The teaser from the three-ticket structure, walked as a sequence. UNC/TCU kicks first, so it is leg 1. Every hedge that cashes banks money that changes what the next one can lock.
In 2026, a serious football bettor writes off the first $20 as the cost of the show. But that $120 teaser stake? That's real money. Leg 1 kicks at noon EST: UNC +17.5 in a 10-point teaser. You're also on TCU -7.5. If TCU wins by 8–17, you win both ways — the hedge cashes and the teaser leg survives. You just got $100 of your $120 back, and you're still alive for legs 2 and 3.
UNC +17.5 teaser + TCU -7.5 hedge = middle window 8–17 · win both · bank $100Line is the number your teaser side takes before the tease, from that side's point of view. Reorder the legs by editing the lines if the kickoff order differs.
| Leg | Both cash when | Keys captured | Survives | Both |
|---|
A highlighted key is captured on both sides of zero — the window straddles pick'em, so that margin counts whichever team gets it. That is the strongest placement available.
| Path | Odds | P&L | 1 in |
|---|
HEDGE = leg busted and the hedge cashed · BOTH = the win-both-ways window · TEASER = leg survived, hedge dead.
Virginia and Stanford are the good news. Both windows straddle pick'em, which means 3 — the single biggest spike in football — counts for you whichever team gets it. A window that catches 3 on both sides is the best placement a ten-point tease can have, and you got it twice by accident of the numbers rather than by design.
UNC is the weaker leg for this purpose. At +7.5 the window sits out at a TCU win by 8 through 17, catching 10, 14 and 17 but not 3 or 7. That is a decent set of secondary keys, but it is the leg where a small line would have helped most — and it kicks first, so it is the one you cannot wait on.
The trap this tool exists to expose: hedging every leg raises the floor and lowers the ceiling, and past a point the expectation falls faster than the floor rises. Push the leg-2 stake up from 50 and watch both numbers move together.
Key-number weights are estimates of the historical margin distribution applied as multipliers on a normal envelope, then renormalized — directionally right, not survey data. Model output, not advice. 21+ where legal. 1-800-GAMBLER.