The 15-Minute Problem: Why NFL Second-Half Lines Stay Soft
Live betting is no longer a side market. In 2025, in-play wagering captured 62.35% of all online sports betting revenue in the United States. NFL is the sport driving the largest share of that handle. And the most chaotic, most exploitable window inside NFL live betting opens every Sunday at halftime and closes 15 minutes later.
That window is where books make their worst pricing errors. Understanding why gives you a specific, research-backed framework for where to look, what to bet, and what to ignore.
What Happens at Halftime
When a first half ends, sportsbooks face a coordination problem. Multiple games are finishing simultaneously. Traders need to set a second-half spread, a second-half moneyline, and a second-half total for each contest, all within roughly 15 minutes. They do this while absorbing a surge of sharp bets the moment lines go up.
Sports Insights documented the structural problem: books must price every game at once during the early window, which means oddsmakers are spread thin and under time pressure they do not face when setting pre-game lines. A pre-game line gets days of refinement. A 2H line gets minutes.
The books also have a liability management problem. A second-half line is not set only to reflect what the book thinks the fair line is. It is also shaped by where the book is exposed on the full-game side. If a book is overloaded on one side of a full-game spread, the 2H line will shade to reduce that liability, not to maximize accuracy. That is a deliberate distortion. Deliberate distortions are findable.
What Research Shows About In-Play Bettor Behavior
The behavioral error on the bettor side runs in the opposite direction from the structural error on the book side. And together, they create a predictable pricing gap.
A 2025 paper in Economic Inquiry by Ötting, Deutscher, Singleton, and De Angelis examined second-by-second in-play betting data across 1,224 professional football matches. The researchers had access to full volume and price data from a major bookmaker, which is a dataset most academic work never gets. Their finding on momentum was direct: after an equalizing goal (a clear momentum event bettors see on the screen), bet volume spiked sharply toward the team that scored last. Bettors chased the momentum signal in their wagering behavior.
But the bookmaker's prices did not move in the same direction. The paper found no evidence that bookmaker odds reflected the momentum created by scoring events. The market itself treated the equalizing goal as roughly neutral to the remaining match probabilities. The bettors moved. The efficient price did not.
That gap, between where bettors push money and where true probabilities sit, is the behavioral edge. Bettors believe in momentum. The underlying data does not support it. The book, priced correctly against the outcome data, takes that money.
The Ötting research covered soccer markets specifically. But the behavioral mechanism, chasing recent scoring events in live betting, operates identically in NFL in-play markets. The bettor sees a team score to go up 21-3, assumes the momentum is real and sustainable, and bets the favorite to extend the lead in the 2H. The book sets a 2H spread that reflects expected score regression. Both err in different directions.
A second 2025 paper in Economic Inquiry, by Fischer and Schmal, studied how betting markets price information from player absence announcements. Their framework confirms the basic mechanism: in-play and near-real-time betting markets incorporate information through price and volume dynamics, but the two signals frequently diverge. Volume chases noise. Prices (when set by disciplined bookmakers) lag toward fundamentals.
A third 2025 paper, by Fodor, Patterson, and Shank in Economics Letters, found that NFL sportsbooks themselves carry an anchoring bias, continuing to incorporate pre-season Super Bowl odds into their closing lines throughout the regular season. That means books are not perfectly calibrated either. They carry systematic biases of their own, which create additional pricing gaps that persist week to week.
The Two Patterns That Have Persisted
The structural chaos of halftime and the behavioral errors of in-play bettors combine to produce two testable patterns. Both have held across large samples.
Second-Half Underdogs of 7 or More Points
When a team is getting 7 or more points on the 2H spread, that typically means the first half ended with a significant score differential. The public has watched one team dominate for 30 minutes. The narrative momentum is entirely on the favorite's side. Casual live bettors load up on the chalk.
The data from Bet Labs tells the story. Since 2005, teams receiving 7 or more points on the second-half spread have gone 152-110-26, covering 58.0% of the time. That is well above the 52.38% break-even on standard -110 juice. The edge holds regardless of home-away status and regardless of divisional opponent status.
The mechanism: books set the 2H spread reflecting some correction for first-half variance, but they also shade toward liability management on the full-game side. Bettors push the chalk because they trust what they watched for 30 minutes. The combination systematically leaves the dog with a number that is too generous.
Teams Leading by 14 or More at Halftime
The flip side of the same data is equally persistent. Since 2003, teams leading by 14 or more points at halftime have gone 425-354-17 against the second-half spread, covering 55% of the time. When the lead reaches 17 or more, the record is 315-244-13, good for 56.4% ATS.
This pattern runs counter to the casual-bettor assumption that big leads always regress. The common belief is that trailing teams make halftime adjustments, dominant teams take their foot off the gas, and second-half performance reverts toward the mean. The data says the opposite. Teams that dominate a first half tend to continue dominating the second half at rates that beat the 2H spread.
Why does the book mis-set this? Two reasons. First, the book is setting the 2H spread quickly, with limited time to model 2H-specific performance tendencies. Second, the book is worried about liability on the full game. If they posted a generous 2H number on the leading team, they risk getting middled if the game stays close and the full-game spread also loses.
The result is a 2H spread that asks you to lay a number on the leading team that is smaller than their true 2H edge. The trailing team gets a number that is larger than they deserve. And bettors who trust the halftime adjustment narrative push more money onto the trailing team, reinforcing the mis-pricing.
How to Use This Framework
There are four rules that keep this edge intact.
First, time your bet. The 15-minute window opens and closes fast. You need to be ready at halftime, not hunting for the line five minutes in. Sharp money arrives immediately when 2H lines post. The worst numbers are available right at the opening. Get your bet placed in the first three to four minutes or skip it.
Second, ignore the score as a narrative. The Ötting research confirms that the actual outcome data does not support in-play momentum. A team scoring to go up 21-0 is not "locked in." They have a 21-point lead built over 30 minutes of football, some portion of which is sustainable talent differential and some portion of which is variance. The book's 2H price is trying to disentangle those two. The casual bettor is not.
Third, focus on the spread, not the moneyline. The 2H moneyline is the most heavily liability-managed number during halftime. A book is not going to offer you a 2H moneyline on the trailing team at anything resembling fair value when the full-game moneyline is also on the board. The 2H spread is where the mis-pricing sits. That is where the Bet Labs data tracked the 58% ATS figure for 7+ point 2H underdogs.
Fourth, know the sample sizes before you size your bets. Since 2005, the 7+ point 2H underdog data covers 288 completed games. That is a credible sample but not a massive one. The leading team data since 2003 covers nearly 800 games. The larger the sample, the more confident you are the pattern is structural rather than noise. Size accordingly.
The Limits of the Edge
Two things erode this framework. The first is market maturity. As live betting grows from 62% to higher shares of total handle, more sharp money will arrive at halftime. Books will have more action to price against, and more data to calibrate their 2H models. The 58% ATS figure for 2H dogs from 2005 through 2024 will compress as the market deepens.
The second is your own limits. Sportsbooks track in-play betting activity closely. A sustained pattern of profitable 2H wagering, particularly at the open of the window, marks accounts for reduced limits or restrictions faster than most bettors expect. The edge is real, but it is not bottomless. Spread your action across books. Use DraftKings, FanDuel, and BetMGM to distribute bets so no single book builds a profile on your halftime behavior.
The 15-minute problem is a structural feature of how NFL second-half lines get built, confirmed by academic research on in-play bettor behavior, and backed by two decades of cover data. Books are under time pressure, bettors are chasing momentum that the data says is not there, and the resulting prices are systematically off in predictable directions. The window is short. But it opens every week.
Citations
- Ötting, M., Deutscher, C., Singleton, C., & De Angelis, L. (2025). Betting on momentum in contests. Economic Inquiry, 63(4), 1066–1089. doi:10.1111/ecin.70008
- Szalkowski, I., & Nelson, R. (2012). The performance of betting lines for predicting the outcome of NFL games. arXiv:1211.4000. arxiv.org/abs/1211.4000
- Action Network / Bet Labs. NFL second-half betting: 2H underdogs and halftime leaders, ATS records since 2003–2005. actionnetwork.com
- Sports Insights. How to bet NFL 2nd half lines. sportsinsights.com
- GlobeNewswire. American football betting market analysis report 2026. Published February 18, 2026. globenewswire.com
- Fodor, A., Patterson, F., & Shank, C. (2025). Anchoring bias in the NFL gambling market. Economics Letters, 250. sciencedirect.com
- Fischer, K., & Schmal, W. B. (2025). Pricing in response to new information: The case of betting markets. Economic Inquiry, 63(1), 236–264. doi:10.1111/ecin.13258