Denver Wins in Overtime; Week 6 Turns to Sunday and Monday Night

Denver won 17–14 in overtime, giving our Week 6 predictions a correct winner and spread pick but a miss on the total. Our published Thursday winner record is now 3–3. The weekend brings two different primetime calls: a modest Green Bay lean against Dallas and a much stronger San Francisco win forecast against Washington.

The Broncos improved to 4–2, while Seattle fell to 3–3. Thursday’s narrow 52.8% Denver selection landed, but both offenses finished below the separate point model’s expectations. Broncos recap, Seahawks recap.

Pressure and a final takeaway decide Thursday

Seattle erased a 14–3 deficit, but Ja’Quan McMillian’s overtime interception of Sam Darnold gave Denver another chance. Wil Lutz finished the ensuing possession with a 53-yard winning field goal. The Broncos recorded five sacks and three takeaways, getting the defensive plays their offense needed. Denver’s game recap.

Seattle’s defense held Denver to 168 yards and 2.9 yards per play. On the other side, Jaxon Smith-Njigba caught 12 passes for 134 yards, but the Seahawks could not turn that production into enough points. Seattle had scored at least 29 in each of its previous four games; this time, both teams’ scoring stayed low even with overtime. Seattle’s rapid reactions.

Denver also ended Seattle’s seven-game primetime winning streak, previously the NFL’s longest. NFL Research. That closes one of the storylines in our Thursday preview, alongside the question of whether Seattle could exploit Denver’s depleted secondary without losing possessions to mistakes.

How the original picks graded

These are the selections issued Tuesday, October 13, at 2:48 p.m. CDT, preserved before kickoff and published in the preview.

Denver’s cover calculation is 17 − 14 + 3 = +6. The combined score finished 9.5 points below the original total. Week 6 therefore opens 1–0 on winners, 1–0 against the spread and 0–1 on totals, with thirteen games remaining. These are three related grades from one game, not three independent tests.

The spread grade belongs to the original +3, even though our Thursday price snapshot had Denver at +2.5. Neither is presented as a verified closing line. Thursday’s report had no qualifying Seattle–Denver recommendation, so correct directional picks should not be turned into a claimed betting return.

The Thursday winner record improves from 2–3 to 3–3: the Bills, Browns and Broncos were correct; the Rams, Packers and Cowboys were misses. That count follows the original published selections on the U.S. Central Thursday schedule. The Wednesday season opener remains excluded, and later model revisions do not rewrite the record.

A better margin call than scoring call

The separate XGBoost margin and total models supplied the point projections below. They are not the five-model winner blend or the spread-cover classifier.

Denver scored 4.64 fewer points than projected, and Seattle scored 6.99 fewer. The projected total was 11.62 points too high, while Denver’s actual winning margin was 2.35 points larger than the model’s estimate. Each difference is calculated before rounding; the displayed team projections therefore need not sum exactly to the displayed total.

Three winner models—Elo, Bayesian and Random Forest—picked Denver correctly. Logistic regression and XGBoost preferred Seattle and missed. The weighted blend’s correct pick does not establish that its precise 52.8% estimate was well calibrated. These remain raw probabilities, and Friday’s prices and reporting have not refitted them.

Friday’s remaining board

The new market capture was taken Friday, October 16, at 12:03 p.m. CDT, covering thirteen remaining games across eleven bookmakers. The completed Seattle–Denver game is excluded. All prices below are saved observations, not live offers.

Among the larger BetMGM changes since Thursday, Philadelphia moved from +1 to +3.5 against Carolina, Cleveland moved from +5 to +3.5 against Baltimore, and the Baltimore–Cleveland total fell from 42.5 to 40.5. Chicago–Atlanta’s total rose from 51 to 52. Those observations establish movement without identifying its cause.

Sunday’s earlier slate still has strong outright forecasts for Kansas City (74.9%), the Rams (74.8%) and Jacksonville (71.0%). The full Week 6 preview contains all fourteen original picks. Here is the closer look at the two remaining night games.

Sunday night: Dallas at Green Bay

Sunday, October 18, 7:20 p.m. CDT at Lambeau Field, on NBC. Green Bay is the saved winner pick at 58.4%, with all five winner models selecting the Packers. Their estimates range from 50.3% to 67.4%, so the agreement includes an almost even forecast. Packers viewing guide.

The other outputs are less aligned:

The point model narrowly favors Dallas by 0.74, while the winner blend chooses Green Bay. These separately fitted models disagree; combining them into one supposedly consistent forecast would hide that. The spread probability supplies almost no separation at −1.

The same-book BetMGM history is:

Green Bay’s moneyline eased Friday, but the spread and total still match the frozen thresholds. Friday’s Under 53 price is −108. A one-point Packers win pushes −1; exactly 53 points pushes the total. The exports do not provide separate push probabilities.

The football question is whether Green Bay can limit Dallas’s explosive plays. CeeDee Lamb and George Pickens combined for 857 receiving yards through five games, while the Packers enter after forcing five Chicago turnovers. Making Dallas sustain drives is a different challenge from expecting another five-takeaway game. Packers matchup analysis.

In the Thursday practice report, Green Bay cornerback Keisean Nixon was limited, while Benjamin St-Juste did not participate. Those are dated practice observations, not final Sunday statuses. Packers–Cowboys injury report.

Monday night: Washington at San Francisco

Monday, October 19, 7:15 p.m. CDT at Levi’s Stadium, on ESPN and ABC. San Francisco is the strongest saved Week 6 winner pick at 76.8%. All five winner models select the 49ers, with estimates between 74.0% and 79.7%. 49ers viewing guide.

San Francisco remains the favored winner even though the separate spread blend slightly prefers Washington +7.5. A 49ers win by one through seven points would satisfy both picks. The point model’s 7.10-point San Francisco margin also falls below 7.5, but that margin is an estimate, not a cover probability.

BetMGM’s four saved observations show the total moving while the spread holds:

Friday has Washington +7.5 at −110, matching the original handicap. The total rose another half point from Thursday and is now a full point above Tuesday’s observation. The saved 51.6% over probability belongs to 50.5, not 51.5. No replacement probability has been calculated at the higher line.

Third downs frame the matchup: San Francisco enters averaging 30.8 points and converting 51.92% of third downs. Washington’s offense has converted 31.88%, while its defense has allowed 29.8 points per game. That puts Washington’s ability to extend possessions—and end San Francisco’s—near the center of the game. Official Washington matchup preview.

Washington’s Stefon Diggs and Jacory Croskey-Merritt missed Thursday practice, and Terry McLaurin was limited. Their final Monday availability is not established by those practice entries. Commanders injury report.

Neither upcoming primetime game has a qualifying offer in Friday’s report. The directional picks remain Green Bay and San Francisco, with their separate spread and total forecasts tied to the original lines. Updated prices and final availability will be later checkpoints; this is the Friday view.

Source note: This post uses the Friday operations package at prediction-repository revision 8ae982c, the unchanged Tuesday forecasts and the linked official reporting. Practice updates above are Thursday’s reports, rechecked Friday; final Sunday/Monday designations and inactives were not verified. Companion evidence retains source hashes, original-line settlement, full-precision point comparisons and all four market snapshots. No models were refitted for this article.