ChallengeFlag

NFL Draft Intelligence — ML-powered prospect rankings

What is ChallengeFlag?

ChallengeFlag is a machine learning model trained on 25 years of NFL combine measurements and college production data. It assigns every draft prospect a probability of becoming a Star, Starter, or Dud in the NFL.

The model does not watch film, read scouting reports, or factor in injury history. It finds statistical patterns that have historically separated stars from busts, then applies them to the current draft class.

What do the labels mean?

Star (~9% of drafted players)

2+ Pro Bowl selections, or ranked top 12 by career Weighted Approximate Value at their position.

Starter (~39%)

An established NFL starter who contributes meaningful snaps but does not reach Star-level recognition.

Dud (~52%)

Started 1 or fewer NFL seasons and career Weighted AV under 10. The majority of drafted players fall here.

What data does it use?

NFL Combine (1980-present)

Height, weight, and six events: 40-yard dash, bench press, vertical jump, broad jump, 3-cone drill, and shuttle run. Each is z-scored within position group and draft year. Participation flags handle missing measurements.

College Production (2004-present)

For skill positions, the model uses the final college season: receiving yards, receptions, yards per reception, and dominator rating (share of team receiving yards). These efficiency metrics are more stable across eras than raw counting stats.

Career outcome labels (2000-2021)

Labels come from actual NFL career outcomes. Players drafted after 2021 are excluded since their careers have not matured.

How much should I trust it?

ConfidencePositionsWhy
HighOT, G, C, DE, DT, LB, CB, SCombine metrics predictive. Large training set. Stable macro F1.
MediumWRCollege stats add signal. Use directionally.
LowQB, RB, TEToo few historical Stars to evaluate reliably. Directional only.

What it does not do