About the Model
MySportsAnalytics · NBA Prospect Projection Model
What This Is
This model projects how a basketball prospect's game will translate to the NBA by finding the
NBA players whose production before they reached the league looked most like his.
What those players went on to do becomes the prospect's projected range.
It is built on a database of every player active in the NBA from 2004-05 to 2025-26: their
NBA seasons (traditional and advanced stats), their college, international, G League and
junior-college seasons, and their awards. The 2027 class covers every prospect on six
current big boards and mock drafts.
Step 1 · Put Every League on One Scale
Twenty points a game in the NCAA is not the same as twenty in the EuroLeague. Each league gets a
league factor: how players' combined points, rebounds and
assists per 36 minutes changed between their last two seasons in that league and their first three
NBA seasons. Leagues with few historical players are blended toward a tier estimate so one outlier
can't swing the number.
log(factor) = (n × log(raw ratio) + 10 × log(tier prior)) / (n + 10)
| League | Factor |
| EuroLeague | 0.945 |
| Spanish ACB | 0.877 |
| EuroCup | 0.842 |
| NCAA Division I | 0.833 |
| G League | 0.799 |
Step 2 · Build a Production Profile
A player is described by his last two to three pre-NBA seasons,
weighted 60% / 30% / 10% from most recent back. The current season counts in full: its per-game numbers are
treated as if played over a full season, so a few games into the year a prospect isn't discounted. An
older, completed season with fewer than 15 games counts proportionally less. Camps, youth tournaments
and cup games are left out. The same rule is applied to the prospect and to every historical NBA player.
season weight = base weight (0.6 / 0.3 / 0.1) × min(1, games ÷ 15) for completed seasons (current season × 1), then rescaled to add up to 100%
Counting stats are converted to per 40 minutes and multiplied by the
league factor. Efficiency and age are added on top. Height, weight and projected draft slot count for a little, so production decides the comps.
Step 3 · Measure Similarity
Each stat is turned into a z-score against the pool of 2,220 NBA players' pre-NBA profiles.
Two players are 100% similar on a stat when their z-scores match, and 0% when they are three or
more standard deviations apart. Overall similarity is the weighted average across the stats both
players have.
stat similarity = max(0, 1 − |zprospect − zNBA player| ÷ 3)
| Stat | Weight | Notes |
| Points / 40 | 12 | league-adjusted |
| True shooting % | 12 | PTS ÷ (2 × (FGA + 0.44 × FTA)) |
| Age | 10 | age in the latest season |
| Rebounds / 40 | 9 | league-adjusted |
| Assists / 40 | 9 | league-adjusted |
| 3PA / 40 | 6 | league-adjusted |
| 3P% | 6 | only with 0.5+ attempts per game |
| FT% | 6 | |
| Draft slot | 5 | prospect: average board rank · NBA player: actual pick (undrafted = 61) |
| Minutes per game | 5 | |
| Steals / 40 | 5 | league-adjusted |
| Blocks / 40 | 5 | league-adjusted |
| FTA / 40 | 5 | league-adjusted |
| Height | 4 | counts for little |
| Turnovers / 40 | 4 | league-adjusted |
| Weight | 2 | counts for little |
Step 4 · Project the Career
The 20 most similar NBA players form the prospect's comparison group. Closer comps count much
more: each is weighted by similarity to the fourth power. The projection reports what that group
actually did in three windows: the rookie season, the
first five seasons (averaged) and the
peak (each stat's best season with 20+ games, e.g. a comp's best scoring year).
| Outlook | Meaning |
| Low | 25th percentile of the comparison group |
| Likely | median (50th percentile) |
| High | 75th percentile |
⚠️ Every comparison player made the NBA, so projections describe a prospect's outcome if he is
drafted and plays. They are not a probability of making the league, and they carry no team context.
Defense, feel, athleticism and character aren't captured. Use this alongside scouting, not instead of it.
NBA Player Cards
Career averages weight each season by games played; TS%, PER and BPM are weighted by minutes; win
shares and VORP are career totals. The peak season is the one with the highest VORP (20+ games).
"Before the NBA" is the player's final college, club or G League season.
Data & Updates
NBA stats come from Basketball-Reference; college stats from Sports-Reference; international,
G League and lower-level stats from RealGM. The 2027 board blends ESPN, Bleacher Report, FanSided,
NBADraft.net, ClutchPoints and CBS Sports. Prospect stats, comparisons and projections refresh
weekly. Early in the season a prospect's numbers rest on only a few games, so his comps can move a lot from week to week.