The tied bet, in one paragraph
A push is what happens when a sports bet finishes exactly on the line. If you back the Cowboys -3 and they win 24-21, the margin equals the spread and the bet is a push: stake returned, no profit, no loss. The same outcome applies to totals (game lands on the number), runlines, pucklines, and any market priced against a whole-number reference. Pushes are the third outcome that recreational bettors forget exists. Sportsbooks do not forget — the entire architecture of half-points, alternate lines, and key-number juice is built around minimising the number of tickets that push, because a push pays the book zero hold.
Understanding the push is understanding how books price keys numbers, why a 'free' half-point is never free, and how parlay tickets behave when one leg ties the line. For a market that pays nothing, the push shapes more of the price you see on screen than almost any other concept in sportsbook math.
How a push settles on your ticket
Mechanically a push is a refund. The book voids the wager, returns the stake to the wallet, and removes the bet from the open-tickets list. On most US apps the ticket appears in bet history with a 'Push' or 'Voided' label. There is no commission, no fee, no penalty for the bettor and no hold for the book. From an accounting perspective, the wager is treated as if it never occurred — though in some operator-level reporting it is still counted as handle for state-tax purposes, because handle in most jurisdictions is gross dollars wagered regardless of outcome.
A small but important quirk: pushes do not roll over into bonus-wagering requirements at most US sportsbooks. If a bettor is working through a 1x or 5x rollover on a deposit match, a pushed bet does not count toward the playthrough at most books — the bet must have action (a win or a loss) for the wager to qualify. The exception is FanDuel and a few European operators, which credit pushes at 'stake-amount' toward rollover. Read the fine print on every promo because this rule meaningfully changes the cost of bonus farming.
Why the half-point exists
The single most important reason books offer 2.5 and 3.5 instead of 3 in the NFL is the push problem. Historical analysis of NFL games from 1978 to 2024 shows that roughly 14% of all games finish with a 3-point margin of victory, and roughly 9% finish with a 7-point margin. Those two numbers are the largest concentrations on the margin-of-victory distribution and the reason 3 and 7 are called 'key numbers.' If a book offered an unhooked -3 spread on every game, it would push 14% of its NFL spread handle — wiping out a significant share of its hold.
Instead the book offers -2.5 or -3.5 and prices the half-point. A move from -3 (-110) to -3.5 (-130) costs the bettor roughly 20 cents of juice. That 20 cents is the book's compensation for absorbing the push risk on the favorite's side. The customer pays for certainty; the book buys insurance against the most common NFL margin. Sharp bettors compute the break-even point: at -130 you need to win 56.5% to break even, and the historical edge from buying off the 3 is roughly 1.5-2% — not enough to justify the half-point at retail prices, but valuable when -3 is available at -105 on a soft book.
NFL key-number push frequencies
The distribution of NFL margins of victory is famously non-uniform. A handful of margins capture an outsized share of outcomes because of how football scoring is constructed: field goals (3), touchdowns (7), two-score swings (10, 14), and field-goal-plus-touchdown combinations (6, 10). The table below summarises the modern-era distribution and what each margin means for spread pushing.
| Margin | Historical frequency | Cumulative | Push significance |
|---|---|---|---|
| 3 | ~14.0% | 14.0% | The biggest push risk in football |
| 7 | ~9.0% | 23.0% | Second key number; common spread anchor |
| 10 | ~6.0% | 29.0% | FG + TD swing |
| 6 | ~5.5% | 34.5% | 2 FGs; relevant in low-scoring games |
| 14 | ~5.0% | 39.5% | Two-TD swing |
| 4 | ~4.0% | 43.5% | Mid-range margin |
| 17 | ~3.8% | 47.3% | 2 TDs + FG; one of the larger spikes |
| 1 | ~3.5% | 50.8% | Late safety / two-point conversion swing |
| Other | ~49.2% | 100% | Distributed across 30+ remaining margins |
Notice that just two numbers — 3 and 7 — account for nearly a quarter of all NFL outcomes. That concentration is the structural reason the books price half-points more aggressively around those margins than around any other. Buying the 3 (from -3 to -2.5) typically costs 20-25 cents of juice; buying the 7 typically costs 12-18 cents. Both are calibrated against the historical push frequency, with the book taking a 1-2% margin on the trade.
Pushes in parlays — the leg-drop rule
The single biggest customer-experience question around pushes is what happens to a parlay when one leg ties. Modern US sportsbook practice — including DraftKings, FanDuel, BetMGM, Caesars, ESPN BET, and Fanatics — is the leg-drop rule: a pushed leg is removed from the parlay and the remaining legs recalculate at the reduced parlay price. A 5-leg parlay with one push becomes a 4-leg parlay; if all four remaining legs win, the bettor is paid at the 4-leg price rather than the original 5-leg price.
The math: a 5-leg parlay of -110 sides pays roughly +2376; a 4-leg version of the same sides pays roughly +1228. The bettor still wins, but the payout is materially smaller because the pushed leg removed a multiplier from the product. This is the correct mathematical treatment — the book is not penalising the bettor, it is recalculating the bet as if the pushed leg had never been placed. Some offshore and older European books still grade pushed legs as full losses, which is a customer-hostile rule the modern US market has largely abandoned. Always check house rules before placing a parlay near a key number.
Same-game parlays (SGPs) are a different beast. Because SGP pricing uses a correlated-probability engine, removing a leg mid-grading is mathematically messy — the remaining legs may have been priced assuming the pushed leg's outcome. Books handle this inconsistently: DraftKings typically applies the leg-drop rule on SGPs, FanDuel sometimes voids the entire ticket, and BetMGM has changed its policy several times since 2022. The safe assumption: never bet an SGP that puts a leg on a whole-number total or unhooked spread unless you have read the operator's current SGP rules.
MLB totals and the whole-number problem
Baseball is the sport where pushes happen most often, because runs are integer-valued and books frequently set totals at 7, 8, or 9. A game total of 8.5 cannot push; a game total of 8 can and does. Most modern MLB books price totals on the half-point precisely to eliminate this risk, but during in-game and live betting some books revert to whole-number totals for simplicity. Live whole-number total pushes are common — perhaps 8-12% of MLB live totals priced on integers will push — and the bettor must understand that the displayed price already includes the push-risk premium.
The runline (a fixed 1.5-run spread) almost never pushes because no MLB game finishes with a 1.5-run margin (impossible by integer scoring), but the related 'alternate runline' at 2.5 runs can push at exactly 2 or 3 runs of margin if the book offers a rare whole-number alt. Bettors should treat any whole-number MLB total as a market with a meaningfully different risk profile than a half-point total.
Tennis tiebreaks and exotic push edge cases
Tennis is push-rich for a structural reason: tiebreaks. A match total set at 22.5 games is unlikely to push, but a set total set at 9.5 games can land on exactly 10 (a 6-4 set extended by tiebreak, scored as 7-6 under some scoring conventions). 'Total games in match' markets at major Slams routinely push at 38 or 39 games because the men's five-set format produces clusters of whole-number outcomes. Books mitigate by offering half-point totals on the most-traded markets and absorbing the push risk on niche markets like 'sets played' (where 3 vs. 4 vs. 5 are the only possible outcomes in a best-of-five).
Other exotic push edge cases include: hockey period totals (often whole numbers, frequent pushes); soccer 'total goals' priced as 2.0 or 3.0 in Asian Handicap variants (which split-push on whole numbers, returning half-stake); and prop bets on whole-number player stat lines (a quarterback over 249.5 passing yards can push on exactly 250 if the book offers the whole-number alt). Every push edge case has the same structural cause: the line lives on an outcome the underlying scoring system can actually produce.
The Asian handicap half-push
Asian handicap soccer betting introduces a quarter-line that produces a unique outcome: the half-push, or 'half-win/half-loss.' A bet on Manchester City -0.75 means the bettor's stake is split into two halves, one on -0.5 and one on -1.0. If City wins by exactly one goal, the -0.5 half wins and the -1.0 half pushes — net result: a half-stake win at the full price. This is mathematically equivalent to pushing on a quarter-line, and it is the reason Asian handicap pricing is preferred by sharp soccer bettors: the granularity reduces the binary push problem and lets the book and bettor share risk more smoothly.
Tax and accounting implications
In US federal tax law, gambling winnings are taxable as gross winnings — meaning each winning bet is taxable income before losses are deducted (subject to the session-based methodology accepted by the IRS for casino and sportsbook play). A push is not a winning bet; it is a refund. No taxable event occurs. The stake comes back, the book reports no win, and the bettor records no income.
State-level treatment is mostly identical, with the operator-side wrinkle that handle (total dollars wagered) typically includes pushes for the calculation of state sportsbook taxes. New Jersey's Division of Gaming Enforcement reports handle-inclusive of pushes; the tax is paid by the operator on hold, not by the bettor. Recreational bettors who file Schedule A (US itemised deductions) can ignore pushes entirely. Bettors filing as professionals (Schedule C) should still log pushes for personal performance tracking, but they do not appear on the tax return.
The psychology of the push
Most bettors describe a push as 'worse than a loss.' The reason is psychological, not mathematical. A push ties up capital for the duration of the game, denies the dopamine of a win, and reveals — in retrospect — that the half-point purchase (or the decision to take the unhooked line) was a coin-flip the bettor lost. Behavioural studies on prospect theory (Kahneman & Tversky) predict exactly this: humans weight a missed win more heavily than they weight a recovered loss, and a push is structurally a missed win.
The corrective frame is to remember that a push is a recovered stake — the same as choosing not to place the bet, with one game of variance reduced and the bankroll fully intact. Treat pushes as neutral events in your win-rate accounting, log them separately from wins and losses, and resist the urge to 'make back' a push with an immediate revenge bet at worse pricing.
Buying off the number — when it's worth it
A common sharp question is whether to buy off the 3 in the NFL — pay 20-25 cents to move from -3 to -2.5, eliminating the push risk on the most common margin. The math is well-trodden. With ~14% of NFL games landing on a 3-point margin, buying the 3 converts roughly 7% of your bets (the half of pushes that would have gone the bettor's way if the line were a hair softer) from push to win, and the other 7% from loss-by-half-point to push. The expected-value swing from buying the 3 is approximately +1.5-2.0% per bet, which is significant.
The cost — 20-25 cents — must be lower than the EV gain. At -120 (a 10-cent buy), the math is positive; at -130 (a 20-cent buy), it is roughly break-even; at -135 or worse, the book has priced the half-point above its expected value. Bettors should check the price of the 3 across multiple books and only buy when the cost is below the EV gain. The same logic applies to buying off the 7 (smaller frequency, smaller half-point premium).
Practical checklist for pushes
- Check the half-point cost before buying off the 3 or 7 in the NFL. The EV gain is real but only at the right price.
- Read parlay rules at every book you use. The leg-drop rule is standard at major US books but not universal.
- Avoid whole-number totals in MLB unless you have explicitly priced the push risk into the bet.
- Log pushes separately in your bankroll tracker. They are not wins or losses but they affect units deployed.
- Watch for SGP push rules, which vary by operator and can void an entire ticket if a correlated leg ties.
- Remember the psychology: a push is a recovered stake, not a missed win. Do not chase pushes with revenge bets.
Sources & further reading
- Levitt, Steven D. "Why are gambling markets organised so differently from financial markets?" Economic Journal, 2004 — bookmaker pricing of asymmetric risk and the role of key numbers.
- Snowberg, Erik & Wolfers, Justin. "Explaining the Favorite-Longshot Bias: Is It Risk-Love or Misperceptions?" Journal of Political Economy, 2010 — bettor behaviour around perceived certainty.
- Kahneman, Daniel & Tversky, Amos. "Prospect Theory: An Analysis of Decision under Risk." Econometrica, 1979 — the psychological weighting of pushes vs. wins.
- New Jersey Division of Gaming Enforcement — Monthly Sports Wagering Revenue Reports, 2018-2025, for handle and hold accounting including pushes.
- Pinnacle Betting Resources & The Action Network historical NFL margin-of-victory studies — push frequency at 3 and 7 in the modern era.