31 prospects → 6 shortlisted → 2 finalists. One model. Three lenses. Click a tab below to see what each role sees.
Analyst
31 prospects · 18 metrics
Builds the board
→
Position Coach
6 cards · scheme + behavior
Narrows to film-worthy
→
Head Coach
2 finalists · R/Y/G grade
Final call
Player names + model scores in the tabs below are real (BB Combined R/Y/G v1, 2026-27 forward predictions). Scheme flags + position-coach narratives are illustrative pending live Coach Suite onboarding.
Roster Retention Risk Roster Retention Risk (RRR): the model’s read on how likely the player is to leave your program, on a 0–100 scale. Higher = more likely to enter the portal. Updated continuously as new signals come in.
Trajectory Performance Trajectory: whether next-season production is heading Up, Stable, or Down. Built from the player’s prior-year rates, scheme context, and behavioral signals.
Scheme Fit Scheme Fit: how cleanly the player’s game maps to your system. Green = clean match. Yellow = borderline / requires adjustment. Red = mismatch.
Behavioral
Social Signal
Combined
Gavin Paull
Tennessee · So
So
n/a
n/a
0.04
+0.86 Rising
0.10 Match
Coachable
Clean
0.06
+ board
Quan Lax
Austin Peay · Sr
Sr
n/a
n/a
0.06
+0.85 Rising
0.05 Match
Coachable
Clean
0.06
+ board
George Perkins
Rice · Jr
Jr
n/a
n/a
0.11
+0.85 Rising
0.20 Match
Coachable
Clean
0.09
+ board
Sencire Harris
Cincinnati · Jr
Jr
n/a
n/a
0.80
+0.34 Rising
50 Borderline
Mixed
Watching
0.59
+ board
Trevon Payton
Arkansas Pine Bluff · Jr
Jr
n/a
n/a
0.67
0.00 Stable
50 Borderline
Authority flag
Frustration
0.59
pass
Analyst flags 3 to position coach for film + behavioral review. Payton filtered off. Risk + behavioral both pointing the wrong way.
Lead-guard shortlist: fit + behavioral read
Position Coach view
Three flagged. Scheme fit, behavioral read, can-the-player-run-our-offense.
Gavin Paull
Tennessee · So · Combo G
Strong fit
Transfer risk (v3)0.04
Trajectory score+86 / Up
Combined R/Y/GGreen / 0.06
PositionCombo G
Scheme transitionMatch (illustrative)
Survey signalpending intake
Illustrative position-coach narrative: Lowest combined risk score on the board (0.06) + Up trajectory. Three years of eligibility. Profile reads as a Year-2 breakout candidate. Behavioral + scheme-specific reads populate from intake survey + scheme-transition signal once onboarded.
→ Recommend to HC for spring visit
Quan Lax
Austin Peay · Sr · Combo G
Strong fit
Transfer risk (v3)0.06
Trajectory score+85 / Up
Combined R/Y/GGreen / 0.06
PositionCombo G
Scheme transitionMatch (illustrative)
Survey signalpending intake
Illustrative position-coach narrative: Sr eligibility (one year window) but Up trajectory at strong production. Mid-major-to-D1 jump is the variable; survey + film read tell us if game speed translates. Use this profile when staff needs a 1-year impact piece, not a 3-year build.
→ Recommend to HC for spring visit
Sencire Harris
Cincinnati · Jr · Combo G
Watch
Transfer risk (v3)0.80
Trajectory score+34 / Up
Combined R/Y/GMid / 0.59
PositionCombo G
Scheme transitionBorderline (illustrative)
Survey signalpending intake
Illustrative position-coach narrative: Interesting case. Up trajectory (+0.34) paired with high transfer risk (0.80). The model says he's developing well but the program-fit context is unstable. Read this as: talent is real, environment is the variable. Survey + social listening would tell us where the friction is. Hold for HC discussion.
→ Hold. Flag for HC if we don't land Paull/Lax
Lead-guard finalists: your decision
Head Coach view
Two finalists. Your call.
Rank 1
Gavin Paull, Tennessee · So
Combo G, Sophomore. Trajectory model reads Up (+86), Roster Retention Risk v3 reads 0.04 (Green), Combined 0.06. Lowest combined-risk score on the board with the steepest up trajectory. Three years of eligibility. Best floor + best development window.Illustrative behavioral + social signal commentary populate from survey + social listening once onboarded.
Analyst confidence
9/10
data fit
Position Coach confidence
9/10
illustrative
Social signal weight
n/a
pending intake
PROCEED: schedule official visit
Rank 2
Quan Lax, Austin Peay · Sr
Combo G, Senior. Identical Up trajectory (+0.85) with low Roster Retention Risk (0.06). One year of eligibility makes this a one-shot impact pick. Not a development build. Right profile if we need a 1-year ceiling raiser at the position.Illustrative survey + film read populate from intake.
Analyst confidence
8/10
data + projection
Position Coach confidence
8/10
illustrative
Social signal weight
n/a
pending intake
WATCH: strong fit IF eligibility window matches roster need
Each confidence score is the model's read. The head coach can override and re-weight. EDGE3 is decision support, not decision replacement.
How we test the models
EDGE3 trains its AI recruiting agents the way coaches train their players and teams: against past seasons that already played out. Every accuracy number below is tested on past seasons the model never trained on.
Roster Retention Risk · RRR v3
78% accuracy
Tested on past portal cycles. Players who actually transferred vs. players who stayed. The model significantly out-predicts chance on data it never saw during training.
Performance Trajectory · v1
73% / 74% accuracy
Tested on whose production actually rose, held, or declined the next season. Improving direction 0.733, declining direction 0.739. Real outcomes that already played out.
How accuracy reads: 50% would be a coin flip. 100% would be perfect. Anything 70%+ means the model is meaningfully better than guessing. We test by hiding past seasons the model never trained on, the same way you scout last year’s tape before this year’s game. Players without a prior-year baseline are flagged "No Baseline" rather than imputed.
Refined Positioning
How staffs actually talk about recruits, now operationalized.
One side of the table: "He’s a can’t miss." The other: "He can’t play for us." Where does he fit in our room? Who else are we bringing in? Who does he remind you of? The two questions every staff meeting opens with, answered continuously. EDGE3 has the read before the room sits down, and updates between meetings, not just at intake.
Player Comparison · Staff Language → System Output
"Who does this kid remind you of? How does he stack up in our room?"
EDGE3’s comparison engine surfaces the answer instantly.
Pulls similar archetypes from your historical roster, national player library, and your current board
Side-by-side rate comparison across performance, scheme fit, and behavioral profile
Live positional ranking against every player you’re evaluating
The comp and ranking appear the moment the analyst flags the player.
Retention + Trajectory · Where the market is broken
"If we take him, how long is he here?" "What’s the likelihood we keep him?"
EDGE3 answers that before it becomes a problem.
Continuous R / Y / G signal powered by surveys, family input, weekly stats, and social signals
Trajectory tier: Up / Stable / Down
Retention probability updates as the season evolves
Drift is flagged weeks before April 1, not after.
Validation case study
What we said in 2024 vs. what the production model says now
In 2024, before Chol committed to UNM, EDGE3 flagged the same friction context the production model still reads today, two years and two model versions later. Drivers don’t move because they were never random.
2024 · Before commitment
EDGE3 Fit & Usage Risk Model · Pre-commitment report
Antonio Chol: projected risk read
Fit Score62.3 / 100
Risk Index37.7 / 100
Risk tierModerate to High
Risk drivers identified: Distance from home (Buffalo NY to Albuquerque ~1,800 mi · NCAA BB transfer avg is 643 to 746 mi). Prior portal history (former Rutgers). Repeat transfer pattern (Rutgers → JUCO → UNM). Socioeconomic stressors.
→
2026 · Production model
BB Combined R/Y/G v1 · Refreshed monthly
Antonio Chol: current signal
Roster Retention Risk Roster Retention Risk (RRR): how likely the player is to leave, on a 0–100 scale. 59% means elevated risk relative to peers.59% (Elevated)
Trajectory Performance Trajectory: direction next-season production is heading. +20 / Up means meaningful improvement expected.+20 / Up
Combined R/Y/G50% · Yellow
Model’s verbatim coach_insight: “Friction context: prior portal entry, repeat transfer, playing far from home.” Same drivers. Different decade. Independent validation.
What this proves
EDGE3’s risk-driver attribution is stable across model versions and validated on real outcomes that already happened. The model didn’t just predict at intake. It persists, updates, and surfaces the same actionable signal years later. That’s the loop in action. Watch the loop in motion →
Coach Olen’s request
The ask
Compare three transfer prospects against UNM’s thresholds. Find the slot fits for the 2025-26 frontcourt.
What EDGE3 delivered
Sports Reference verified rates. FIBA performance + injury records. Prior transfer + opportunity context. Three slot-fit grades against the 2025-26 baseline: Anya A-, Butka B+, Petraitis C+.
EDGE3 Personalized AI Recruiting Agents
Configured to surface what’s known, and what’s typically hard to access. Sports Reference advanced rates, FIBA + international records, conference media guides, and public archives, consolidated into one read. Today: data and decisions, delivered in one read.
Model coverage
D1 coverage and refresh cadence at a glance.
D1 players scored
4,979
2026–27 forward
Programs covered
365
all D1
Refresh cadence
Monthly
model + data
One use case. One live engine.
See a real EDGE3 deliverable. Then play with the live engine yourself.
Use case · UNM · 3 portal candidates
Transfer Portal Player Comparison Report
Three portal candidates graded against the program’s stated thresholds (DReb%, OReb%, Stl%, Blk%, height), with verified advanced rates, 2026-27 roster churn, and a slot-fit grade per candidate.
Search every D1 player. Compare any two side-by-side. Drag survey + social signals on a real Stretch 4 (Antonio Chol, UNM) and watch the R/Y/G signal recompute live. The same engine your staff would use, opened up for you to play with.
BB Roster Retention Risk v3 (78% accuracy), Performance Trajectory v1 (improving 0.733 / declining 0.739), Combined R/Y/G. Tested on past seasons the model never trained on. Coaching system fingerprints across 365 programs.