Patriots vs. Raiders — Week 5 Preview
Raiders at Patriots: The Model Says New England, the Injury Report Says Otherwise
Las Vegas rolls into Foxborough at 3-1 with the AFC's second wild-card spot in hand. New England sits at 2-2, fresh off a 29-26 road theft in Buffalo, and the market has installed the Patriots as -3.5 home favorites with the total parked at 44.5. Neither number has moved off the open. That's the first thing worth noting: books posted this line and nobody has argued with them yet.
What makes this game genuinely interesting is the collision between a model that likes New England and a roster report that screams Las Vegas.
Las Vegas Is a Passing Team Now, and It's Working
Kirk Cousins has thrown for 1,026 yards, 11 touchdowns and 4 interceptions with a 99.5 passer rating. The Raiders are averaging 256.5 passing yards per game on 36 attempts, and they're scoring 28.8 points per game — more than seven points per week better than the Patriots. Cousins has been sacked just 1.3 times per game, which means protection is holding and he's getting to his second and third reads.
The run game is the weak link. Las Vegas manages only 93.0 rushing yards per game, and Ashton Jeanty is grinding out 3.4 yards per carry. This is a one-dimensional offense, but it's a very good one-dimensional offense, and the dimension it's missing is the one New England is best equipped to defend.
Defensively, the Raiders are giving up 21.0 points per game, generating 3.0 sacks and 2.3 takeaways per contest, and sitting at a +1.3 turnover margin. That last number matters more than anything else in this preview.
New England's Offense Has a Drake Maye Problem
Maye's raw line is ugly: 854 yards, 4 touchdowns, 7 interceptions, 62.4% completion rate, 71.0 rating. He has thrown at least one pick in every game, including three in the opener against Seattle. His legs are a real weapon — 136 rushing yards on the season, 54 of them in the Buffalo win — but the Patriots are turning the ball over 2.3 times per game and carry a -1.0 turnover differential.
That's a losing formula against a team that takes the ball away 2.3 times a week. When you pair New England's 16.3 points per game with that giveaway rate, you're asking a defense to be perfect for sixty minutes. Mike Vrabel's unit has been good — 19.3 points allowed — but not that good.
The Matchup That Decides This Game
New England ruled out six active-roster players, and the list is brutal: corners Christian Gonzalez, Carlton Davis III, Karon Prunty and Channing Canada, fullback Reggie Gilliam, and receiver Mack Hollins. Both starting perimeter corners are gone, plus two more bodies from the same position group. Wednesday's practice report wasn't any kinder.
On the other side, Las Vegas runs a two-tight-end-heavy passing game with Brock Bowers (4.0 catches, 50.5 yards per game) and Michael Mayer (5.0 catches, 42.2 yards). Tre Tucker is averaging 58.2 yards on 15.5 yards per catch. That's a receiving corps built to attack the middle of the field and stress linebackers and nickel defenders — exactly the area a cornerback-depleted defense has to patch with help. If New England rolls coverage inside, Tucker gets one-on-one outside. If they don't, Bowers and Mayer eat.
The one counter: the Raiders' run defense has been leaky enough that Josh McDaniels may lean on Rhamondre Stevenson (40.8 yards per game) and TreVeyon Henderson (35.2) to shorten the game and keep Cousins on the sideline.
What the Model Says
The projection is Las Vegas 24, New England 28 — Patriots by four, which covers the -3.5 by the width of a shoelace. The model's ATS and moneyline picks are both New England, at a confidence of 0.5367. Read that number honestly: it's barely a coin flip. The model scores the game 52 points, comfortably over the 44.5 total, but no over/under probability was published, so I'm not going to pretend the model is banging the over drum. Treat the total as a lean, not a play.
Betting Angles
Start with the moneyline/spread gap. New England at -196 implies roughly a 66% win probability, yet the spread sits at just -3.5. That combination tells you the book expects a low-scoring game decided by a field goal — consistent with a 44.5 total that hasn't budged.
The frozen lines are informative. When a number opens at -3.5 and stays there all week while a team loses four cornerbacks and a starting receiver, one of two things is happening: the market already priced the injuries, or the books are happy to take balanced action and let the sharps decide. Given that the Raiders are only +164 on the moneyline, I lean toward the former — this line was built with the New England attrition baked in.
But here's the angle that matters most: Las Vegas is +1.3 in turnover margin and New England is -1.0. That's a 2.3-turnover-per-game swing, and turnover margin is the single most predictive in-week stat in the sport. Give Cousins 2.3 extra possessions on short fields and 28.8 points per game becomes a real number against a secondary missing its top four corners.
Situational Factors
Both teams are on normal rest, and this is a conference game, not a division one — no divisional weirdness to account for. Gillette Stadium is outdoors with an artificial surface, and the forecast is 61°F with 6 mph wind and mostly cloudy skies. No weather advantage for either side.
The head-to-head is genuinely useful here: Las Vegas has won three of the last five meetings, including a 20-13 win at Gillette Stadium in Week 1 of 2025. This isn't a venue where the Raiders historically melt.
Final Verdict
The model likes New England, and I understand the logic: home field, a defense allowing 19.3, a coaching staff coming off a road win in Buffalo, and an offense that can control the clock on the ground. But 0.5367 confidence is noise. When a near-coin-flip projection collides with four cornerbacks on the injury report and a 2.3-turnover-per-game margin gap in the underdog's favor, the disciplined play is the other side.
The pick: Las Vegas +3.5. Cousins is playing the best football of his career by rating, the Raiders take the ball away, and New England's depleted secondary walks into the worst possible matchup for it. If you're backing the model's side, do it on the moneyline at -196 rather than laying 3.5 — a four-point win is the model's margin, and it clears the number by half a point. That's not a margin worth trusting with real money.
Confidence is low. This is a lean, not a hammer.