Cowboys vs. Giants 2026: The Clock Cooperated. The Scoreboard Didn’t. | SR / NB / 007
SR / NB / 007 — The Retriever’s Notebook
Classification: NFL Prediction Audit
Game: Dallas Cowboys at New York Giants — September 13, 2026
Final Score: New York Giants 28, Dallas Cowboys 20
The Giants took more than eight minutes to score a touchdown.
Then, on their next possession, they did it again.
For anyone holding an Under, the clock appeared to be cooperating. The scoreboard wasn’t.
New York beat Dallas 28–20, producing 48 total points and a losing result for SharpRetriever’s pregame Under 47.5 read.
The margin against the betting line was half a point.
The difference from the model’s 42.78-point projection was 5.22 points.
Those are different measurements. An honest audit keeps both.
This game wasn’t a parade of instant touchdowns. Long possessions repeatedly ended in the end zone, and a late Dallas score pushed the total beyond the published number.
The clock kept moving.
So did the score.
Cowboys vs. Giants Result at a Glance
Published read: Under 47.5.
Displayed model probability: 64%.
Projected total: 42.78 points, displayed as 42.8.
Halftime score: Giants 14, Cowboys 7.
Final score: Giants 28, Cowboys 20.
Final total: 48 points.
Read result: Lost by 0.5 points against the published line.
Projection difference: Actual scoring exceeded the projection by 5.22 points.
Decisive play: Dallas’s 17-yard touchdown reception with 5:21 remaining.
Notebook verdict: The Under lost. Sustained touchdown drives—not simply fast scoring—were central to how the game exceeded the projection.
The Original SharpRetriever Prediction
SharpRetriever’s strongest pregame read was Under 47.5 at a displayed 64% probability, supported by a projected total of 42.78.
That placed the projection 4.72 points below the published line.
The accompanying explanation referred to recent scoring indicators, the divisional matchup and the limited early-season sample. It presented a lower-scoring expectation without identifying one dominant statistical factor.
That distinction matters.
The published explanation did not establish a specific prediction about eight-minute possessions, third-down stops or red-zone efficiency. We should not retrofit those claims into the model after seeing the game.
What we can evaluate is clear:
The published Under result.
The difference between projected and actual scoring.
The possessions and scoring events that challenged the read.
The original price for Under 47.5 and the closing total are not included in the preserved material. This audit therefore does not assign a financial return or a closing-line-value grade.
A 14–7 Halftime Score Left the Under Unresolved
At halftime, Dallas and New York had combined for 21 points.
Under 47.5 remained possible. Another 26 points would have left the final total at 47; another 27 would have beaten it.
But the path to halftime already contained a warning about how quickly a scoring picture can change.
With the game tied 7–7, the Giants lost a fumble late in the second quarter. Dallas then gave possession back when Dak Prescott was intercepted by Jevón Holland.
New York took over at its own 45 with 49 seconds remaining.
The Giants needed only 29 seconds to turn that opportunity into a touchdown, finishing with Jaxson Dart’s 15-yard pass to Isaiah Likely with 20 seconds left.
That sequence changed a 7–7 game into a 14–7 halftime lead.
It did not settle the total. It demonstrated that a game with several lengthy possessions could still produce a quick scoring burst.
Two Eight-Minute Drives. Two Touchdowns.
The Giants opened the second half with a 12-play, 74-yard touchdown drive lasting 8:01.
Dart found Likely for a two-yard score. New York led 21–7.
Dallas answered with a 12-play, 75-yard touchdown drive lasting 6:22. Javonte Williams finished it with a one-yard run, bringing the score to 21–14.
Then the Giants put together another extended possession:
13 plays. 65 yards. 8:11.
Dart’s nine-yard touchdown pass to Devin Singletary made it 28–14.
Those three consecutive touchdown drives consumed 22 minutes and 34 seconds, using the possession durations in the supplied play-by-play.
They also added 21 points.
This was the central tension of the night. The offenses were consuming time, but they were not leaving those possessions empty-handed.
Long drives reduce the time available for more possessions. They do not erase the points scored on the possessions already happening.
At 28–14, the combined score stood at 42.
The Under could survive five more points.
It could not survive another touchdown.
Third-Down Conversions Kept the Scoring Drives Alive
The duration of a drive tells only part of its story.
Whether it ends in a punt, a field goal or a touchdown matters just as much.
On Dallas’s third-quarter touchdown drive, Prescott found CeeDee Lamb for 15 yards on third-and-12, moving the ball to the Giants’ two-yard line.
That was not merely another completion. It extended a possession that ended in seven points.
On New York’s following touchdown drive, Singletary caught a 13-yard pass on third-and-nine. Later, he scored on a nine-yard reception on third-and-five.
Those conversions help explain why the scoreboard kept advancing despite the time consumed.
The Under did not need the offenses to stop gaining yards altogether. Stops that forced punts or field-goal attempts could have changed the scoring path.
Instead, important third downs became fresh opportunities—and eventually touchdowns.
The Touchdown That Beat Under 47.5
Dallas received the ball trailing 28–14 with 7:26 remaining.
This possession moved much faster than the three touchdown drives before it.
The Cowboys covered 74 yards in nine plays and 2:05.
With 5:21 left, Prescott found Williams for a 17-yard touchdown reception.
The score became:
Giants 28, Cowboys 20.
The combined total became:
48.
That was the moment Under 47.5 lost.
Brandon Aubrey subsequently missed the extra point, but the total had already crossed the line. The kick could have changed 48 to 49. It could not return the game to 47.
The Giants then held possession for the remaining 5:21, converting two third downs before eventually kneeling out the game.
There was no additional scoring.
There did not need to be.
Why This Was More Than a Half-Point Story
“Lost by half a point” accurately describes the result against 47.5.
It does not fully describe the projection.
SharpRetriever projected 42.78 points. The teams scored 48.
That is a 5.22-point underestimate.
Both facts belong in the record:
The betting-line result was narrow.
The scoring projection missed by a larger amount.
Calling the model “basically right” because the Under lost narrowly would blur those two measurements.
The pregame read was not Under 48.5. It was not a prediction that the game would finish somewhere near 48.
It was Under 47.5, backed by a 42.78-point projection.
That is the version this Notebook grades.
What the Audit Can—and Cannot—Conclude
The Game Showed Why Pace Alone Is Not Enough
The Giants’ two eight-minute touchdown drives are the clearest teaching point.
Time consumption can limit scoring opportunities. Efficient finishes can still turn those opportunities into enough points to beat an Under.
That is an observation about this game, not proof that the model explicitly relied too heavily on pace. The available pregame explanation does not reveal enough of the underlying calculation to make that diagnosis.
Penalties Helped Extend Scoring Opportunities
Dallas’s first touchdown drive included defensive pass-interference penalties of 38 and 19 yards.
Its third-quarter touchdown drive included two more defensive pass-interference penalties.
Those events belong in the reconstruction because they advanced the offense and sustained scoring opportunities.
They do not justify declaring that the prediction would have won “without the flags.” Removing a penalty also changes the down, field position, subsequent decisions and remaining clock.
We can identify what happened without inventing an alternate final score.
Four Straight Second-Half Possessions Ended in Touchdowns
The first four possessions after halftime produced:
A Giants touchdown.
A Cowboys touchdown.
Another Giants touchdown.
Another Cowboys touchdown.
That sequence added 27 points to the halftime total of 21.
The missed extra point explains why those four touchdowns produced 27 rather than 28. It did not change the Under’s losing result.
One Loss Does Not Settle a 64% Probability
A 64% forecast leaves room for the other outcome.
This game landing Over does not, by itself, prove that the probability was poorly calibrated. Nor would an Under result have proved it was well calibrated.
Calibration requires a collection of comparable forecasts and outcomes.
The result here is narrower and certain: this published Under lost.
The model’s broader reliability remains a question for the cumulative record.
The Notebook 007 Verdict
Outcome: Under 47.5 lost. Giants 28, Cowboys 20 produced 48 points.
Projection accuracy: The 42.78-point forecast underestimated scoring by 5.22 points.
Decisive moment: Williams’s 17-yard touchdown reception with 5:21 remaining moved the total from 42 to 48.
Game mechanism: Long possessions repeatedly ended in touchdowns. Third-down conversions helped sustain that scoring.
Complications: Penalties extended opportunities, and a 29-second Giants touchdown drive changed the halftime picture. Neither observation excuses the result or proves a specific model defect.
Process assessment: The preserved explanation is too broad to establish exactly which model input failed. One outcome cannot establish calibration, and the missing price and closing line prevent a complete market-value audit.
Lesson: A game can use the clock exactly as an Under bettor hopes and still score too efficiently for the Under to survive.
Notebook 006 recorded two winning predictions.
Notebook 007 records a losing one.
Both belong in the same record.
The SharpRetriever Sports Puzzler
THE SITUATION
Giants 28. Cowboys 14.
The published read is Under 47.5.
Dallas scores a touchdown. The extra point misses.
Did the missed kick save the Under?
The Answer
No.
The touchdown alone made it 28–20.
That is 48 total points, already half a point above 47.5.
The missed kick kept the total from reaching 49. It did not rescue the original read.
Frequently Asked Questions
What was the Cowboys vs. Giants final score on September 13, 2026?
The New York Giants defeated the Dallas Cowboys 28–20, for a combined total of 48 points.
Did SharpRetriever’s Under 47.5 prediction win?
No. The final total of 48 exceeded the published line by 0.5 points.
What did SharpRetriever project for Cowboys vs. Giants?
The model projected 42.78 total points, displayed as 42.8, and listed Under 47.5 at a displayed 64% probability.
When did the game go Over 47.5?
Dallas’s 17-yard touchdown reception by Javonte Williams with 5:21 remaining made the score 28–20 and pushed the total to 48.
Why didn’t the missed extra point save the Under?
The touchdown had already pushed the combined score past 47.5. The extra-point attempt could only add to that total.
How did long drives still produce an Over?
The first four second-half possessions all ended in touchdowns. Three of those drives consumed substantial time, but their scoring efficiency still helped produce 27 second-half points.
Read the Record. Follow the Trail.
A winning prediction deserves examination.
A losing prediction deserves the same.
SharpRetriever’s Notebook preserves the original read, follows the game that tested it and records the outcome without moving the line afterward.
The clock cooperated.
The scoreboard didn’t.
The record keeps both.
Source basis: SharpRetriever’s supplied pregame card and the supplied Cowboys–Giants play-by-play. Possession durations follow that play-by-play.
SharpRetriever publishes probabilistic sports analysis, not guarantees. Model probabilities describe uncertainty; they do not eliminate it. Informational purposes only—not betting advice.