Auditing 2024-25 NHL season predictions

Now that the season is over, we can turn back to six months ago when preseason predictions were made. Here’s what we had:

Atlantic Division
Metropolitan Division

Pretty decent here, with two exceptions. Boston completely imploded, the move to fire one of the NHL’s best coaches ended up backfiring spectacularly (gee…) and before the season ended the season ended the Bruins traded their captain Brad Marchand and several other key players like Trent Frederic, Brandon Carlo and Charlie Coyle. That didn’t go well, but who could have seen it coming?

On the flip side, Montreal did well to out-perform preseason expectations. Their goal differential of -20 might be fitting for a last place team but they found a way to win some games when not suffering blowout losses and make the playoffs.

Otherwise, not too bad, all the teams relatively ended up about as projected, for this division at least.

That story isn’t the same in the Penguins’ home division..

Aside from getting the layup that the Hurricanes would be safely in the playoffs, most of the rest of the picks here are all over the map. We liked the Devils too much and the Rangers way, way too much — but like Boston that was another case of negative inputs from the executive/management level trickling down to hinder a team.

We were able to call Washington returning to the playoffs over the other contenders, but not as successfully as they ended up.

The biggest whiff of all was counting Columbus out. Can’t be too mad about being wrong about that since the Blue Jackets provided one of the more inspiring and impressive seasons of anyone in recent memory.

Grading the predictions

As far as margins of error goes, here are some models and betting lines performances this year. To give an idea of accuracy, the average pick is about 11 points off per team in one direction or the other, it’s a very inexact and difficult task to blindly predict a hockey season.

Perhaps to be expected but the betting lines ended up doing better than what the fan consensus was. That’s why Las Vegas and the online betting industry stays in the green — they tend to come closer across the board more accurately than the people they’re taking money from. It might be said that sometimes lines are set to ensure even betting, but attempts for accuracy pay off when it comes to cash being on the line.

Here’s a chart of the Pensburgh picks.

To take a victory lap, we intentionally predicted 737 points to be earned in each division. The Atlantic division ended up getting exactly 737 points. We over-shot some (hello, Boston!) and under-projected others (Ottawa, Montreal) but absolutely nailed the end result. That and a 10.25 variance made for a very strong overall outlook.

In the not so good part, the Metropolitan division calls ended up going very poorly. Pretty brutal stuff on NJ, NYR and CBJ, The division as a whole only took 713 total points, and the Metropolitan did not earn any Wild Cards as a result with only three teams getting over the 90-point plateau this season (which would have been very difficult to call ahead of time).

Due to the mess in the Metro, our average error was 12.4 points, worse than all the models. That makes sense, being as I spent about an hour or so and wasn’t going off any detailed data to come up with these shots in the dark. On the plus side, that 10.25 point variance in the Atlantic more than holds its own as a very good shot at it. (Last time gloating, promise!)

We were able to accurately call six of the eight playoff teams; Boston and NYR didn’t make it, Ottawa and Montreal stepped up. It was a “close but no cigar” outlook on the Senators, with just not quite enough courage or insight to drop the Bruins down to where we see they belonged. (Of course, back in October, no one knew Boston would panic fire a good coach and eventually trade their captain, plus Brandon Carlo and Charlie Coyle, among others..)

Of all the non-playoff teams to potentially make a move, I like Ottawa the most to step it up. The addition of Linus Ullmark should help them in net, but after a few players they get really thin really quickly on defense, which is troubling when trying to put a lot of faith in them making a major step forward. A wild card could be the coaching of Travis Green, but at this point that could end up being a positive or negative factor.

Next time we probably shouldn’t predict each division to take the exact same point total and get a little more independent to attempt for more accuracy. Other than that, the inexact science of making predictions can be a cruel exercise. The attempts to be better will have to wait for next season.