Analysis.

The top 10 prospects by model probability score. M. Celebrini leads at 94 with the SHL continuing to outperform all leagues on aggregate score. Notable movers: two USHL players enter the top 15 for the first time this window.

OHL draftees converted to 200 NHL games at 22.4% vs the WHL's 19.8% over the 10-year dataset. The gap narrows considerably in rounds 3—7. WHL leads on volume with 18% more picks in the full dataset.

14.2% – the NHL conversion rate for round 4 picks in the 2019 draft class. Against a 10-year average of 11.8%, the 2019 fourth round significantly outperformed the model. Full data context, one number at a time.

End-of-season summary across all draft-eligible QMJHL prospects. Average model score, position breakdown, top performers, and historical comparison to 2019—2024 QMJHL cohorts. 47 prospects tracked this season.

The 2019 4th round converted at 14.2% against a historical average of 11.8%. 8 of 32 picks reached 200 GP. Full round breakdown with model scores, career GP, and conversion flag for every pick in the round.

Annual revisit of the 2020 class now that five years of career data exists. Model scores vs actual NHL outcomes. Which picks outperformed their probability? Which franchises came closest to the model's preferred slate?

Every defenceman picked in the 2025 draft, arrayed by model score, round, and league. Positional average score, conversion benchmark from comparable classes, and top overperformers relative to draft position.

Players the model scored above 55 who went undrafted in 2024. Historical conversion rate for undrafted players with similar scores: 6.2%. Six players identified – two have since signed AHL deals.

Colorado's 2021 class averaged a model score of 71.2 against the model's preferred slate of 74.8. The largest divergence came at pick 25 – the model preferred a Swedish defenceman at 78, Colorado selected a forward at 61.

Drafted in the 2nd round with a model score of 41.3 – 18th percentile of his class – Kucherov went on to play 900+ NHL games. Of 47 players drafted with similar scores in the dataset, only 6 reached 200 GP.

Two elite defencemen compared across every model variable at the time they were drafted. League, age-adjusted score, position rank in class, and how their actual careers mapped to the model's projection band.

Yakupov entered the 2012 draft with a model score of 89 – top 2% of all prospects in the dataset. He played 401 NHL games. The Miss series profiles high-probability players whose careers didn't follow the model's projection.

Full retrospective on the completed 2018 class. Overall conversion rate vs model prediction, franchise-by-franchise grade, best and worst value picks by round, and the five picks the model would most like back.

A GM profile built entirely from draft record. Wilson's San Jose tenure across 17 drafts: average pick score vs model preferred score, round efficiency, positional tendencies, and the three picks that defined his tenure.