A data-driven framework for evaluating the greatest quarterbacks in NFL history using era-adjusted statistics, normalization techniques, and multi-model scoring.
Rather than relying on a single metric or subjective ranking, this project builds four separate analytical models that evaluate different dimensions of quarterback performance. The models are combined into a final GOAT Score to produce a comparative ranking across eras.
Because NFL passing environments have changed dramatically over time, the analysis uses era adjustment and z-score normalization to compare players relative to their contemporaries rather than raw totals.
The core challenge of this project was comparing quarterbacks from the modern passing era to those of earlier eras. To solve this, the model utilizes a Relative Value Framework:
- Era-Adjustment (Relative to Mean): Each season's statistics are compared against the league average for that specific year. This ensures that a 4,000-yard season in 1980 is weighted appropriately compared to a 4,000-yard season in 2024.
- Z-Score Normalization: Every metric is converted into a z-score, representing how many standard deviations a player sits above or below the era mean. This allows "Greatness" (awards) to be mathematically combined with "Efficiency" (ANY/A).
- Weighted Aggregation: The final GOAT Score is a composite index. Each of the four pillars (Greatness, Best, Dominant, Clutch) is assigned a specific percentage weight based on historical significance and data reliability.
The workbook is engineered for scalability and clarity, following a three-tier software-style structure:
- Presentation Layer:
DASHBOARDandFINAL_GOAT_RANKINGS. These utilize data validation and conditional formatting for high-level insights. - Calculation Layer: Hidden sheets that handle the z-score math, weighting distributions, and era-lookup tables.
- Data Layer:
DATA_QB_SEASONSandDATA_AVG_SEASONS, acting as the "Source of Truth" for the model.
The model evaluates quarterbacks through four analytical lenses:
Greatest Measures career legacy and accomplishments, including major awards, championships, and long-term achievements.
Best Evaluates overall performance level using efficiency metrics such as ANY/A+ (Adjusted Net Yards per Attempt relative to league average).
Dominant Measures how much a quarterback separated himself from league averages during his career using season-level statistical comparisons.
Clutch Analyzes postseason performance and whether quarterbacks elevated their play in playoff situations.
Each pillar produces an independent ranking, which are then combined into a final composite GOAT Score.
The model evaluates 25 historically significant NFL quarterbacks spanning multiple eras of professional football. I relied on Pro Football Reference and StatMuse for data collection.
The dataset includes:
- career statistical totals
- season-by-season quarterback performance
- league averages by season
These inputs allow the model to construct era-adjusted and relative performance metrics.
This project was designed as a portfolio example of applied data analysis using spreadsheet tools.
Statistical Modeling
- Z-score normalization
- Weighted scoring models
- Era-adjusted statistical comparisons
Advanced Spreadsheet Techniques
- Multi-sheet analytical architecture
- XLOOKUP and conditional logic
- Data normalization workflows
Data Presentation
- Interactive dashboard design
- Visual ranking comparisons
- Structured analytical reporting
The spreadsheet is organized into several layers:
Presentation Layer
- Dashboard
- Final Rankings
- Quarterback Roster
Model Layer
- Greatest Model
- Best Model
- Dominant Model
- Clutch Model
Calculation Layer
Hidden sheets contain the underlying normalization calculations, raw data tables, and intermediate metrics used to construct the models.
- Start with the Dashboard sheet for a visual overview of the rankings.
- Review the Model Results sheet to see how quarterbacks rank across each analytical model.
- Explore individual model sheets to understand how scores are calculated.