[DATA THAT THINKS]

Notes

Worked examples and methodology notes. Concrete numbers, no buzzwords. Each post walks math you can verify against the live product surfaces.

How a Chiefs/Rams spread gets computed end-to-end

2026-05-15

We walk a single NFL spread pick through all seven pipeline layers — de-vig, Elo, prediction markets, AI modifier, EV, Kelly, CLV. Concrete numbers at every step. No hand-waving.

Why we publish per-sport calibration N — and why you should ignore us at low N

2026-05-15

Calibration buckets are meaningful at N≥200 per sport. Below 50 they're noise. We tell you N on every bucket. Here's the honest reading guide.

Steam detection: what 'sharp money' actually looks like in our logs

2026-05-15

Coordinated line moves across multiple sharp books inside a 5-minute window. Two real examples from the steam log + what we filter out as noise.