As my followers can tell you, I’m obsessed with the “market value” of NHL hockey players, using performance statistics to approximate their expected salary. I have been on this journey for more than 20 years, and have used many different algorithms to quench my thirst, each one better than the last. The market is continuously evolving, so you can’t just make one perfect formula to use forever. In the past when I tried to predict future free agent contracts, I’d use a combination of algorithm and the averages from lists of statistically comparable players at similar ages.
My evolution for these 2023 predictions was to move away from comparables lists, where outliers can have a negative misleading impact. Instead, I made lists of 15 free agents with comparable statistics and the contracts they received for most of the deals signed in the last 5 years. I then took the average stats and the average contract for each list, and made a database of averages. So for my 2023 free agents, I’d take their stats (second half of 2021/22 and first half of 2022/23) and isolate the averages that were the closest to that player. We’ll find out in July whether this was a smart strategy decision.
At this point in my journey, I’m less concerned with writing an algorithm that dictates what everyone should get paid (like some places try to do with goals above replacement models), but rather trying to actually predict what their contract WILL be, not what it SHOULD be. The market ain’t perfect and I’ve spent plenty of time breaking down bad contracts (writing an entire book about them in 2020), but still believe that on a whole, the market is smarter than me. NHL teams employ dozens of people, many of them smarter than me. I’d rather guess right than wrong.
The predictions you are about to browse are not the results of my algorithm, or the average of the averages. Those inputs were simply consulted, with me making a “judgement call”. Many forecasters will create a good formula and just trust the results for every player. Not me. The problem is, not every player accepts the highest bid. It’s very common for players to take lesser bids to play in more desirable destinations. If your algorithm says Zach Parise should be making $3M, but he’s the same player he was last summer when he signed for $750K, then that’s a likely outcome on the next deal. Zach has plenty of money.
So here are my midseason free agent contract predictions for the summer of 2023 (337 total). There will be a final list of predictions coming out in June. I like to get an early list out for all those who sign extensions in the next 6 months, but the list that comes out in the summer is generally much closer to the actual outcomes. The second half of the season has a strong influence over the salary that player will procure in July. Releasing a midseason list also helps those of you in fantasy dynasty where real cap hits count in your league can use this to make decisions on pending FAs.
To view last
summer’s list, click
here:
My #NHLFreeAgency predictions for 2022 were remarkably accurate for restricted free agents, with a 94% R^2, out of 112 signed thus far. My trend line also fits perfectly along the x=y axis, so I neither over-predicted or under-predicted the group. How did Evolving Hockey perform? pic.twitter.com/61CxG7q5QY
— Hockey Economist (@Hockeconomics) August 5, 2022
UFA Forwards
RFA Forwards
UFA Defense
RFA Defense
Goalies
$4.1M is Mackenzie
Blackwood’s required qualifying offer, though he should be worth considerably
less. Otherwise he’s be down in the $2M range.
No comments:
Post a Comment