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.

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?

• 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.

• 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.