The F19 draft-year factor

The draft-year valuation problem

The NHL Entry Draft uses a fixed cutoff: a player is first eligible in the calendar year he turns 18 (between September 16 of one year and September 15 of the next). That hard cutoff means two prospects born nine months apart can be drafted in the same year, one as a 17-year-old and one as a near-19-year-old. Raw production in their draft-year season is not directly comparable — the relative age effect, the league each one plays in, ice time, role, and team strength all change how that point total should be read.

F19 is the layer of the ProdigyChain algorithm that converts a draft-year season into a single normalised draft-readiness score on the same axis for every eligible prospect, regardless of league, exact age, role, or supporting cast. It reads the cross-league normalisation from F17 as input and adds the draft-specific weights on top.

How the draft-year weighting works

For every draft-eligible player-season, F19 computes a draft-readiness score from five inputs:

• Level-adjusted production. Raw points are first run through F17 so cross-league play in the WHL, OHL, QMJHL, USHL, NTDP, NCAA, J20 Nationell, MHL, Liiga U20, or the Russian KHL all enter F19 on the same scale.
• Birthday-relative age. A September 16 birthday and a September 14 birthday in the same draft year are read very differently — F19 applies a relative-age curve so the younger half of the cohort gets credit for production they delivered at a chronologically earlier point.
• Draft-window weight.The draft-eligible season (D−0) carries the heaviest weight; D−1 (the pre-draft year) and D−2 enter at lower weights so a prospect's arc — not just his peak season — informs the score.
• Role estimator. Top-line minutes against first-pair opposition matter more than equal-strength minutes in a checking role. F19 inherits the shot-share and on-ice deployment estimator that SBI produces and feeds it back into the draft-year score.
• Team-context adjustment.A 70-point USHL season on a top-seeded team with a high-end linemate reads differently than the same 70 points on a bottom-feeder roster. F19 normalises against the prospect's own team's relative strength and schedule difficulty.

The output is a single per-season scalar on a draft-readiness axis comparable across every eligible cohort. The 47-factor model then layers downstream weights on top — defensive impact, projection confidence, comp-finder distance — to produce the published rank.

The D−0 / D−1 / D−2 draft window

F19 reads up to three seasons of production per prospect:

• D−0.The draft-eligible season itself — the year ending on the June draft date. Weighted highest because it's the most recent and the season scouts have been actively evaluating.
• D−1. The pre-draft season (one year before eligibility). Weighted lower because the prospect was younger and the league context typically differed.
• D−2. The earliest year F19 considers. Weighted lowest of the three but still in the model — it catches early breakouts (a 15-year-old MHL appearance, a 16-year-old call-up to a senior European league) that signal trajectory.

Re-entry-eligible prospects (those passed over in their first draft year) are scored with the same window shifted forward — the re-entry year becomes the new D−0 — so the algorithm doesn't double-count or under-weight the years already seen.

Why this matters for rankings

Without F19, a draft-year ranking inherits whichever league the prospect plays in, whichever month he was born, and whichever team he happens to play on as confounding variables. A late-2008-born forward in the USHL who scores 55 points on a struggling team looks worse than an early-2008-born WHL forward scoring 70 points on the league's best line — but once F17 normalises the league gap and F19 layers age, role, and team context on top, the two often converge inside a much tighter band.

F19 is also why ProdigyChain can rank a Phase I USHL pick, a CHL Import Draft selection, an SHL-loaned Liiga prospect, and an NTDP forward on the same draft board without league-bias rank floors. The algorithm learns the weights from historical draft + NHL career outcomes — not from scouting bias toward any single league — then re-applies them every week as new data arrives.

Known limitations

• Mid-season league moves. A prospect who splits his D−0 season across leagues gets a weighted F19 blend that favours the higher-leverage stretch. The split is surfaced on the player profile so readers can see the underlying segment breakdown rather than just the blended score.
• Single-tournament samples.IIHF U18 World Championship production enters F19 with a higher prior toward regression-to-the-mean than a full league season — a hot three-game sample doesn't overwhelm the rest of the year.
• Injury-shortened seasons.F19 doesn't pad short samples. A draft-year prospect with only 18 games due to injury is scored on the per-game production he delivered, with the confidence indicator dropped to reflect the sparse sample.
• Late-bloomer / re-entry prospects.A re-entry eligible prospect has F19 re-anchored to the re-entry year as D−0. The earlier D−0 is retained as historical context but isn't treated as a current draft-year season.

These edge cases are why each player profile shows a per-factor confidence indicator next to F19 — readers should know when the draft-year score is operating on robust data and when it's working from a sparse or split sample.