Denver Broncos at Kansas City Chiefs. Lean: Denver Broncos +2.5. Pick: Under 43.5.

Spread.Take Denver +3, currently displayed at -120 by DraftKings. Our model makes Kansas City +0.3 versus the market's -3, but its poor ATS validation means that 3.3-point disagreement is not a demonstrated edge. The stronger argument is protection: Simmons remains in rehabilitation and Benson is the likely replacement, creating a plausible advantage for Denver's edge rush against a returning Mahomes. That is an inference, not a confirmed performance limitation. Mahomes' expected availability supports Kansas City's favoritism, and Arrowhead plus coaching continuity makes a Chiefs rebound credible. These injury concerns are public already: DraftKings shows 76% of spread handle and 70% of tickets on Denver while retaining Chiefs -3. That does not establish sharp action or a missed injury adjustment. I prefer the key-number protection, but the expensive price and incomplete official injury information keep this a measured lean rather than a strong position.

Total.Take under 43.5 at DraftKings' freshly displayed -110. The bias-adjusted model reference is 41.11, a 2.39-point cushion, but its worse-than-market error limits conviction. Kansas City's protection uncertainty supports fewer sustained drives and potentially more conservative play selection; neither reduced pace nor restricted quarterback mobility is verified. The over deserves respect: Denver's Waddle/Sutton receiving options, Mahomes' expected return and Bieniemy's return as coordinator provide offensive upside. The dossier's DraftKings total rose from 42.5 to 43.5, genuine counterevidence to an under, although the cause is unverified. The preliminary forecast suggests modest wind and possible rain, not conditions warranting a large weather deduction. I favor the under narrowly because protection risk matters more than speculative offensive improvement.

Where our number looks wrong.Returning-QB flags conceal recovery risk, and the injury-blind ratings cannot evaluate Kansas City's current left-tackle problem. Conversely, treating Denver's slight modeled superiority as precise ignores the model's losing ATS record. The totals model also cannot distinguish quarterback continuity from full health. Applying its stated 1.2-point historical overprediction adjustment lowers 42.31 to 41.11, but that correction is not independently validated for this matchup.

Spread.Our v3 number makes this Denver by 0.3; the market says Kansas City by 2.5-3, a three-point gap on the Broncos. I am fading our own number. It is a frozen preseason fit with no demonstrated ATS edge (47.6% held out), and it is anchored to a 2025 Chiefs season that misrepresents them: 6-11 with a +2.0 per-game point differential is a luck outlier, not a 6-11 roster. The live news is all Denver's. Mahomes is nine months off a Dec. 15 ACL/LCL repair and was only a 'pretty good chance' to start as of 9/8, and left tackle Josh Simmons (back) is likely out with an undrafted rookie behind him. That is precisely why the price is just 2.5-3. What moves me is what has not happened: from the 8/30 open through 9/9 not one book in our capture has moved off -2.5/-3 while 66-73% of tickets and 71% of DraftKings handle sit on Denver. Books absorbing that much one-way public money without giving a half point is the strongest signal on the board, and it says this number is not soft.

Total.Our totals model wants 42.3, and after the documented +1.2 preseason bias closer to 41 - an under lean of roughly three points. I will not lean hard on it (holdout MAE 10.46 versus the market's 10.19), but the fundamentals agree. Denver's 2025 defensive EPA of -0.089 is the best number in this dossier by a wide margin, and it draws a quarterback nine months off an ACL/LCL repair behind a likely undrafted rookie left tackle. Division openers are the league's most familiar matchups and Week 1 offences are usually the ones behind. Weather is neutral to mildly under-friendly: NWS has 85F falling to the mid-70s, SSW wind 9-10 mph and a 46% thunderstorm chance at kickoff.

Where our number looks wrong.Two places. First, rating_v3 takes Kansas City's 6-11 2025 record largely at face value while their point differential was +2.0 per game - a team that unlucky regresses hard, and the frozen preseason fit cannot see that the market has already repriced them. Second, the model is blind to injuries by its own admission and the injuries table is empty for the NFL, so it cannot tell that the news flow here (Mahomes' knee, Simmons out) runs entirely in Denver's favour - which means our -3 'gap' is not new information, it is the same information the market used to set 2.5 instead of 5. The calibration map also nudges toward Denver on a rating built from a 14-3 season that is itself likely a positive outlier (+5.3 point differential does not usually support 14-3). Given the stated 47.6% held-out ATS, the honest reading is that this number carries no independent weight against a market that has refused to move for ten days.

Spread.So the model's number isn't the case here - the fundamentals and fresh news are. Denver went 14-3 in 2025 with a top-tier defense (-0.089 EPA/play) while Kansas City collapsed to 6-11; Chiefs -2.5/-3 at home is a short number for that form gap, largely a Mahomes/Reid/home-field premium. That premium looks shakier this week: starting LT Josh Simmons (back) is 'in danger of missing' Monday and Mahomes himself is questionable with a knee, both per ESPN's injury report (dated ~9/8, accessed 9/9). Denver's pass rush facing a compromised or backup-protected Mahomes is an edge the market hasn't obviously priced - the number hasn't moved since the 8/30 open despite this news. Reported sharp money still favors Kansas City, which tempers conviction, but the O-line/QB health risk plus last year's form gap is enough to take the field.

Total.Our totals model sits under the market by 1.4 points (42.3 vs 43.7) even before applying its documented historical over-bias of about 1.2 points, which would push the model's bias-adjusted read closer to 41, a cleaner under signal than the raw output shows - though the model's own note flags that this under-lean is partly mechanical at a below-average total level, not purely a football read. Real-world support for the under: both defenses graded well in 2025 (Denver -0.089 EPA/play, Kansas City -0.027), this is a division rivalry (typically plays tighter than non-division games), and Kansas City's likely starting QB (knee) and starting LT (back) are both questionable heading into Monday night, per ESPN's injury report, adding offensive-execution risk. The total has crept up only marginally, from an open range of 42.5-43.5 to 43.5-44 now, not enough to override the defensive and injury case for staying under.