How the Prospect Model Works
April 2026 Β· Draft Chalkboard
The PWHL has held three entry drafts (2023, 2024, and 2025), drawing primarily from the NCAA women's hockey pipeline. That's 90 picks worth of data β enough to ask: which college signals actually predict where a player gets drafted?
We tested everything we could extract from public college data β awards, national team experience, production stats, scoring trajectories, conference strength. Most of it was noise. Four signals survived.
The 4 Signals
Each distinct college award = 1 point. All-American, All-Conference, Patty Kazmaier finalist, Conference Player of the Year, Rookie of the Year, All-Rookie Team.
We count unique award types, not repeat wins. A two-time First Team All-American still scores 1 for that category. Team achievements (national championships, Frozen Four) are excluded β they reflect the program, not the individual.
Has the player represented their country at the senior level? Olympics, World Championships, or named to a senior national team roster.
This is the strongest single binary signal: +39% separation between early-round and late-round picks. If a college player has already worn the maple leaf or stars and stripes at the senior level, the PWHL has noticed.
U18 Worlds, World Juniors, NTDP, or any youth national team program.
A baseline qualifier. Most drafted players have youth NT experience. Its value is as a filter β players without it are fighting uphill.
Did the player produce β₯0.75 points per game over β₯25 games in their first NCAA season?
The threshold is deliberate: 25 GP filters out small samples, and 0.75 PPG as a freshman is genuinely impressive (most freshmen on ranked teams are in the 0.3-0.5 range). This signal showed a +28% early-vs-late delta in validation. Raw production, not freshman awards β Rookie of the Year and All-Rookie were not predictive on their own.
What Didn't Make It
Several signals that seem intuitively useful didn't hold up:
- Current production β PPG in the draft year didn't separate early from late picks beyond what freshman impact already captured
- Scoring trajectory β Year-over-year improvement was too noisy with only 42 data points
- Conference multiplier β Weighting WCHA/ECAC higher than CHA added complexity without improving rank correlation
- Frozen Four / championships β Team achievements, not individual signals
- Academic awards, MVP awards β Not predictive of draft position
Total = Awards + SrNT + YthNT + Freshman
The total signal score is a simple sum: award count + senior NT (0 or 1) + youth NT (0 or 1) + freshman impact (0 or 1). No weights, no multiplication, no adjustments.
This simplicity is intentional. With 42 validation points, a complex model would overfit immediately. The rank correlation (Ο = 0.425β0.492 with draft position across entry drafts) comes from signal selection, not signal weighting.
Class-Year-Adjusted Tiers
A senior has had 4+ years to accumulate awards and NT caps. A sophomore with the same total signal score has done it in half the time. The tier system adjusts for this:
| Tier | Sr / 5Y | Jr | So | Fr |
|---|---|---|---|---|
| Elite | β₯8 | β₯6 | β₯4 | β₯3 |
| Tier 1 | β₯5 | β₯4 | β₯3 | β₯2 |
| Tier 2 | β₯3 | β₯2 | β₯2 | β₯1 |
| Tier 3 | β₯1 | β₯1 | β₯1 | β₯1 |
Draft Class Mapping
The board organizes prospects into projected draft classes based on NCAA tenure (the higher of listed class year and seasons of stats data):
- 2026 Draft β Seniors and 5th-years (tenure β₯4)
- 2027 Draft β Juniors (tenure = 3)
- 2028 Draft β Sophomores (tenure = 2)
- 2029 Draft β Freshmen (tenure = 1)
Goalies
Goaltenders share the same signal framework as skaters β awards, senior NT, youth NT β but freshman impact is set to 0 (their stats are saves and GAA, not points). PPG shows as β--β in the table. They appear alongside skaters in each draft class because the PWHL drafts them from the same pool.
Data Sources & Limitations
The board scans 37 NCAA Division I women's hockey programs. Player bios (the source of award and national team signals) come from school athletics websites. Stats come from school-provided APIs.
The main limitation is bio coverage. Some schools have sparse or missing bios β these players show a warning in their expanded view. A player with no bio isn't necessarily undecorated; we just can't see their awards. The βno bio dataβ flag is there so you know what you're looking at.
With only two PWHL entry drafts to validate against, this model is directional, not definitive. It tells you which players have accumulated the signals that correlate with being drafted early. As more drafts happen, the model can be re-validated and refined.
International Prospects
Not every PWHL draft pick comes from NCAA hockey. Across three drafts, roughly 10β15% of selections have been European professionals β players from the SDHL (Sweden), ZhHL (Russia), Naisten Liiga (Finland), and other top leagues.
The NCAA signal model doesn't apply to these players (they have no college awards or freshman impact to measure). Instead, we identify international draft targets by archetype matchingagainst the European pros who have already been drafted.
The Drafted European Archetype
Looking at the non-NCAA players selected across three PWHL drafts, a clear profile emerges:
- Senior national team experience β Olympics and/or IIHF World Championships. This is nearly universal among drafted Europeans.
- Multi-year pro career β Established veterans in a top league (SDHL, ZhHL, Naisten Liiga), typically 4+ seasons.
- Age 23β31 β Old enough to have accumulated credentials, young enough to contribute for multiple PWHL seasons.
- League awards or statistical dominance β Scoring titles, Defender of the Year, Goalie of the Year, All-Star selections.
Speculative Tiers
International prospects are placed into two speculative tiers based on how closely they match this archetype:
| Tier | Criteria |
|---|---|
| Tier 1 | Olympic/Worlds veteran + league award winner + confirmed or expected to declare for the PWHL draft |
| Tier 2 | Senior NT experience + established pro career + expected to declare, but fewer marquee credentials |
These tiers are directional, not modeled. There's no formula β it's pattern matching against a small sample of drafted Europeans. As more international players enter the PWHL draft, this can evolve into something more rigorous.
Who's Not Included
Several high-profile European players are excluded because they're already in the PWHL system or have committed to staying in Europe:
- Nadia Mattivi β Signed by MontrΓ©al Victoire (already PWHL)
- Noora Tulus β Previously drafted (2024, NY Sirens 3rd round)
- Jenni Hiirikoski β Re-signed with LuleΓ₯ HF, not entering the draft
- Elisa Holopainen β Re-signed with FrΓΆlunda HC through the 2026 Olympics
Case Study: Caroline Harvey
Wisconsin defenseman Caroline Harvey scores a 10 β the highest total signal on the 2026 board. Here's how it breaks down:
She maxes out every signal category. Made the US Olympic team at 17, became Wisconsin's first freshman All-American since 2001, and has won conference defender of the year twice. The model's job is to surface players like this β and to show you exactly why.
Case Study: Petra Nieminen (International)
Finnish forward Petra Nieminen is the top-ranked international prospect for the 2026 draft (THN #11 overall). Here's why she fits the drafted European archetype:
Nieminen checks every box: Olympic experience, dominant league production, a scoring title, and confirmed intent to enter the PWHL. She's the archetype the speculative tier system was built to identify.