Four Favorites Fell on Week 2 Sunday: +110.6 Units

Sunday, September 20, 2026 was the best betting day of our season: $113,412 wagered returned $124,472.93, a profit of $11,060.93, or 110.6 units. Four favorites lost outright. Baltimore fell 24-17 to New Orleans as an 8.5-point home favorite, Tampa Bay lost 24-19 to Cleveland by the same number, Minnesota won 9-3 in Chicago as a 4.5-point underdog, and the Chargers went down as well. When a whole Sunday tilts one direction, live prices lag every single game.

The day before this one, we lost 61.1 units. Both are on the results page. This is what the good end of that swing looks like.

What Happened

Baltimore 17, New Orleans 24. The Ravens were 8.5-point home favorites. They led 17-9 entering the fourth quarter and did not score again, while New Orleans closed with two touchdown drives.

Tampa Bay 19, Cleveland 24. Also an 8.5-point home favorite, also beaten outright. The game included a two-hour lightning delay before the final two minutes were played.

Chicago 3, Minnesota 9. A 4.5-point underdog won a field-goal fight on the road. Six total points in sixty minutes.

Dallas 37, Washington 20. Dak Prescott threw four touchdown passes as Dallas got to 1-1.

The Ravens, Chargers, Bears and Texans were all upset on the same afternoon, with Baltimore and the Chargers each laying more than six points.

Why This Kind of Sunday Pays So Well

The mechanism is the same in each game, and it is the mirror image of what cost us 61 units on Saturday.

Live models are trained to expect favorites to reassert themselves. When Baltimore led 17-9 into the fourth, the in-game price still strongly favored Baltimore, and historically that is the right lean. Teams with a lead and a talent edge usually close.

The gap we work in is between how far the model moves after each event and how far the game has actually moved. On a normal Sunday that gap is small and the money is grinding. On a Sunday where four favorites lose outright, the gap opens in the same direction in game after game, and every correction the model makes is a beat behind the field.

The Ravens game is the clean example. Leading by eight entering the fourth quarter is a specific, highly-priced state. When a team in that state stops scoring entirely and the opponent produces two drives, the model has to walk a long way from where it started, and it walks slowly because the prior is strong.

Tampa Bay's game had an extra wrinkle. A two-hour lightning delay before the final two minutes is precisely the kind of event no in-game model handles well. There is no training data for a game that stops for two hours and restarts with 120 seconds left.

The Game That Went the Other Way

Not everything broke toward the underdog, and the exception is worth noting because it is where a lot of bettors lost money chasing the pattern.

Dallas beat Washington 37-20 with Dak Prescott throwing four touchdown passes. That is a favourite doing exactly what a favourite is supposed to do, on an afternoon when almost nothing else did.

This is the trap on a day like this one. Once three favourites have gone down, the temptation is to treat "fade the favourite" as the read of the day and apply it everywhere. That is not a read, it is a pattern noticed after the fact, and it is how a good afternoon turns into a bad one in the late window.

The actual signal was never "favourites are losing today." It was game-state specific: a leading team that has stopped scoring, priced as though it had not. Baltimore fit that. Tampa Bay fit it. Dallas never did, because Dallas kept scoring.

The distinction matters. One is a condition you can observe in the game in front of you. The other is a story about the day. Conditions repeat and can be priced. Stories about the day are how people lose the money they made in the early window.

Why Week 2 Specifically

There is a seasonal reason this kind of Sunday clusters early.

Week 2 numbers are built on one game of current-season data plus last year. A team that looked dominant in Week 1 gets rated up, sometimes far up, on a single result against an opponent nobody has measured either. Baltimore laying 8.5 at home and Tampa Bay laying 8.5 at home were both numbers carrying more confidence than three weeks of evidence could support.

By November those numbers are built on real form and the board stops producing four outright upsets in an afternoon. September is when the market is guessing with conviction, which is the most useful combination there is.

The Documented Day

Note that the handle is almost identical to Saturday's $113,724, the day that lost 61 units. Same size card, opposite result. That is the clearest illustration on this site of why single days prove nothing.

What This Game Teaches About Football Handicapping

A leading team that stops scoring is the single best live signal in football. Not a team that is behind. A team that is ahead and has gone quiet. Baltimore entering the fourth at 17-9 and finishing at 17 is the whole pattern. The model prices the lead; the game is being decided by an offense that has stopped functioning, which the model reads as noise until it is too late.

Weather and stoppages are free edges. A two-hour lightning delay with two minutes left is unmodellable. Anyone watching knows more than the algorithm in that moment, and that is rare enough to be worth acting on. The same is true of wind in the fourth quarter and of a kicker warming up in it.

The same board-wide correlation that destroyed Saturday created Sunday. This is the honest symmetry. Our approach concentrates risk on a shared read, which produces both the 61-unit loss and the 110-unit win. That is not a flaw to be engineered away so much as the actual shape of the product, and anyone buying NFL picks against the spread or live alerts from anyone should understand they are buying that shape.

The practical filter for football handicappers is whether they explain this or hide it. A service showing you only Sundays like this one, and never the Saturday before, is describing a different business than the one it runs.

What This Doesn't Prove

One very good day proves as little as one very bad day.

The 110.6 units here came with roughly the same handle as a Saturday that lost 61.1. If you saw only this article, you would form a badly wrong impression of what a week looks like. That is exactly why the losing days are published in the same place and the same format.

We are not claiming a win rate, this week or ever. Individual dated statements are the entire claim. Past results do not guarantee future performance.

Where the Week Went

Five days, Thursday to Monday:

That is +66.4 for the stretch with two losing days inside it. September stands at +226.6. The week's card is on best bets today and the season ledger is on NFL picks.

Frequently Asked Questions

Which favorites lost on September 20, 2026?

Baltimore lost 24-17 to New Orleans as an 8.5-point home favorite, Tampa Bay lost 24-19 to Cleveland as an 8.5-point home favorite, Minnesota won 9-3 at Chicago as a 4.5-point underdog, and the Chargers were also upset. Baltimore and the Chargers each laid more than six points.

What happened in the Ravens-Saints game?

Baltimore led 17-9 entering the fourth quarter and did not score again. New Orleans closed the game with two touchdown drives to win 24-17.

Why was the Buccaneers-Browns game delayed?

A two-hour lightning delay stopped play before the final two minutes. Cleveland won 24-19.

How much did The Best Bet on Sports make that Sunday?

$11,060.93 on $113,412 wagered, or 110.6 units at $100 a unit. It is the best single day published on this site, and the statement is posted with it.

Why is a leading team that stops scoring a good live betting signal?

Because live models weight the scoreboard heavily and treat a stalled offense as noise until several possessions confirm it. A team ahead by eight that has genuinely stopped functioning is priced as though it is still ahead by eight and still functioning.

Does one big day mean the service is profitable?

No, and it should not be read that way. This day came with nearly identical handle to a Saturday that lost 61.1 units. Only a long ledger with every day in it says anything meaningful.

How do I get these alerts during the games?

By Email and Discord the moment a mispriced live number appears, with the side, the price at dispatch and the unit size. Packages start at $39.95 a month on the packages page, or reserve a free live pick first.