Probability-based frameworks for interpreting NFL Combine performance
GridStrength quantifies how athletic testing shifts the odds of NFL career success — without overpromising prediction.
The problem with athletic testing
NFL Combine performance contains meaningful signal, but no physical profile reliably separates players who will succeed from those who will not. Group-level differences exist, yet individual outcomes overlap substantially. GridStrength is designed to help decision-makers interpret this reality with discipline rather than false certainty.
A probabilistic framework for interpretation
GridStrength is built on a simple but critical premise: athletic testing should inform probability, not dictate outcomes. Combine performance is most useful when interpreted at the population level, where group differences can be identified, and at the individual level, where overlap and uncertainty must be acknowledged. Rather than applying hard thresholds, GridStrength evaluates how multiple performance characteristics jointly shift the likelihood of future success.
Three principles that guide interpretation
First, group-level differences are real. When athletes are examined collectively, meaningful differences in testing performance emerge across career outcome groups. These trends provide useful context for understanding which attributes tend to matter more at the population level. Second, individual outcomes overlap. Despite group-level trends, successful and non-successful players frequently share similar athletic profiles. This overlap limits the utility of rigid cutoffs and explains why classification accuracy remains modest even in well-specified models. Third, odds shift rather than flip. Individual performance characteristics are best understood as gradually increasing or decreasing the likelihood of success. Improvements in speed, agility, power, or size move probabilities in a favorable direction, but they do not eliminate uncertainty or guarantee outcomes.
What the data can — and cannot — say
Combine performance data is well suited to identifying population-level trends, informing screening decisions, and framing probabilistic expectations. When interpreted appropriately, these data help contextualize athletic profiles, prioritize further evaluation, and support communication across departments.
At the same time, Combine metrics cannot predict individual careers, eliminate false positives or false negatives, or account for the many non-physical factors that shape professional outcomes. Film, role fit, durability, coaching, and opportunity remain decisive and must be integrated alongside any analytical framework.