NFL Week 10 Preview: Nine Unanimous Picks and a Changing Market

Nine Week 10 winners have all five models behind them, led by Jacksonville at 69.0% and Buffalo at 68.6%. That group follows a 5–0 Week 9 and an 11–1 run across the last two weeks. Green Bay also has unanimous support, even as its market handicap has shrunk from 5.5 points to three. Tonight’s Giants lean is narrower, with one model choosing Washington and the total falling before kickoff.

This preview covers all 14 games, November 12–16, including the three primetime matchups. Forecasts were issued Tuesday, November 10, at 1:54:26 p.m. CST, trained through completed Week 9. They remain frozen. The latest line comparison uses DraftKings at 11:33:11 a.m. CST Thursday, a dated snapshot rather than a closing price. Official personnel and weather reports were checked at approximately 12:28 p.m. CST Thursday.

Week 9 ended 12–3 on winners, 7–8 on saved spreads and 7–8 on totals, with Bayesian going 14–1 on winners. The recap includes the full results and the improvement since the Week 4 rebuild. That progress gives us a useful starting point, but nine unanimous selections are not a promise of nine wins.

Green marks the predicted outright winner; opponents are neutral gray. Spread and total selections answer separate questions.

All fourteen frozen picks

All kickoffs are Central Standard Time. “Models” counts the five base winner models selecting the ensemble’s side. Their probabilities are weighted differently, so two models can still tip the aggregate toward a team. LAR means the Rams; LAC means the Chargers. Every spread and total percentage belongs to its displayed original benchmark, and these leans are not a refreshed list of qualifying bets.

San Francisco–Dallas is in Sunday’s late-afternoon window. Pittsburgh–Cincinnati is the Sunday night game. New England–Detroit supplies the early Sunday start. NFL Week 10 schedule.

Nine unanimous picks, with different levels of confidence

“Unanimous” means Elo, Logistic, XGBoost, Bayesian and Random Forest all choose the same winner. We include every 5/5 selection, with no minimum win-probability filter. The ensemble selects that same side in all nine games.

San Francisco’s 58.4% estimate is much closer to even than Jacksonville’s 69.0%, although both have five-model support. Agreement counts the selected sides; confidence here refers to the ensemble win probability. The models share information, so five votes are not five independent pieces of evidence.

The handicap adds another distinction. Jacksonville leads the winner board, but the saved spread estimate slightly prefers Tennessee +10.5 at 51.6%. The separate point model projects Jacksonville 25.39–Tennessee 22.72, a 2.67-point margin and 48.12 total. Jacksonville winning and Tennessee covering are compatible outcomes.

Buffalo’s point projection is 26.90–23.26 at the Jets, a 3.64-point margin against a current −8.5 handicap. Its separate 55.1% cover estimate applies to the original −8. The Rams are 67.4% outright at Arizona, but their −9.5 cover estimate is only 51.9%. Stronger winner support does not establish a similarly strong large-margin prediction.

The record behind the agreement

Across Weeks 5–9, all five models agreed on 32 completed games, going 23–9, or 71.9%. Week 10’s pending games are excluded. The latest two weeks account for 11–1, but Week 6’s 5–4 remains part of the same record.

Each point pools all 5/5 winner picks from Week 5 through the labeled week. Counts vary by week; there were no ties in this sample. This is straight-up winner accuracy, not spread performance or wagering returns.

As the recap’s probability breakdown showed, those historical picks averaged 65.0% predicted win probability. Their 71.9% result is an observed hit rate, not a calibrated probability to assign to this week’s nine selections. The chart captures an encouraging recent run and the losses that preceded it.

Three primetime previews

Thursday: Washington at NY Giants

Tonight, November 12, at 7:15 p.m. CST. Four models choose the Giants, with Random Forest choosing Washington. The aggregate is NYG 52.9%, much closer to even than any of this week’s unanimous picks. Our published Thursday winner record stands at 6–3 after Jacksonville’s Week 9 win.

The separate point model projects Giants 22.78–Washington 20.97, a 1.81-point margin and 43.75 total. The saved spread lean is Washington +2.5 at 52.6%, and the total lean is Over 41 at 52.2%. A one- or two-point Giants win would satisfy both the winner selection and Washington’s spread selection.

DraftKings holds NYG −2.5 across Tuesday, Wednesday and Thursday. The Giants’ moneyline moved −130 → −142 → −135, with Washington +110 → +120 → +114. The total fell 41.5 → 41.5 → 40.5. Thursday’s spread prices are NYG −2.5 (−108) and WAS +2.5 (−112); at 40.5, Over is −118 and Under −102.

Washington has ruled out Lucas Patrick (concussion) and Colson Yankoff (knee). Jayden Daniels practiced fully all three days. The Giants list Malik Harrison (groin) questionable; J.J. McCarthy and Cam Skattebo progressed from limited Monday to full Tuesday and Wednesday, with no game designation. Malik Nabers also has no game designation after full participation throughout. These are final injury-report designations; tonight’s complete inactive lists remain pending at midday. Washington report, Giants report.

The immediate question is whether the Giants can create separation or Washington keeps a close game inside the handicap. The point model is above the current scoring threshold, but the saved 52.2% belongs to Over 41, not Over 40.5. Rain in the venue forecast adds a ball-handling and kicking consideration, discussed below; it does not manually change the frozen percentage.

Sunday night: Pittsburgh at Cincinnati

November 15, at 7:20 p.m. CST. Cincinnati is the 53.9% winner lean, with four models on the Bengals and Elo on Pittsburgh. The original market selections are Pittsburgh +3.5 at 54.8% and Over 48.5 at 51.5%.

The point projection is Cincinnati 25.77–Pittsburgh 22.19, a 3.59-point margin and 47.96 total. Its margin sits just beyond the saved handicap, while the separately fitted spread model prefers Pittsburgh. These estimates express different views near the same boundary; they are not one coherent exact-score forecast.

DraftKings keeps CIN −3.5 throughout, with the total moving 48.5 → 49.5 → 49.5. Moneylines moved CIN −175 → −180 → −180 and PIT +145 → +150 → +150. Both Thursday spread sides are −110; Over 49.5 is −108, Under −112.

A three-point Bengals win would hit Cincinnati’s winner lean and Pittsburgh’s +3.5 selection. A 49-point game would go over the saved 48.5 and under the current 49.5. Those examples explain why the forecast’s exact line stays attached to its probability when the market changes.

The early personnel question is in the secondaries. Wednesday’s preliminary report lists Cincinnati safety Kyle Dugger (quadricep) as a non-participant, Bengals cornerback DJ Turner II (ankle) limited, and Pittsburgh safety DeShon Elliott (ankle) limited. Their eventual availability could affect passing opportunities, but these practice statuses are not final Sunday designations. Official Bengals report covering both teams.

Monday night: LA Chargers at Baltimore

November 16, at 7:15 p.m. CST. Baltimore is the primetime game with unanimous winner support, at 67.6%. The separate point projection is Baltimore 23.29–Chargers 18.26, a 5.03-point margin and 41.55 total.

The saved market leans are much less emphatic: Baltimore −5 at 50.1% and Over 38.5 at 50.2%. Five models choosing Baltimore to win does not turn either of those near-even market estimates into a strong cover or total signal.

DraftKings moved BAL −5.5 → −4.5 → −4.5, with the total unchanged at 38.5. Baltimore’s moneyline moved −218 → −218 → −225, while the Chargers moved +180 → +180 → +185. Current spread prices are BAL −4.5 (−108) and LAC +4.5 (−112); both total sides are −110.

A five-point Ravens win would lose Tuesday’s DraftKings −5.5, push the model’s saved −5 and cover Thursday’s −4.5. The viewing question is whether Baltimore can translate the winner agreement into enough margin. Its 50.1% saved cover probability remains attached to −5.

Monday’s final personnel picture remains pending at this Thursday check. The Ravens and Chargers reports will matter before kickoff, especially when the projected margin sits so close to the saved handicap.

Sunday’s other stories

Green Bay’s unanimous support meets a smaller handicap. The Packers are 61.5% to win against Minnesota, with a separate 24.57–18.52 point projection: GB by 6.05, totaling 43.09. DraftKings eased GB −5.5 → −3.5 → −3, while the total moved 40.5 → 42.5 → 42.5. The saved 53.6% cover estimate belongs to −5.5. A smaller available handicap changes the required margin, not the frozen probability.

Wednesday’s official report lists Josh Jacobs (finger) and Sean Rhyan (ankle) as non-participants for Green Bay. Minnesota lists Kyler Murray (hamstring) as a non-participant and Carson Wentz (ankle) limited. That puts backfield, protection and quarterback uncertainty alongside the contracting spread, but it does not establish a final absence or prove why the line moved. Packers report covering both teams.

Seattle’s market move is larger than its model agreement. The Seahawks moved −3.5 → −6.5 → −6.5, while the total fell 43.5 → 40.5 → 40.5. Seattle is a 62.0% winner lean, with Logistic dissenting for Las Vegas. The separate point model actually gives Las Vegas a tiny lead, 22.12–21.79, or 0.33 points. A 50.9% cover estimate at the original Seattle −3.5 cannot be reused at −6.5.

Miami–Indianapolis is the closest winner split. Elo, Bayesian and Random Forest choose Indianapolis; Logistic and XGBoost choose Miami. Their weighted probabilities put the ensemble on Miami at 51.6%, while the point model projects Indianapolis 22.81–Miami 21.11, a 1.70-point Colts margin and 43.91 total.

DraftKings moved IND −2.5 → −3.5 → −3.5, with the total falling 40.5 → 38.5 → 38.5. The saved Over 40.5 at 57.3% is this slate’s strongest aggregate total lean. It has not become a 57.3% estimate at the new 38.5 threshold. Carolina–New Orleans has another winner/market disagreement: the aggregate leans Carolina 55.2%, while Thursday’s book favors New Orleans −2.5.

Kansas City’s handicap reaches three. All five winner models select the Chiefs at 62.5%, while the saved spread model prefers Atlanta +3.5 at 56.2%. DraftKings moved KC −3.5 → −3.5 → −3, and the total 45.5 → 45.5 → 46.5. A three-point Kansas City win would push the new favorite handicap and cover the original Atlanta +3.5.

Detroit and San Francisco frame Sunday’s long viewing window. Detroit’s early game carries 62.4% winner support, but only 51.6% to cover the original −3. DraftKings is at −2.5. San Francisco’s afternoon winner lean is 58.4%, with a point projection of 49ers 29.74–Dallas 27.09, a 2.65-point margin and 56.83 total. The current handicap is SF −3.5; the saved spread estimate is 51.6% at −4. Both are unanimous selections with more modest margin cases.

The full Tuesday-to-Thursday line report

The table uses DraftKings at Tuesday, November 10, 10:56:29 a.m. → Wednesday, November 11, 11:40:44 a.m. → Thursday, November 12, 11:33:11 a.m. CST. Tuesday is our first saved project snapshot, not a verified sportsbook opener. These are observations, not closing lines.

Spreads name the market favorite. The model’s original benchmarks are a separate set of saved lines and can differ even from Tuesday’s DraftKings snapshot. For example, the forecast uses Over 41 for Washington–Giants, while Tuesday’s DraftKings total was 41.5. We preserve each source’s number rather than treating that difference as a later move.

Two spreads and two totals changed since Wednesday. Kansas City and Green Bay each eased half a point to −3; Kansas City–Atlanta’s total rose one point to 46.5, while Washington–Giants fell one to 40.5. Compared with Tuesday, seven spreads and eight totals changed.

What can sit behind a line move?

Injury information can affect both scoring and the expected margin. A missing quarterback or offensive lineman can reduce offensive expectations; a missing pass rusher or defensive back can help the opposing offense. Replacement quality, likely workload and what the market already anticipated matter. A lower total or smaller handicap does not identify the responsible player or report by itself.

Weather can affect passing, ball handling and kicking. The National Weather Service’s East Rutherford forecast, updated Thursday at 10:28 a.m. EST, calls for rain tonight, a 90% precipitation chance and northeast wind of 8–14 mph. Wet conditions and that breeze are practical considerations alongside Washington–Giants’ drop from 41.5 to 40.5. The forecast covers the night rather than a precise kickoff reading, and the snapshots do not establish that weather caused the move. NWS venue forecast.

Sportsbooks also respond to wagers, potential payouts and competitors’ prices. A move can change the scoring threshold, the handicap, the price attached to it, or several at once. Tonight’s Over 40.5 costs −118 while Under is −102, so the number alone does not describe the offer. These captures provide no bettor identities or betting-volume evidence that would justify labeling a move “sharp money.”

The confidence check

Tuesday’s latest monitoring receipt is GREEN, with 0.6205 log loss across Weeks 7–9. The wider ensemble winner record since Week 5 is 48–23, or 67.6%, excluding one tie; the unanimous subset is the 23–9 sample shown above. These are different populations, not interchangeable success rates.

The models still use the raw fallback, T = 1, with calibration unvalidated. Separate winner, margin, spread and total models can disagree, as Seattle and Miami–Indianapolis demonstrate. Modest cover probabilities alongside unanimous winner votes should remain modest in the write-up.

The dashboard’s last portfolio import uses Tuesday’s capture. Thursday’s market observations are not a newly selected betting portfolio, and settled-bet evidence remains insufficient to calculate ROI. This week’s test is whether the nine unanimous selections can sustain their recent winner record while the original spread and total predictions receive their own accounting.

Source note: the project’s November 12 preview handoff, Thursday DraftKings line report and supplied five-model record chart; the frozen Week 10 forecast issued November 10 at 1:54:26 p.m. CST; and the Week 9 recap for historical results and probability context. Point scores, margins and totals are rounded independently from full precision. Original forecast lines remain the grading benchmarks; Thursday’s quotes are labeled separately.