🏈 NFL DECODED: A Forensic System Architecture Investigation PIECE 7 of 18 — The Combine Pipeline ← Piece 6: Media Capture | Piece 8: Injury Designations in a Gambling Universe →

🏈 NFL DECODED: A Forensic System Architecture Investigation
PIECE 7 of 18 — The Combine Pipeline
← Piece 6: Media Capture  |  Piece 8: Injury Designations in a Gambling Universe →

The Combine Pipeline

Before a player earns his first NFL dollar, before he signs his first contract, before he is anyone's employee — the NFL measures him. Completely. Permanently. And the data never expires.

Every February, approximately 300 college athletes travel to Indianapolis for the NFL Scouting Combine. They are measured, timed, weighed, scanned, interviewed, and psychologically evaluated over four days. Sensors in their equipment track their movement to within inches, ten times per second. Machine learning models calculate a "Draft Score" — a composite athleticism rating benchmarked against every player in the historical database. The results flow to all 32 teams simultaneously.

None of these players are NFL employees yet. None of them have signed contracts. None of them have union representation at the combine — the NFLPA does not represent players until they are drafted. They attend because not attending is effectively career suicide. They consent by showing up.

What they are consenting to has never been fully mapped.

In 2011, the CBA that introduced the rookie wage scale — analyzed in Piece 3 — also contained a provision that most players, agents, and analysts have never focused on: players agreed to have their on-field location and health metrics tracked. That consent was granted in 2011. The tracking technology being built on top of it in 2025 — real-time biometric sensors, AI movement prediction, machine learning Draft Scores, gambling data integration — did not exist when the consent was given.

The consent was signed for a world that no longer exists. The data infrastructure built on top of it operates in a world the signatories could not have anticipated. This is the Combine Pipeline — and it is the foundation of the NFL's data moat.

What the Combine Actually Collects

📊 THE COMBINE DATA ARCHITECTURE — Public Documentation, 2025

Players invited annually: ~300 prospects
Duration: 4 days, Indianapolis (Lucas Oil Stadium)

Physical measurements collected:
Height, weight, arm length, hand size, wingspan
Body fat percentage
40-yard dash (to hundredths of a second)
Vertical jump, broad jump, 3-cone drill, 20-yard shuttle
225-lb bench press repetitions

Sensor data collected (2023–present, via biometric sensors):
Movement tracking: tenth-of-a-second accuracy
Speed, acceleration, deceleration at every point
Route paths and movement patterns
Positional data within inches

Outputs generated (2025 Combine):
Draft Score — AI composite athleticism rating vs. historical database
Position-specific performance benchmarks
Combine IQ dashboard — publicly launched February 2025
All data flows to all 32 teams simultaneously

Additional evaluations (not all public):
Wonderlic cognitive test (or successor)
Psychological evaluations
Team interviews (20 minutes per team, up to 60 teams-player sessions)
Medical examinations including injury history review
Drug screening

The physical measurements are the visible layer. The sensor data and the machine learning outputs built on top of them are the architectural layer. And the psychological and cognitive evaluations — conducted by team-hired consultants, results shared league-wide — are the layer almost never discussed in public Combine coverage.

Source Layer: The 2011 Consent That Keeps Expanding

⬛ FSA — Source Layer The source of the NFL's data collection authority over players is a single CBA provision: the 2011 collective bargaining agreement in which players agreed to have their on-field location and health metrics tracked. That consent is the legal foundation for a data infrastructure that has expanded continuously since 2014, when Zebra Technologies began embedding RFID location sensors in shoulder pads. The players who signed the 2011 CBA consented to tracking that did not yet exist — and the players subject to that tracking today inherited consent they did not personally give.

The sequence matters architecturally:

• 2011: CBA signed. Players agree to location and health metric tracking. The tracking technology: nascent.
• 2014: Zebra Technologies begins embedding RFID sensors in shoulder pads. Location tracking at scale begins.
• 2017: AWS partnership launches. NGS moves to cloud infrastructure. Machine learning begins processing tracking data.
• 2018: PASPA repealed. Legal sports betting explodes. Player tracking data becomes inputs for a $13.7 billion annual gambling industry.
• 2023: Biometric sensors introduced at the Combine. Draft Score AI model launched.
• 2025: Combine IQ dashboard goes public. For the first time, fans and analysts have access to the same AI athleticism ratings teams have used since 2024. More than 200 new data points are generated on every play of every game.

At no point in this sequence did players vote on whether the expanding infrastructure was consistent with their original consent. The 2011 provision was written broadly enough — location and health metrics — to accommodate technologies that did not exist in 2011. The consent was structural. The data architecture built on top of it was not.

The 2011 CBA gave the NFL a blank check to track players. The NFL has been cashing it ever since — with increasingly sophisticated technology, for increasingly commercial purposes, in an increasingly valuable data market.

Conduit Layer: The Data Infrastructure That Runs It All

⬛ FSA — Conduit Layer The NFL's player tracking infrastructure runs on a four-partner architecture: Zebra Technologies (RFID sensors and hardware), Wilson Sporting Goods (ball-embedded chips), AWS (cloud computing, machine learning, and storage), and the NFL itself (data governance and licensing). Each partner has a distinct commercial relationship with the data. Together they form a pipeline that converts player movement into a permanent, AI-processable, commercially licensable data asset.

The Hardware Layer: Zebra and Wilson

The tracking system captures player data such as location, speed, distance traveled and acceleration at a rate of 10 times per second, and charts individual movements within inches. This is accomplished through RFID sensors embedded in every player's shoulder pads and chips inside game balls. Altogether, an estimated 250 devices are in a venue for any given game, requiring a team of three operators at every game to confirm all tracking systems are functioning properly.

Every player. Every play. Every game. Every season. The data accumulates into a longitudinal record of each player's physical capabilities, movement patterns, and performance trajectory that follows them throughout their career — and beyond it, as historical data for AI model training.

The Cloud Layer: AWS and Machine Learning

The raw data is used to automate player participation reports, calculate performance metrics, and derive advanced statistics through machine learning on AWS. More than 200 new data points are created on every play of every game.

The AWS partnership — active since 2017 — is not simply data storage. It is the machine learning infrastructure that converts raw position data into predictive models. The NFL's Big Data Bowl, now in its eighth year, crowdsources data science talent to build new analytical models using NGS tracking data. Over 75 Big Data Bowl participants have been hired in data and analytics roles in sports, with more than 60 joining the NFL family. The competition is simultaneously a talent pipeline and an open-source model development program — the NFL gets external data science talent to improve its proprietary analytical infrastructure at a prize cost of $100,000 per year.

The Combine Layer: Where It Starts Before Employment

As players showcase their strength and agility during the Combine, biometric sensors track player movement data with tenth-of-a-second accuracy. The NFL's Next Gen Stats team takes this data to calculate the Draft Score and inform other metrics using machine learning models.

The 2025 Combine introduced the Combine IQ dashboard — visualizing data for fans and analysts, showcasing comparative player rankings and Draft Scores created in partnership with AWS to measure player athleticism and provide insight into a player's potential success in the NFL using historical comparisons and position-specific benchmarks.

Three years of Combine sensor data now exist. The NFL's own analytics team noted: "What really matters is you need a runway of seasons of data points to be able to find out what's actually the signal in the noise." The runway is being built. The signal extraction is underway. And it begins before the player has any employment relationship with any NFL team.

Conversion Layer: Five Ways the Data Becomes Value

⬛ FSA — Conversion Layer Player tracking data converts into five distinct forms of value simultaneously: team competitive intelligence, broadcast enhancement, gambling market inputs, AI model training, and the data licensing revenue stream mapped in Piece 5. The player generates all five. The player is compensated for none of them specifically. The data is the asset. The player is the instrument.

Value Stream 1: Team Competitive Intelligence

The most obvious use: all 32 teams receive NGS data to inform game planning, player evaluation, injury risk assessment, and roster construction. The tracking data tells teams not just what a player did but how he did it — his acceleration patterns, his route efficiency, his load accumulation, his physical deterioration over a season. This intelligence is the primary justification for the tracking system and the one most clearly aligned with player interests (better injury management) as well as team interests.

Value Stream 2: Broadcast Enhancement

NGS data feeds directly into broadcast graphics — the speed, route, and probability overlays that appear during game telecasts on every network. NFL media and its broadcast partners can find new storytelling opportunities beyond the box score. The tracking data makes broadcasts more engaging, which supports viewership, which supports the $10 billion annual media rights value mapped in Piece 6. The player's movement data is a production asset for broadcasts the player does not share revenue from beyond their base salary.

Value Stream 3: Gambling Market Inputs

As mapped in Piece 5, the NFL's official data licensing to sportsbooks — through Sportradar and Genius Sports — converts NGS tracking data into real-time betting market inputs. In-game and micro-betting products — wagering on individual plays, drives, and possessions — require exactly the granular, real-time positional data that NGS produces. The next frontier for Next Gen Stats is likely to involve the integration of biometric data into player analytics — real-time monitoring of vital signs, fatigue levels, and physiological responses. When that frontier is crossed, player biometric data — heart rate, fatigue indicators, physiological stress — will become gambling market inputs in real time. Players will generate that data with their bodies. The gambling industry will price it into betting lines. Players will receive nothing specifically for that contribution.

Value Stream 4: AI Model Training

Every season of NGS data becomes training data for the next generation of machine learning models. The Big Data Bowl makes this architecture explicit: external data scientists compete to build models using historical NGS data, the winning models are productionized into official NGS statistics, and those statistics inform team decisions, broadcast content, and eventually gambling products. The players whose historical performance data trains these models — including retired players whose career data remains in the system — have no ongoing relationship to the models built from their data.

Value Stream 5: The Draft Score as Permanent Record

The Combine's Draft Score — an AI-generated composite athleticism rating — is the newest and most consequential conversion. It creates a permanent, quantified, AI-generated assessment of a player's physical profile before he enters the league. That score is benchmarked against every player in the historical database, updated as career data accumulates, and available to all 32 teams for every subsequent roster and contract decision the player is involved in.

A player's Draft Score is, in effect, a permanent data file that follows him throughout his career. His negotiating position at every contract renewal is partially determined by data he generated before he signed his first contract — data collected when he was not yet an employee, not yet a union member, and not yet represented by anyone whose job was to protect his interests.

Insulation Layer: The Consent Architecture

⬛ FSA — Insulation Layer The data collection system is insulated from challenge through three overlapping consent mechanisms: the 2011 CBA provision (broad enough to cover technologies not yet invented), attendance consent (players consent by showing up to the Combine and games), and the safety narrative (tracking data is presented primarily as a player health and injury prevention tool, which is true — and which makes the commercial applications of the same data harder to challenge).

The Safety Shield

AWS and the NFL report 700 fewer injury-related game absences in 2023 and the fewest concussions on record in 2024 since tracking began. These outcomes are real and significant. Player tracking data has demonstrably improved injury management. The safety benefit is genuine.

It is also the primary public framing for a data system that simultaneously serves competitive intelligence, broadcast production, gambling markets, and AI model development. The safety narrative is the most powerful insulation mechanism because it is true — and because challenging the data system feels like challenging player safety. The commercial applications ride behind the safety justification.

The Attendance Consent Problem

Combine attendance is voluntary in a technical sense. No player is legally required to attend. In practical terms, declining to attend is a significant negative signal to every team in the league — one that most agents would counsel strongly against. The voluntariness is architectural: the system is designed so that non-participation is prohibitively costly, making attendance — and the consent it implies — functionally mandatory.

The players who attend are not yet NFL employees. They are not yet NFLPA members. The NFLPA does not represent pre-draft prospects. The most comprehensive data collection event in a player's career occurs at the moment when he has the least institutional protection and the most desperate career incentive to comply.

The Expanded Consent Problem

The 2011 CBA consent covers "on-field location and health metrics." The current NGS system collects positional data, speed, acceleration, route paths, and movement patterns. The next frontier — real-time biometric data including vital signs and fatigue levels — pushes against the boundary of what "health metrics" means. The NFLPA has maintained veto power over new tracking technologies under the CBA. But the veto operates after proposals are made — it does not prevent the architecture from expanding to the boundary of existing consent before new negotiations begin.

⚑ ANOMALY 18 — Consent Given Before the Technology Existed The 2011 CBA tracking consent was granted for a technological infrastructure that did not yet exist. RFID sensors in shoulder pads began in 2014. AWS machine learning integration began in 2017. Combine biometric sensors began in 2023. Gambling data integration exploded after 2018. Each expansion occurred within the boundaries of a consent framework designed for none of it. The players who signed the 2011 CBA did not consent to feeding a $13.7 billion gambling data market. They consented to tracking. The industry that tracking feeds has since changed completely.

⚑ ANOMALY 19 — The Pre-Employment Data Harvest The most comprehensive data collection event in a player's career — the NFL Scouting Combine — occurs before the player has any employment relationship with any team, before he has any union representation, and before he has signed any contract. The data collected at the Combine follows the player for his entire career and informs every contract negotiation he will ever be party to. No other American industry conducts comprehensive biometric and psychological profiling of workers before they are employed, without those workers having any union representation or legal employment relationship with the collecting entity.

⚑ ANOMALY 20 — The Big Data Bowl as Free Labor The NFL's annual Big Data Bowl invites external data scientists to build analytical models using NGS tracking data — data generated by players' bodies — for a total prize pool of $100,000. The winning models are productionized into official NGS statistics used by all 32 teams. The players whose career data trains and validates these models receive no compensation for their data's use in this process. The NFL receives permanent, productionized analytical tools. The cost: $100,000 in prizes. The data: generated by players earning an average of $3.3 million per year.

⛔ FSA WALL — Unknown Unknown Marker 005 The full commercial value of NFL player tracking data — across team intelligence, broadcast licensing, gambling data contracts, and AI model development — is not publicly consolidated. What is documented: the infrastructure, the partners, and the directional applications. What is not documented: the aggregate revenue generated by player data across all commercial channels, the specific terms of the Sportradar and Genius Sports data licensing arrangements, and the extent to which Combine psychological and cognitive evaluation data is retained, shared, or used beyond initial draft evaluation. The data moat has visible edges. Its depth is unknown.

Structural Findings — Piece 7

Finding 26: The NFL's player tracking infrastructure converts athlete movement into a permanent, AI-processable, commercially licensable data asset through a four-partner pipeline (Zebra, Wilson, AWS, NFL) that generates more than 200 new data points on every play of every game. The player generates the data. The infrastructure captures, processes, and monetizes it. The player is compensated through salary — not through data licensing, gambling integration, or AI model training revenue.

Finding 27: The 2011 CBA tracking consent was granted for a technological world that no longer exists. It has been used to authorize a data infrastructure whose commercial applications — gambling market inputs, AI Draft Scores, crowdsourced model development — could not have been anticipated by the players who signed it. The consent is structurally valid. Its application to the current data landscape is architecturally extraordinary.

Finding 28: The NFL Scouting Combine is the most comprehensive pre-employment biometric data collection event in American labor — conducted on workers who are not yet employees, without union representation, under conditions of functionally mandatory participation. The data generated follows the player for his entire career and informs every contract negotiation he will be party to. No equivalent pre-employment data collection exists in any other American industry.

Finding 29: The safety narrative — real, documented, and significant — functions simultaneously as the primary public justification for the tracking system and as its most effective insulation against scrutiny of commercial applications. Challenging the data system feels like challenging player safety. The architecture ensures these two things are inseparable.

The Combine Pipeline is the NFL's data moat. It begins before employment. It never ends. And the players who fill it have never been compensated for the full range of value it generates.

HOW WE BUILT THIS — FULL TRANSPARENCY

Human-AI collaboration: Randy Gipe (FSA methodology and investigative direction), Claude/Anthropic (research and drafting). All claims sourced from public record. FSA Walls mark where public data ends.

Confirmed sources used in this piece:
• NFL Football Operations (operations.nfl.com) — Next Gen Stats official documentation
• CIO Dive — "How the NFL gave the Combine a data makeover" (February 27, 2025): Combine IQ dashboard, AWS QuickSight, Draft Score mechanics
• AWS Sports page (aws.amazon.com/sports/nfl) — NGS infrastructure, injury reduction figures
• Wikipedia — Next Gen Stats: 2011 CBA consent provision, Zebra Technologies partnership history
• NFL Football Operations — Big Data Bowl official documentation (7th and 8th annual competitions)
• SlashGear — "How The NFL's Next Gen Stats Technology Actually Works" (January 2024): next frontier biometric integration analysis
• AWS Science blog — NGS defense coverage pattern ML model documentation

FSA Wall acknowledged: Combine psychological evaluation data retention and sharing practices are not publicly documented. The specific commercial terms of player tracking data licensing to sportsbooks are not publicly consolidated. Both are registered as Unknown Unknown Markers in this piece.

Coming next in this series:
Piece 8: Injury Designations in a Gambling Universe — The system designed in 1947 to prevent information asymmetry now produces it at industrial scale. Every Wednesday, every team knows something the betting market doesn't. The architecture of that gap — and who profits from it — has never been mapped.