The Only Signal That Held Up: Elimination Timing

We tested eleven strategy ideas against real market prices and real results. Ten of them came back flat. This is the one that did not — and the useful thing about it is that it costs you nothing to use. It is not a pattern to spot. It is a decision about when you submit your pick.

What elimination timing actually means

Late in the season, teams stop being equally motivated. By weeks 15 through 18 a chunk of the league is mathematically eliminated, and the important information about those games — who is actually going to play — does not exist on Tuesday. It arrives Friday and Saturday: a starting quarterback shut down, a running back rested, a coach on his way out managing snaps for players he wants to see in the spring.

Markets absorb that news as it lands. So the price on Saturday night is a genuinely better estimate than the price on Tuesday, in a way that is not true of a normal week where both teams are chasing the same playoff spot. That gap between the midweek price and the closing price is what we set out to measure.

Why this one survived the test

The test used 183 real games from 2021 through 2025, weeks 15 to 18, priced twice each: once at midweek and once at close, both from real bookmaker moneylines and both devigged so the two sides sum to 100%. Games were split into two arms — 144 in the elimination arm, where one side had nothing left to play for, and 39 in a clinch arm, where a team was already locked into its seed.

The crucial part was the control. Lines get better as kickoff approaches in every game, for boring reasons, so proving anything about motivation required a comparison group. We used week 5 and week 10 games from the same seasons, where every team is still playing for something, and asked whether the elimination games improved more than those did.

Sample

183 games

Elimination arm

144 games

Signal

z = 3.21

Window

2021–2025, wk 15–18

They did. On the elimination arm, closing lines were significantly more accurate than midweek lines relative to the control, at z = 3.21 — comfortably past the point where chance is a reasonable explanation. The clinch arm, at 39 games, is too small to call in either direction, so we report it as inconclusive rather than dressing it up as a second finding.

What to actually do with it

1. From week 15 onward, do not lock your pick early. Draft your intention on Tuesday, submit it as late as your pool allows.
2. Specifically hold when either team in your game has nothing left to play for. That is where the information gap is real.
3. Before you submit, re-check the price. If the number has moved hard against your team between midweek and Saturday, the market has learned something you have not.
4. Treat a large late move as a reason to switch, not a reason to argue. The move is the news.
5. Check whether your pool locks all picks at the first kickoff of the week rather than at your game's kickoff — if so, the Sunday-morning window you were counting on may not exist.

For every other week of the season this changes nothing. Weeks 1 through 14, the honest advice is unchanged: take the devigged favourite, budget your teams across the remaining schedule, and do not try to outthink the price.

What this finding does not say

It does not say eliminated teams lose more often — the market already prices that. It does not say you should fade resting teams; by the time you know they are resting, the line has moved. And it is not a licence to shop for edges in December generally. The measured effect is narrow: in a specific window, in specific games, later information is better information.

The full methodology — the arms, the control construction, the devigging and the significance test — is in the Edge Lab writeup. The complete register of all eleven factors, including the ten that did not hold, is in the Edge Lab, and how every win probability on this site is built is on the methodology page.