NFL Preseason Week 3: The Two-Market Problem in Today's Totals
Coaches call it the dress rehearsal. Bettors treat it like a real game. Neither framing tells you where the edge in the total lives.
Here is the real structure of Week 3 preseason totals: every game gets priced somewhere between 36 and 40. Books use the same band regardless of whether the starting quarterback is playing a half, playing a series, or watching from a hotel in athletic shorts. The market treats "dress rehearsal week" as a single entity. It is not. It is two completely different betting markets dressed in the same jersey.
Understanding which type of game you are looking at today, before the opening whistle, is how you find value in a market most bettors dismiss as unreadable.
The Two Types of Week 3 Games
Every Week 3 game falls into one of two categories:
Type A: Starters play a quarter or more. The head coach wants a real tune-up. The starting offense gets live reps. A capable quarterback moves the ball against a real defense for 12-20 plays. These games produce scoring.
Type B: Backups run all night. The head coach has already made his depth chart decisions and wants no injury risk. The starting quarterback either watches from the sideline or plays a single token series. These games look like glorified practice squads fighting for roster spots.
Books price both game types within the same 36-40 range. The actual scoring output of these two game types diverges by 15-20 points on average.
The 2025 preseason illustrated this split in consecutive days.
What 2025 Week 3 Showed
On August 22, 2025, Patrick Mahomes played a quarter for Kansas City. He went 8-for-13 for 143 yards and a touchdown. The Chiefs built a 20-3 lead in about 15 minutes of real game action before Andy Reid pulled his starters. The Bears came back to win 29-27. Combined score: 56.
On August 23, 2025, Jacksonville played Miami without Trevor Lawrence, Travis Etienne, or any significant Jaguars starter. Tua Tagovailoa played one series for Miami. The Dolphins won 14-6. Combined score: 20. The total had been set at 38.5.
The same total range. A 36-point gap in actual scoring. That is the two-market problem.
Books cannot perfectly price this split because coach decisions on Week 3 starter involvement often come out Wednesday or Thursday, get updated Friday, and sometimes change again Saturday morning. By the time sharp money acts, the line has already moved. In most games, it has not moved enough.
The Academic Foundation
This is not a new phenomenon. Davis and Krieger (2016) studied NFL preseason betting markets across 1995-2014 and found systematic mispricing in preseason point spreads, published in Applied Economics (Vol. 49, No. 12). Their core finding: preseason spreads are too large, creating a profitable underdog strategy. The mispricing is more pronounced as spreads get bigger.
One specific finding is directly relevant: the underdog advantage disappears in Week 3, when clubs "expel a higher level of effort." Teams competing for roster spots in the final preseason week play harder than in Weeks 1-2. Coaches give backup players more competitive reps. This changes the spread dynamic. It also affects totals. Games with legitimate depth-chart competition at every position produce more real scoring than Week 1 or Week 2 games when everyone knows their role.
The implication for totals: Week 3 games with confirmed starter absences should trade at a discount to games where starters play. The market rarely applies this discount with precision.
The Historical Totals Picture
An analysis of NFL preseason totals from 2002 through 2025 shows a 57.2% Under rate in games with totals set between 39 and 41.5. Across the full preseason, the Under hits roughly 56-60% depending on the total range. The highest-set games show the most consistent Under lean.
The "preseason unders hit 80%" claim is marketing. It has no sample size behind it. The real figure from 23 years of preseason data is closer to 57%, which is a genuine edge, not a guarantee.
A 57% win rate at -110 is worth 3.0 units per 100 bets. Over a season, if you identify six Week 3 games with the specific conditions below, you are playing a positive-expectation market, not guessing.
The 2026 Week 3 Starter Split
Here is where this year's dress rehearsal week falls after Friday and Saturday morning's lineup reporting:
| Game (Sat 8/23) | QB1 Status | QB2 Status | Game Type |
|---|---|---|---|
| Rams vs. Browns | Stafford: Sitting | Shedeur Sanders: First half | Type B/Hybrid |
| Seahawks vs. Packers | TBD | TBD | TBD |
| Jaguars vs. Dolphins | Lawrence: Likely sitting | Tua: Limited | Type B |
| Raiders vs. Cardinals | TBD | TBD | TBD |
Confirmed sitting from Friday reports: Josh Allen (Bills), Kyler Murray (Vikings, with JJ McCarthy starting), Lamar Jackson (Ravens), Matthew Stafford (Rams, described as having the full day off). Most Cowboys starters are also out. The Falcons game Thursday (8/22) had Jack Strand running Atlanta's offense with the Cowboys in a similar position.
The Browns game is the one nuance in the current slate. Cleveland has Shedeur Sanders, their rookie starter, taking the first half, with Deshaun Watson completing the second half. This is genuinely unusual. The Browns are giving Watson live work while simultaneously evaluating Sanders' ceiling. Two quarterbacks getting real reps means more first-team offensive execution than a pure backup game, but neither quarterback runs the offense for a full game at full speed. Treat this as a hybrid.
Three Questions Before You Bet a Week 3 Total
The Line Movement Signal
Sharp money in preseason Week 3 moves on starter confirmation, not on team talent or recent form. Those factors are noise in a game where the starting quarterback is on a stationary bike.
When a total opens at 38 and moves to 35.5 by kickoff, the news driving it is almost always a confirmed starter absence. When a total stays at 37.5 after the same news, the market either missed the announcement or is not adjusting enough. These are the totals worth targeting.
Check the opening line against the current line at DraftKings or FanDuel for each Saturday game. A game where two starters are confirmed out but the line has barely moved from opening is the specific setup you are looking for.
What the Market Gets Right (and Where It Stays Wrong)
Books are not oblivious to this dynamic. They do price Week 3 totals lower than regular-season totals by design. A 38-point total in the preseason would be a 44 in the regular season.
Where they consistently underadjust: games with confirmed double-starter absences still trade too high relative to games where starters play a half. The Jaguars-Dolphins 2025 game at 38.5 with both teams' starters sitting is the clearest example. The book used the same total it would have given a game with one or two limited starter appearances. The resulting miss was 18 points.
The books also face an information timing problem. Starter status in Week 3 shifts late in the week. By Friday evening, some teams still have not confirmed plans. Books open totals mid-week with uncertainty baked in. By Saturday morning, when the picture is clear, they have moved lines, but the sharpest line movement happens in the 12-hour window after a coach confirms a starter is sitting. Retail bettors are often still looking at the original number.
The Specific Takeaway for Today
Week 3 of the 2026 NFL preseason is running with more confirmed starter absences than typical for this week. Allen, Murray, Jackson, and Stafford are out. On the other side of those games, their opponents are also running backup quarterbacks.
For each game on today's board:
- Confirm whether both starters are out using current beat reporter reports.
- Check opening total against current total at your book of choice.
- Look for totals in the 37-40 range where both starters are confirmed absent. Those have the sharpest structural Under lean.
- Avoid games where one team's starter is confirmed playing even a quarter. One capable quarterback is enough to push a total.
The dress rehearsal narrative encourages people to bet these games like mini-real-games. They are not. For the majority of this week's slate, you are watching backups fighting for roster spots on the final cut weekend. The market knows this in aggregate. In specific games where the starter confirmation landed late or the line has not moved, it does not know it quite well enough.
That gap is where you find the edge.
Sources
- Davis, J. & Krieger, G. (2016). "Preseason bias in the NFL and NBA betting markets." Applied Economics, 49(12). Taylor & Francis | Semantic Scholar
- NFL Preseason totals trends (2002-2025), 57.2% Under on totals 39-41.5. Tony's Picks
- Dolphins 14, Jaguars 6 (August 23, 2025). Total was set at 38.5. Trevor Lawrence and Jaguars starters did not play. Tua Tagovailoa played one series. ESPN | Jaguars.com
- Bears 29, Chiefs 27 (August 22, 2025). Mahomes went 8-for-13, 143 yards, 1 TD before starters were pulled. CBS Sports
- 2026 NFL Preseason Week 3 starter lineup updates: Allen, Murray, Jackson, Stafford all out. Sanders playing first half for Cleveland. Covers.com | Yahoo Sports