Miami Dolphins at San Francisco 49ers. Lean: San Francisco 49ers -13.5. Lean: Under 45.5.

Spread.I take San Francisco -13.5 at FanDuel (-104). Edge Labs makes this -10.2, but its roster-blind prior understates Miami’s offensive teardown and San Francisco’s changed personnel. My judgmental projection is approximately 29-14, making this a small favorite lean. Willis taking five sacks in his opener reinforces the protection concern; Purdy took none against Los Angeles. Evans and Samuel give San Francisco receiving options the rating does not directly recognize. Those facts are documented below. The market has already charged substantially for this mismatch: DraftKings moved from -10.5 to -13.5 in the dossier, and FanDuel independently shows -12.5 to -13.5. Reported favorite-heavy betting provides no additional edge. Miami’s strongest case is the inflated price, Willis’s running ability and late scoring against a team protecting a lead. San Francisco’s Melbourne return also makes its extra rest less straightforward. Missing Week 2 injury designations and a projection only modestly above market keep this well below strong-position confidence.

Total.Under 45.5 at FanDuel (-105). Applying the specified 1.2-point historical correction takes 45.99 to 44.79, only 0.71 below market. My personnel-based judgment is nearer 43: Miami’s inexperienced passing offense and protection threaten sustained drives, while a San Francisco lead could encourage clock-consuming rushing. That pace expectation is an inference; I could not verify current neutral-situation pace. The over has credible routes: favorable weather, Evans improving scoring opportunities, Achane and Willis creating explosive runs, and turnovers producing short fields. San Francisco also needed takeaways and a goal-line stop to hold Los Angeles to seven, so repeating that defensive score is too optimistic. The dossier’s one-point DraftKings decline has already absorbed some under value. I still prefer Miami scoring below the model’s expectation, with moderate conviction.

Where our number looks wrong.The -10.2 spread likely understates the personnel gap, especially Miami losing both Hill and Waddle. Conversely, stretching the raw eight-point margin through calibration is not evidence of predictive accuracy. The totals model's Miami projection of 19.56 looks optimistic for this receiving group and protection, while SF's 25.49 looks conservative. Its zero QB-disruption adjustment also deserves an audit given the verified quarterback change. The injury scan omits important reserve-list context and relabels Week 1 information with a recent retrieval date.

Spread.Our number (-10.2) trails the market (-13/-13.5) and would lean Miami, but our own model warns it is still mostly the preseason prior and reads roster teardowns only through flags. The flags fire correctly here: Miami has a new HC (Jeff Hafley) and a first-year starter (Malik Willis), and Week 1 confirmed the downgrade is real, not modeled noise. Willis was sacked five times behind a leaky line and Miami managed 13 points at Las Vegas (NBC Sports/local10 9/13), while SF dismantled the Rams 27-7 and held Stafford to 155 yards (Covers 9/15). The line has only climbed since it opened (DK -10.5 8/30 to -13.5, FanDuel -12.5 to -13.5) with no pushback, so sharp and public agree. SF is home, healthy (Kittle, Bosa cleared and played Wk1 per ESPN/CBS 9/11), and rested 10 days versus Miami's 7. The Dolphins offense and pass protection are the worst pairing on the board against Bosa. Two-TD road covers and Shanahan's mercy pulls cap this below full size, but SF is the side.

Total.Model (45.99) sits right on the market (45.5) and our totals model is worse than the market and runs ~1.2 points high, so it offers no real over signal. The qualitative case favors under: a projected two-touchdown SF win invites a second-half game script that shortens the game and pulls starters. SF's defense is elite (Rams to 7, Stafford 155 yds, Covers 9/15) and Miami's offense is punchless and can't protect Willis (5 sacks Wk1), so the Dolphins are unlikely to trade scores. Weather at Levi's is warm, dry, low wind for a September afternoon (non-factor, slightly scoring-positive). The live risk is Willis's mobility manufacturing garbage points and SF simply outscoring a low number. That keeps this a lean, not a lock.

Where our number looks wrong.The -10.2 is preseason-anchored and undervalues SF. It cannot see that Miami's Week 1 offense was broken (5 sacks, 13 points) or that SF's stars are healthy and dominant, so it prices this closer to a coin flip than the field does. Our -3 gap 'reading Miami' is a stale-prior artifact, not a real edge; the whole market moving to SF is the tell.

Spread.Our number (-10.2) sits 3 points shorter than the market (-13.2/-13.5), and the model's own known-weaknesses note it is still mostly the preseason prior after one game and has zero proven ATS edge historically, so I'm not leaning on the model itself, I'm leaning on how the number got here. The line opened SF -10.5 (DraftKings, 8/30) and has since been bid up to -13.5, the largest spread on the Week 2 board, almost entirely off SF's 27-7 Thursday win over a mediocre Rams team and Miami's 27-13 loss at Vegas where Malik Willis was sacked five times. That's real signal, but 13.5 is a two-score number laid against a rested, home, healthy favorite in Week 2, exactly the spot where garbage-time scripts produce backdoor covers even in a real talent mismatch. Miami's rebuild is ugly, but SF doesn't need to blow this out by three scores to make Dolphins +13.5 the right side; it just needs the fourth quarter to go the way these games usually do.

Total.Our totals model's 45.99 is barely above the market's 45.5, but its own documented bias runs about 1.2 points hot versus the closing number, which nets out to roughly 44.8, already an under lean before adding context. Miami managed only 13 points at Vegas behind a line that allowed five sacks, and now travels to face a 49ers front that just held the Rams to 7. San Francisco's offense is fine, but this game likely turns into a blowout script by the second half: once SF is up two-plus scores against a rebuilding roster with a backup-caliber QB, expect run-clock offense and fewer competitive-game possessions for Miami, which caps the total more than it inflates it. No dome/wind concerns are in play (outdoor grass, mild late-September Bay Area weather, no material forecast risk found this far out).

Where our number looks wrong.The rating still reflects mostly last year's records plus one 2026 game, so it hasn't fully absorbed how thin Miami's offensive line looked in five sacks of Willis, nor how much a fully healthy Bosa/Warner front raises SF's real ceiling versus 2025 when both missed most of the season, both cut against the Dolphins side I'm taking. I'm still on Miami because the market has already priced that news into a 13.5-point number, and the model's own documented history (47.6% ATS holdout, no edge at any gap size) means I trust the standard 'large road dog vs. a not-yet-elite favorite' pattern more than either the model's -10.2 or blind faith in the market's -13.5.