Every value in this dashboard is a projection of a player's EvanMiya BPR (Bayesian Performance Rating) — a single number capturing total on-court impact. Each projection is a floor / base / ceiling : the base is the most likely outcome (leaned slightly optimistic, since the downside is dominated by unpredictable injuries and role changes), and the band around it is the realistic range. A team's projection aggregates its players' projections, weighted by each player's expected share of the minutes.

• EvanMiya — BPR (the target), its adjusted and box-score components, team ratings, and transfer-portal projections, 2016-17 through 2025-26. A paid subscription — downloaded, not scraped.
• RealGM — the canonical player identity and full career history across every level (D1, D2, JUCO, NAIA, international, NBA, G-League), plus rosters and biographical data.
• 247Sports & On3 — high-school recruiting rankings and ratings.
• NBA draft boards (NBADraft.net, Tankathon) — preseason consensus boards used to flag elite freshmen and international prospects.

Each player is routed to one model based on where he came from. Specific inputs and coefficients are proprietary and not shown here.

Returning players

A returner is anchored to his own prior-season BPR, and the model projects the change from there — a development curve by class year, a reversion pull on players who over- or under-shot their true level, and recovery credit after an injury year. Players who have never logged real minutes (walk-ons, redshirts) lean on program quality and recruiting pedigree instead. Pitfall: a player three-plus years in who has still never cracked the rotation is held to replacement level — he has already shown he is not a contributor.

D1-to-D1 transfers

Built from EvanMiya's own portal projection for the player, then adjusted for the new program and his experience; a transfer with no portal projection falls back to his prior D1 production. Pitfall: portal-heavy roster overhauls underperform their paper talent on average (new system, no chemistry, transfers who shined in an old role do not always replicate), so the model stays deliberately conservative on “new-look” teams.

Non-D1 transfers (JUCO / D2 / D3 / NAIA)

Translates lower-level production — scoring, efficiency, minutes — by level and by the quality of the D1 program the player is joining. Pitfall: there is a hard ceiling on how predictable a jump from low-level ball is (the competition gap is large and noisy), so these carry the widest uncertainty bands.

Freshmen

Driven mainly by recruiting rank and rating plus program quality, with a special elevation layer for the few consensus top-tier NBA-draft prospects the base model would otherwise cap too low. Pitfall: high-school recruiting services are a weak signal — rank explains only a fraction of how freshmen actually perform, cannot separate the #1 recruit who busts from the #1 who becomes a star, and systematically under-rates the very best; unranked freshmen get a program-quality baseline.

International (foreign pro → D1)

A tiered system that rates a prospect by the strength of his pro league and how much he dominated within it, with carve-outs for very young players buried on elite rosters, thin pro samples read through national-team events, and projected NBA picks read off draft boards. Pitfall: the data is seeded from players who eventually reached D1, so any league- or age-level statistic is survivorship-biased — we never compute “league averages,” and never compare a player across different leagues.

Team projections

Individual projections are rolled up using a projected depth chart (who plays, and how much, by position and role), so a team's number reflects both its talent and how the minutes are likely distributed. Pitfall: program quality is measured from recent history, so a team with a brand-new coach reads cautiously — validation shows roster-overhaul teams underdeliver often enough that the caution is earned.