The two 4.5-point prices can break a FLEX tie. They cannot pick a lineup by themselves. This snapshot has team spreads and nothing about routes, carries, targets, or injuries. That makes it useful as a last-step game-environment check, not a substitute for ranking the players actually in your decision.
The two loudest prices
Provenance tier: live_pregame; source table: game_odds. The Chargers were minus 4.5 at home against the Rams . Miami was a 4.5-point home underdog to Atlanta . Those were the widest cited NFL spreads, so they deserve attention when two FLEX options have comparable roles.
The favored side gets a modest nudge because the market expected it to control more of the game. The underdog side gets a caution flag because its offense may be forced away from the preferred plan. Neither statement tells us which player receives the ball. A pass-catching back on the dog can benefit from a chase script; a low-volume receiver on the favorite can disappear. The player role still carries the decision.
The field-goal tier is where overconfidence starts
Provenance tier: live_pregame; source table: game_odds. Tennessee was minus 3 against Chicago , and Philadelphia was minus 3 against Cincinnati . A field-goal spread describes a competitive game, not a clean fantasy split.
That means the original “start the favorite, bench the dog” scaffold is too blunt. The Bengals side can create catch-up volume. The Bears side can still support a player with a secure workload. The Titans and Eagles sides can stall despite being favored. Use these rows to understand the expected shape of the game, then go back to carries and targets.
The target-share and air-yards guide supplies the player-level half of that work. A spread becomes useful only after you know who is likely to earn the opportunities it may create. The start/sit tool turns that role evidence into the actual flex decision.
The narrow games should barely move a FLEX call
Provenance tier: live_pregame; source table: game_odds. Buffalo was minus 1.5 against Pittsburgh . Kansas City was minus 1.5 against Seattle , and Dallas was minus 1.5 against New Orleans .
Those are close-game prices. Any article that turns them into a confident position ranking is pretending the spread contains more information than it does. In this tier, start the better role. Use the line only when everything else is genuinely tied, and even then treat the result as a lean.
What the market gap really is
There is no published player projection beside these spreads, so there is no honest points gap to advertise. The actual gap is the missing comparison market. Every cited row came from one book. Zero range at one book is not consensus; it is one observation.
That matters because an early quote can move when another shop posts or when injury information reaches the market. The cure is not a stronger adjective. It is another current line. Until that arrives, the Chargers and Atlanta favorite sides stay useful context and nothing more.
This is also why the favorite-versus-underdog framing should never erase position type. A runner with receiving work, a slot receiver with short-area volume, and a touchdown-dependent tight end can react differently to the same game script. The spread is shared context; the fantasy paths are not.
A FLEX process that survives Sunday morning
Rank the players by secure opportunities first: carries, routes, targets, and red-zone work. Remove anyone whose role or health changed. Use league scoring to separate rushing, receiving, and touchdown-dependent profiles. Only then bring in the spread.
On this board, the 4.5-point favorite sides provide the largest script adjustment. The field-goal favorites provide a smaller one. The 1.5-point games are close enough to ignore unless the player cases are otherwise even. Recheck the market after news, but do not turn a team price into a player guarantee. That is the difference between using the board and letting the board use you.
NFL ATS cover-margin distribution
Bars count completed NFL schedule rows by closing-spread cover margin using the repository canonical home-margin grading convention.
Model calibration from graded predictions
Calibration points render only when a verified source binds prediction probabilities to settled outcomes for the same observations.
