Probetrics — Sports Analytics
← NFL picksGame data: lines, stats, form, props →
Kansas City Chiefs
2-0
Model predicts
23 – 19
Sun, Sep 27, 1:00 PM ET
Hard Rock Stadium · 86° · Mostly sunny
Miami Dolphins
0-2
Spread KC -10.5 (opened KC -7.5)Total 46Moneyline KC -650 · MIA +4767 books · best KC -625 pinnacle · MIA +520 fanduel
Save imageSquare
WinnerProbability the picked team wins. Calibrated: when the model says 65%, it has been right about that often.Solid
KC -650
61%
model margin KC by 3.8
SpreadThe model's lean on this number. It can name the other team from the Winner tile; see the note below when it happens.Lean +
MIA +10.5
53%
6.7 pts from the number · best MIA +11.5 at pinnacle
TotalThe model's lean on this total, shown so you can see its number against the posted one.Lean +
Under 46
52%
model total 41.7 · best under 46.5 at hardrockbet
ⓘWinner and Spread don't match — that's not a contradiction. The model still has KC winning the game outright; it just doesn't think they win by enough to cover this number, so the spread lean is the other side, MIA. A slight favorite can easily be the value side to take points with.

The read

A quarterback change matters here: Malik Willis is starting for Miami Dolphins. The clearest mismatch is the passing game: KC passing ranks #13, MIA defends it at #27; the other way MIA #23 into KC's #15. The model lands on KC by 3.8 against a market KC by 10.5, so the spread lean is MIA by 6.7 points — a lean for reading the game, not the call itself. On the total the model sees 41.7 against 46.

Who has the edge

Model: KC by 3.8 · market KC by 10.5
Passing game
Clear edge
KC
KC passing ranks #13, MIA defends it at #27; the other way MIA #23 into KC's #15.
context · not in the margin
Rushing game
Even
—
KC rushing ranks #15, MIA defends it at #20; the other way MIA #9 into KC's #9.
context · not in the margin
Third down & red zone
Slight edge
KC
MIA third-down conversion ranks #27, KC defends it at #28; the other way KC #22 into MIA's #18.
context · not in the margin
Form (last 5)
Clear edge
KC
MIA −18.0 per game over the last five, KC +12.0.
context · not in the margin
Market
Clear edge
KC
Consensus has KC by 10.5; the model says KC by 3.8.
context · not in the margin

Where the lines sit Every NFL final since 2012, counted by margin and by total. The dashed lines are the market's number and the model's; the bright bars between them are the share of history that landed in the gap — the games the two numbers actually disagree about. Key numbers (the most common margins) are the pale bars.

3,862 NFL games since 2012
Marginmarket KC by 10.5 · model KC by 3.8
Mkt KC 10.5
Model KC 3.8
-28028

16.2% of NFL games land between the two numbers — the zone where the model’s side covers and the market’s does not. The model’s number is across the key numbers 7 and 6. Most common margins: 3 (14.4%), 7 (8.6%), 6 (6.7%), 14 (5.1%).

Totalmarket 46 · model 41.7
Mkt 46
Model 41.7
204672

11.5% of NFL games finish between 46 and 41.7 — the games where the under cashes on the model’s read but not by much. The league’s median total since 2012 is 45.

Sharp read Sharp money cannot be seen; who moved first can. Pinnacle and the other market makers move on the bets they respect and retail books copy them hours later. This tracks the market maker's number against retail, its move since the open, and whether the market followed. Recording started September 2026 and it is graded on its own — it is not a pick until a season of results says it should be.

— tap to see the market-maker read ↓
Spread
Sharp side: KC · model disagrees
Opened
KC -8.5
Pinnacle now
KC -11.25
Retail now
KC -10.5-0.8 behind
Moves
30 across 7 books · 3 led

First sharp move Mon 9:15 AM: KC -11 → KC -10.5, 0 books followed within six hours; retail was still KC -10.5.

Total
Sharp side: the over · model disagrees
Opened
44.5
Pinnacle now
46(+0.8)
Retail now
46
Moves
16 across 7 books · 1 led

First sharp move Mon 9:15 AM: 44.5 → 46, 0 books followed within six hours; retail was still 45.5.

Informational only. Winner probabilities are backtested and calibrated; spread and total calls are tracked leans. Model reads, not personalized advice. 21+, 1-800-GAMBLER.