Every game was priced from real bookmaker moneylines pulled from the historical odds archive, then de-vigged so the two sides add to 100%. That de-vigged number is the market's honest win probability. A factor only counts as an edge if games matching it beat that probability by more than chance would explain. Real head coach for each team in each game, from the historical game record. For every game, the prior head-to-head record between those two coaches computed using only games that had already happened — no hindsight leaking backwards. Games where one coach held a strong prior edge tested for whether that coach beat the de-vigged market probability.

edge lab · factor writeup · 6 min read

Coaching Head-to-Head: Do Some Coaches Really Own Other Coaches?

Q: What the numbers said

Sample: 5,295 games; Window: 2006–2025; Signal: z < 1.17 Flat. Coaches with dominant head-to-head records did not outperform their prices in the next meeting. The obvious reason is that most of these records are four or five games long, and most of what they actually measure is which coach happened to have the better roster during those years. The trap: A 7-1 record is impressive-sounding and statistically almost weightless. Eight games cannot separate coaching from quarterbacking, and you would need a lot more than eight to try.

Q: What it means for your picks

Ignore the graphic. Coaching quality does matter, and to the extent it does, it is baked into every team's price week after week. There is nothing left over for you in the head-to-head footnote. Team-by-team histories are in Explore; every tested factor is listed in the Edge Lab.

Head-to-head coaching records look like destiny in a graphic. Tested against the market across 5,295 games, they predict nothing the line hasn't already priced.

Broadcast graphics love this one: 7-1 lifetime against that coach. It reads like a curse. It is almost always a small sample wearing a nice suit.

What we tested

The hypothesis: a coach's prior record against a specific opposing coach predicts future results beyond what the market expects — that some matchups are genuinely owned.

How we tested it

Every game was priced from real bookmaker moneylines pulled from the historical odds archive, then de-vigged so the two sides add to 100%. That de-vigged number is the market's honest win probability. A factor only counts as an edge if games matching it beat that probability by more than chance would explain.

Real head coach for each team in each game, from the historical game record.
For every game, the prior head-to-head record between those two coaches computed using only games that had already happened — no hindsight leaking backwards.
Games where one coach held a strong prior edge tested for whether that coach beat the de-vigged market probability.

What the numbers said

Sample: 5,295 games
Window: 2006–2025
Signal: z < 1.17

Flat. Coaches with dominant head-to-head records did not outperform their prices in the next meeting. The obvious reason is that most of these records are four or five games long, and most of what they actually measure is which coach happened to have the better roster during those years.

The trap

A 7-1 record is impressive-sounding and statistically almost weightless. Eight games cannot separate coaching from quarterbacking, and you would need a lot more than eight to try.

What it means for your picks

Ignore the graphic. Coaching quality does matter, and to the extent it does, it is baked into every team's price week after week. There is nothing left over for you in the head-to-head footnote.

Team-by-team histories are in Explore; every tested factor is listed in the Edge Lab.

More: back to the research register · methodology · guides