How EDGE3 models are different
Other tools tell you what already happened. EDGE3 tells you what’s coming — and re-scores the moment any signal moves.
Continuous, not batch.
Most products refresh on a schedule — daily, weekly, monthly batch. EDGE3 re-scores on anomaly detection: the moment any input crosses a threshold. R/Y/G can shift between meetings, not just at intake.
Cross-reference architecture.
Athlete survey, parent survey, weekly snaps, public posts. Any one source can be gamed; all four cannot be gamed at the same time. EDGE3 is the only platform that joins the four.
Personalized AI recruiting agents.
EDGE3 configures recruiting agents for each program — HC voice, scheme bank, position taxonomy, behavioral dimensions weighted to your system. The decision intelligence engine is yours, not generic.
Inputs: structured + unstructured, layered.
Coaches don’t scout from one feed. EDGE3 layers the sources every staff already trusts — on the structured side AND the unstructured side — into one continuous read.
The scouting pipelines. Verified player rates, recruiting class data, advanced metrics — the inputs every program already pays for.
- PFF · weekly snap counts, position grades, room dynamics
- Tackle Football (formerly SSA) · advanced FB analytics
- Sports Reference / BartTorvik · verified player rates
- 247sports / On3 · HS recruiting class quality + composite rankings
- CFBD / CBBD · transfer portal cycles + program-level data
- MaxPreps · HS competition strength + production
- More structured sources in the agent integration queue.
The reads only EDGE3 surfaces. Behavioral, contextual, and environmental signals coaches feel but can’t quantify alone.
- Smart surveys · HC-personalized, scheme-specific, NIL realism
- Parent cross-reference · family-side validation of athlete answers
- Social listening · follower drift, engagement patterns, frustration markers
- Socio-economic context · ACS hometown economics + SAIPE district data
- Search-trend velocity · recruiting-market interest signals
- More unstructured sources added as the agent layer grows.
EDGE3 configures personalized recruiting agents for each program. Each agent is tuned to the staff’s scheme, position taxonomy, HC voice, and decision priorities — layering both data sides into a customized decision intelligence engine the program owns. Same architecture, different fingerprint per program.
The four input streams
Each stream is independent. The loop doesn't break if one slows down.
Athlete survey
HC-personalized. Captures behavioral dimensions, NIL realism, scheme experience, injury history, and social-listening hooks.
Parent survey
Cross-reference points. Flags divergence between what the athlete says and what the family sees from the outside.
PFF weekly snaps
Snap counts, positional rank, room dynamics, grade changes — the on-field reality the athlete is living.
Social listening
Public sentiment trajectory, influence chain, brand-deal velocity, frustration markers — from self-reported handles only.
How the loop works
Four streams in. One re-scoring engine. One signal out — refreshed the moment any input crosses a threshold.
HC-framed
cross-reference
auto-pulled
handles only
What the coach actually sees
Three things. One screen. One decision per player.
The current signal
- Player name → R/Y/G dot
- Trajectory tier — Rising / Steady / Declining / No Baseline
- Top three drivers behind the current signal
The story behind it
"When Coach Olen asked you about playing time, you said…"
Plus the drift event that pushed it ("sentiment dropped 28% over 14 days; 3 frustration markers logged") and any cross-reference flags ("parent says distance worry; athlete said comfortable").
The recommended action
- Specific and time-bounded — "schedule a check-in this week"
- Tied to the most recent input — "reference the role conversation from October"
- Routed to the right staff member — HC, position coach, or director of player development
Why the personalization matters
Athletes reveal more when they believe the head coach is asking. Validated in focus groups with student-athletes and parents.
Every survey question is wrapped in the head coach's voice — not generic third-party survey copy:
"Coach Olen would like to know — when you think about your athletic career, what's the biggest financial goal you're working toward?"
This unlocks honesty an anonymous survey can't reach. The data quality is materially higher. The QwikU partnership amplifies the loop — synthetic-media avatars deliver a 5–8 second HC video intro per section, lifting engagement and completion.
Why the loop is defensible
Three properties make this hard to copy.
Cross-reference architecture
Athlete vs. parent vs. social posts vs. snap-data reality. Any one source can be gamed. All four cannot be gamed at the same time. EDGE3 is the only platform that joins all four.
Threshold-driven re-scoring
Most products run a daily / weekly batch. EDGE3 re-scores on anomaly detection — the moment any signal crosses a threshold. Coaches see drift faster than the rest of the market.
The personalization stack
HC-framed surveys plus synthetic-avatar delivery surface behavioral data the athlete won't disclose to a generic platform — the input that most heavily weights survey-driven model features.
What's live today vs. what's next
Roadmap status across the loop. Updated April 2026.
- Survey instrument v2.1 with personalization templating, scheme questions for every position, NIL realism, injury adversity, S&C / weight context, and social-listening hooks
surveys/survey_config.yaml - Scoring math: per-question → per-dimension → model feature columns
scoring_logic - Models with placeholder columns wired for survey + social-listening data
bball_performance_model_config_v1.json - Combined R/Y/G Red/Yellow/Green signal
football_combined_riskfit_2025.csv·bball_combined_riskfit_2026.csv
- QwikU synthetic-avatar integration — HC video intros per survey section
- Annual parent re-survey cadence trigger
- Coach-facing weekly digest email — "three players whose signal moved this week"
Companion docs
Where this loop connects to the rest of EDGE3.