NHL Machine Learning Draft

In this post, I'll try to highlight some potential 'sleepers' who are eligible for the 2019 NHL draft.

For our purposes, a sleeper is a player who's predicted to have significantly a better career than is expected given their projected draft round. Of course, the 2019 ML model-based predictions are used to estimate the player's NHL career, and we use draft projections found from around the web (or estimate from scouting service rankings) when the player will be chosen in the draft.

Okay, let's look at some players:

Brayden Tracey -- (Predicted NHL Career: 1019 GP, 752 P. Projected 2019 NHL Draft Round: Early 2nd)

Tracey probably can't be considered a true sleeper at this point, but given his metoric rise up the Central Scouting Service rankings (73rd at midterm, 21st at final), it's fair to consider his ceiling may be even higher.

An excellent season in the WHL (36G, 45P in 66GP) earned him Rookie of the Year honors. Combine that with his big frame and being on the younger side of the 2019 class, and you've got a guy who seems to check a lot of the boxes.

The ML model predicts he's top-end material; we'll see if scouts agree and he ends up being a first-rounder in the 2019 NHL draft.

Matias Maccelli -- (Predicted NHL Career: 0 GP, 0 P. Projected 2019 NHL Draft Round: 3rd).

Obviously, the first thought here would be: why is a player predicted to play zero NHL games a sleeper?

The model used by NHL ML Draft for the 2019 forward predictions produce intermediate data from which a final prediction is generated. The result of this is that we can say we think two players might be superstars, but we have different levels of confidence for each prediction. We could also a player was 'just barely' predicted the way he was, and that's the case for Matias Maccelli, who, according to the internals of the model, narrowly misses on having a significant NHL career.

In addition to information provided by the model, Matias Maccelli also moved up about 30 spots in the Central Scouting Service's rankings since the midterm. He had an impressive season with the Dubuque Fighting Saints of the USHL (72 P in 62 GP).

At 170 lbs, one would hope he can put on weight, and, it's worth noting the most probable outcome of the ML model still has him never hitting NHL ice. However, there's certainly reason for optimism here.

Nathan Legare -- (Predicted NHL Career: 650 GP, 418 P. Projected 2019 NHL Draft Round: 3rd)

(Sorry about missing accents. That's a site-wide issue that needs to be addressed at some point).

Legare had a superb season in the QMJHL with Baie-Comeau (45G, 42A, in 68GP). He gained 12 spots in the NA CSS rankings between miderm and final (was 64th, now 52nd).

Legare's age is a bit of a mitigating factor -- he's on the older side of the 2019 draft cohort. At 200 lbs., he also doesn't have the 'headroom' for more healthy weight that some prospects may have.

The ML model predicts he'd be an excellent value for a 3rd round pick. We'll see if he goes earlier in the draft or if someone ends up with what would be a 3rd round steal.

Robert Mastrosimone -- (Predicted NHL Career: 471 GP, 144 P. Projected 2019 NHL Draft Round: Late 3rd / Early 4th).

Similarly to Matias Maccelli, Robert Mastrosimone's sleeper prediction makes use of intermediate data from the ML model. The model feels most comfortable giving him a yeoman's career, but was not far from giving him a more impressive NHL career.

Losing three spots in the CSS rankings between midterm and final doesn't bode well, and he's a touch undersized at 5'10". He produced an impressive statline with the USHL's Chicago Steel(31G, 29A in 54GP), although he's on the older side of the 2019 draft class.

Mastrosimone also scored one point in four games wtih the US National U18 squad.

An outstanding playoff with the Steel (7G, 8A in 11 GP) also provides some intrigue.

Alex Beaucage -- (Predicted NHL Career: 703 GP, 146 P. Projected 2019 NHL Draft Round: 4th)

One of the youngest forwards elgible for the 2019 NHL entry draft, Beaucage had a terrific season with the Rouyn-Noranda Huskies of the QMJHL (39G, 40A in 68GP). At 6'2", he's certainly not lacking in the size department.

The ML model estimates he'll have a long career as a contributor, which would be a great value for a 4th rounder. However, given his production and other 'tangibles', there must be some reason he's only ranked 64th among North American skaters by the Central Scouting Service. Although the ML method certainly supports his success, he may also be a guy whose game doesn't quite translate to the NHL level.

Michael Gildon -- (Predicted NHL Career: 650 GP, 418 P. Projected 2019 NHL Draft Round: 6th)

Gildon was part of the 2018-2019 U.S. National U18 team, where he was 8th in scoring on a team loaded with talent. It's difficult to determine how the talent around him would have affected his season and how it will affect his draft stock. It's also notable that he was one of the younger players on the NTDP U18 roster.

Given that his Central Scouting Ranking is a lowly 110th, it appears as though the scouts have some reservation toward Gildon. As a comparison, his teammate John Beecher had a very comparable season (15G, 28A in 63 GP), but has a Central Scouting Ranking of 49th.

Still, considering only the ML prediction, he looks to be worth more than a 6th round pick.

---

Well, there's six names the ML model has identified as potential sleepers for the 2019 NHL draft. It'll be interesting to watch how they're picked on draft day, but we probably won't know the truth for four or five years. See you then!