Division

TeamPlayoff Chance
Predators 82%
Stars 70%
Wild 65%
Blackhawks64%
Blues 62%
Jets 59%
Avalanche 21%

The Central is by far the tightest division in the league, with very little to choose between among the middle five teams. Colorado are the clear runt of the litter, not least because they have to play the other Central teams so much. Nashville is considerably improved by the acquisition of Subban, which moves them to favourites for the division.

Pacific Division

TeamPlayoff Chance
Kings 89%
Sharks 80%
Ducks 61%
Flames 53%
Oilers 37%
Coyotes 30%
Canucks 27%

Even with clear recent improvement, the Oilers still have lots of weaknesses; Vancouver has made a number of very poor moves recently and is almost certain to be very weak; and Arizona's playoff hopes rest on several of their young players showing strengths they have not previously shown signs of. The traditional powerhouses of Los Angeles and San Jose look every bit as strong as last year.

Stat Breakdowns

Based on their likely rosters throughout the season, the expected attributes for each team are as shown to the left. Some teams are all-around power-houses, like Los Angeles, Pittsburgh, Nashville, or San Jose, where some have many weaknesses, like Buffalo or Colorado.

Some very good teams are unusually reliant on goal-tending, most obviously Montreal and Philadelphia, both of whom would be middle of the pack without their excellent goaltending. Such high rankings are not undeserved---after all, wins from great goaltending are still wins---but they are more fragile and prone to variation, both that typical to the position and injuries, than rankings founded in other stats. Conversely, a few teams have unusually weak goaltending, such as Carolina and Dallas; they also could change considerably in strength.

Rest

Not all teams are equally affected by rest. The strongest effects from rest are seen when teams play after playing the night before. The table below shows how many times each team plays 'tired' in this sense, as well as how many times they play against a team that is tired.

Wooden Spoon

Since the league hasn't yet implemented Gold drafting, many teams will once again play many games at the end of the year which they would rather lose. Although there will be many trades and confusions between now and the end of the year, these are the early probabilities (including all three lotteries) for who will get the 2017 first overall draft pick.

Team(s)Chance of First overall pick
L.A, NSH, PIT, S.J, WSH1%
BOS, CHI, DAL, DET, FLA, MIN, MTL, NYI, PHI, STL, T.B2%
ANA, CAR, CGY, NYR, WPG3%
OTT4%
EDM, TOR5%
ARI, CBJ, N.J6%
VAN7%
COL8%
BUF9%

Weaknesses

By far the largest weakness in my approach that I can discern is that no attempt is made to account for coaching effects. For single players who change teams I feel that this effect is reasonably small but for teams who change coaches the effects can be substantial. Some teams (Colorado and Ottawa, noticeably) have substantially improved their coaching staffs since any NHL games were played; those teams will probably do better than listed here although I am at a loss to estimate just how much better. Some other teams (notably Anaheim) have substantially weakened their coaching staffs, they will probably do somewhat worse than projected here. Other teams have changed their coaching staffs non-trivially also though I am not sure if for the better or for the worse.

Another (related) weakness is the failure to explicitly account for any context to the individual elements that go into game predictions except for score-state. My previous work has suggested that zone deployment is largely negligible; competition deployment may or may not be negligible; and teammate deployment is certainly not negligible: none of them are explicitly included. When teammates remain together the latter fault is small (it is implicitly present in the data and preserved) but when players change teams it is larger. Thus my approach will only partially account for players who change teams. I don't see a way to tackle this without modelling the way in which two (or three, or four ...) players will affect one another's on-ice results; at the moment this simply seems like a gargantuan undertaking.

Finally, player aging is not explicitly included, so players under the age of twenty-four or so will be slightly underrated by Cordelia and players over the age of twenty-six or so will be slightly overrated. I consider only results from the past two years in part to try to detect aging effects, but teams with lots of young players will likely overperform their estimates here and teams with heavy icetime to older players vice versa.

Post-Season Update

Now that the dust has settled, how did I do?

TeamMean points (expected)Actual pointsError (in standard deviations)
Penguins 100.01111.3
Capitals 98.11182.4
Lightning 97.8 940.5
Canadiens 96.41030.8
Panthers 95.7 811.8
Islanders 95.3 940.2
Flyers 94.6 880.8
Red Wings 94.2 791.8
Bruins 94.0 950.1
Rangers 92.31021.2
Hurricanes 90.4 870.4
Senators 90.2 980.9
Maple Leafs 87.8 950.9
Devils 87.3 702.1
Blue Jackets 87.11082.5
Sabres 83.1 780.6

  • Hits: Lightning, Islanders, Bruins, Hurricanes, Sabres, all off by no more than 0.6 standard deviations (that is, about four or five points off).
  • Misses
    • Pittsburgh, Rangers, and Washington, probably since their shooting talent was undervalued;
    • Ottawa, likely wrong because of not accounting for coaching improvement;
    • Toronto, likely wrong by underestimating youth.
    • Detroit, likely wrong by overestimating age.
    • Flyers and Canadiens, for no reasons that I can perceive.
  • Big Misses
    • Columbus, badly wrong, probably due to underestimating young players and failing to predict a Vezina-calibre season from Sergei Bobrovsky.
    • Panthers, mostly due to severe injuries.
    • New Jersey, at least in part due to an unusually weak goaltending performance.

TeamMean points (expected)Actual pointsError (in standard deviations)
Kings 101.0 861.8
Predators 98.7 940.6
Sharks 97.7 990.2
Stars 95.2 792.2
Wild 94.01061.4
Blackhawks 93.91091.8
Blues 93.5 990.7
Ducks 93.01051.4
Jets 92.9 870.7
Flames 91.1 940.3
Oilers 87.91031.8
Coyotes 86.2 702.0
Canucks 85.6 692.0
Avalanche 84.2 484.4

  • Hits: Flames, Predators, and Sharks, off by only a handful of points.
  • Misses
    • Kings, likely due to overestimating old players, especially as their shooting talent falls off;
    • Stars, perhaps due to unexpectedly bad goaltending performances.
    • Minnesota, likely due to coaching effects.
    • Chicago, due to unexpectedly strong goaltending performance.
    • Jets, due to unexpectedly weak goaltending performances.
    • Oilers, likely due to underestimating young players and also unexpectedly strong goaltending.
    • Blues and Ducks, for reasons I can't see.
  • Big Misses
    • Colorado, badly wrong, due to failing to account for aging, coaching, and unusually weak goaltending.
    • Arizona and Vancouver, and I don't know why, although weak performances early (and subsequent decisions near the trade deadline) exaggerates the discrepancy.


© Micah Blake McCurdy 2017, supported via patreon