LineThe setup
A middle exists when the line on a game moves enough that you can hold both sides at different numbers. You bet the underdog at +9 on Tuesday; by Sunday the line has moved and the favorite is available at -6.5. Take it. Now: favorite wins by 7, 8, or 9 and both tickets cash. Any other result, one wins and one loses, and you're out only the vig — roughly 4.5 cents per dollar on the losing side.
That asymmetry is the whole idea: a small, known cost on most outcomes against a double win on some outcomes. Whether it's a good bet is a single arithmetic question — does the middle hit often enough to pay for the vig?
The break-even math
With two standard -110 bets, hitting the middle roughly doubles your money on the ticket, while missing costs about 4.5% of one stake. Run it through and the break-even hit rate lands near 1-in-21 — about 4.8%. So every middle reduces to: is the probability the game lands in my window better than 5%?
In football, key numbers answer that instantly. A window that contains 7 — like the +9 / -6.5 example, covering 7, 8, 9 — comfortably clears 5%, since 7 alone decides a meaningful share of NFL games. A window containing 3 is even better. A two-number window on dead numbers like 11 and 12 usually isn't worth the vig. This is the same margin-clustering logic that powers Wong teasers, pointed at a different weapon. Totals middle the same way: under 47.5 at one book, over 44.5 later or elsewhere, and 45–47 wins both.
Middling vs. arbitrage vs. hedging
People conflate the three. Arbitrage locks a small guaranteed profit no matter the result — no scenario loses. A middle risks a small guaranteed-ish loss (the vig) for a shot at a large double win. A hedge reduces the risk of a position you already hold, usually sacrificing EV for certainty. Middles are the only one of the three that's a genuine positive-expectation bet when the window covers key numbers — you're being paid for line movement the market handed you for free.
How middles actually show up
You rarely create middles on purpose; you harvest them. The two natural sources: early positions that the market moves through (bet a good number early in the week, and when steam pushes the line 2.5+ points, check whether the other side now completes a key-number window), and slow books disagreeing (one book hangs 6.5 while another sits at 9 — bet both immediately, because that gap will close in minutes). The discipline points: both sides must clear the 5%-adjusted break-even at the actual prices, extra juice on either leg raises the required hit rate fast; size both legs so a miss costs a fraction of a unit, not a unit; and log middles in your record as one combined position, or the string of small vig losses between hits will look like a losing streak instead of the cost of doing business.