EDGE3/Basketball/Continuous Intelligence Loop
EDGE3 Methodology

Drift gets seen before it becomes a transfer.

EDGE3 is a recruiting and decision intelligence platform. We configure personalized AI recruiting agents for each coaching staff so they build their own customized decision intelligence engine — layering structured scouting data with the unstructured signals every staff already cares about.

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.

Difference 01

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.

Difference 02

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.

Difference 03

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.

Structured data

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.
+
Unstructured data

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.
Personalized AI recruiting agents

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.

Stream 01

Athlete survey

Intake + quarterly pulse

HC-personalized. Captures behavioral dimensions, NIL realism, scheme experience, injury history, and social-listening hooks.

Feedssurvey_satisfaction · survey_culture_fit · survey_authority_signal · survey_external_pressure · scheme_transition_signal
Stream 02

Parent survey

Intake + annual

Cross-reference points. Flags divergence between what the athlete says and what the family sees from the outside.

Feedsexternal_influence (validated / contested)
Stream 03

PFF weekly snaps

Every Tuesday in-season

Snap counts, positional rank, room dynamics, grade changes — the on-field reality the athlete is living.

FeedsAll Performance Trajectory features
Stream 04

Social listening

Continuous · 6h sweep

Public sentiment trajectory, influence chain, brand-deal velocity, frustration markers — from self-reported handles only.

Feedssocial_sentiment · social_external_pressure · social_engagement_delta · social_frustration_flag

How the loop works

Four streams in. One re-scoring engine. One signal out — refreshed the moment any input crosses a threshold.

In · monthly
Athlete survey
HC-framed
In · annual
Parent survey
cross-reference
In · weekly
PFF snaps
auto-pulled
In · always-on
Social listening
handles only
Re-score engine
EDGE3 models
Roster Retention Risk v8/v3 · Performance Trajectory v1 · Combined R/Y/G
Out · what the staff sees
Coach dashboard — Red / Yellow / Green per player
Most products refresh on a schedule (daily, weekly, monthly batch). EDGE3 re-scores on anomaly detection — the moment any signal crosses a threshold. R/Y/G can shift daily, not just at intake.

What the coach actually sees

Three things. One screen. One decision per player.

1

The current signal

  • Player name → R/Y/G dot
  • Trajectory tier — Rising / Steady / Declining / No Baseline
  • Top three drivers behind the current signal
2

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").

3

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.

1

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.

2

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.

3

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.

Live today
  • 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
Coming soon
  • 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.