Every NFL pick published on this page comes with three things attached: the reasoning, the line it was made at, and, once the season is underway, a running record of how the calls have gone. That last one matters most. The picks industry mostly hides its results, and an entire website exists just to track what the experts got wrong. We would rather save it the work. Our NFL picks are free, written against lines actually posted at US sportsbooks, and never sold as anything more certain than they are. 21+ in most states.
Week 3 NFL Picks
| Date | Match | Signal pick | Win prob |
|---|---|---|---|
| Thu Sep 24 |  ATL@GB  | GB -6.4 | 73%73% |
| Sun Sep 27 |  KC@MIA  | KC -5.4 | 69%69% |
| Sun Sep 27 |  CAR@CLE  | CAR -1.1 | 54%54% |
| Sun Sep 27 |  TEN@NYG  | NYG -8.4 | 78%78% |
| Sun Sep 27 |  NE@JAC  | JAC -0.2 | 51%51% |
| Sun Sep 27 |  LAC@BUF  | BUF -9.4 | 81%81% |
| Sun Sep 27 |  NYJ@DET  | DET -9.0 | 80%80% |
| Sun Sep 27 |  HOU@IND  | HOU -2.7 | 60%60% |
| Sun Sep 27 |  SEA@WAS  | SEA -7.5 | 76%76% |
| Sun Sep 27 |  CIN@PIT  | CIN -0.3 | 51%51% |
| Sun Sep 27 |  ARI@SF  | SF -8.7 | 79%79% |
| Sun Sep 27 |  MIN@TB  | MIN -2.8 | 61%61% |
| Sun Sep 27 |  LV@NO  | NO -3.0 | 61%61% |
| Sun Sep 27 |  BAL@DAL  | BAL -2.9 | 61%61% |
| Sun Sep 27 |  LA@DEN  | DEN -2.6 | 60%60% |
| Mon Sep 28 |  PHI@CHI  | PHI -1.2 | 54%54% |
Straight-up pick, projected line and win probability from the Squawka Signal model — a model read, not a betting tip. Tap a win probability to trade the game on Kalshi (US) or back it at bet365 (rest of world) — also on Polymarket and ProphetX. 21+ where applicable; eligibility varies by US state. Updated as the season plays out. How the model works.
The 2026 season kicks off on Wednesday 9 September. Weekly picks begin with the Week 1 slate: every game called straight up and against the spread, with best bets flagged and the reasoning written out. Until then, this page covers how the picks are made and how to use them. Most bettors skip that part. They should not.
How we make our NFL picks
Every pick on this page comes from the Squawka Signal model, not a tipster’s gut. The model rates all 32 teams with an Elo power rating built on every game since 2021, adjusts each team for its starting quarterback — the biggest single lever in the NFL — and simulates the rest of the season tens of thousands of times. That yields a win probability and a projected margin for every game. We then set our number against the live prediction-market price, so you can see where the model and the market disagree. No black box: the method is identical every week, and the full detail is in our methodology.
Straight up or against the spread
Picks come in two currencies and it pays to know which one you are reading. A straight up pick answers one question: who wins the game. It is what office pools and most casual conversations run on. An against the spread pick answers a harder question: who beats the number the sportsbook set. A team can win the game and lose you the bet, and a 14-point underdog that loses by 10 was the right call. Our weekly picks give both, side by side, because they serve different bets. If the second currency is new to you, the markets section of our NFL betting guide covers how spreads are read and priced.
What about computer picks and consensus?
Our picks are model picks — and we treat that as the point, not a disclaimer. A model is disciplined and tireless, with no loyalty to a brand name or last week’s headline. What a model on its own won’t tell you is where it disagrees with the market, so we do: every pick sits beside the live Kalshi, Polymarket and ProphetX price, flagged where our probability is higher (value) or lower (a fade). Consensus — which side the public is loading — is useful mainly as a contrarian tell. Our edge is the model versus the price, not the crowd.
Our record, in the open
From Week 1, every pick gets logged and the running results published right here — straight up and against the spread, tracked against the closing line. Cold streaks will show, and that is the feature working: a picks page that never shows one is editing its history. The honest benchmark is modest — clearing roughly 53% against the spread across a full season is genuinely good, and any model claiming far more deserves suspicion. The record fills in as the season is played.
How to use NFL picks without outsourcing your judgment
A pick is a starting argument, not an instruction. The ones worth following tell you why, so you can disagree.
Check the price, not just the side. Picks on this page reference the line at time of writing, and lines move. If we liked a team at +3.5 and your book now shows +2.5, half the reason for the pick is gone. This is the strongest case for holding accounts at two or three books: the same game is priced slightly differently across them, and our NFL sportsbook ranking ranks the ones we rate for football. In a state without sportsbooks, prediction market contracts price the same outcomes their own way, so the check matters just as much there.
And size every bet like the pick might be wrong, because some of them will be. Decide a bankroll for the season, divide it into units, and bet in units. The habit sounds dull. It is also the entire difference between a losing streak that stings and one that ends your season in October. And if the mechanics of accounts, deposits and bet slips are still new, our guide to how to bet on the NFL covers them step by step.
What we publish each week
Each week, a game-by-game call for the full slate: the straight-up winner, the model’s win probability, and the projected line, so you have the against-the-spread read too. Every pick is logged and the results published in the open, tracked against the closing line. Player prop picks live with our dedicated NFL player props coverage rather than here.
Survivor pools: one pick, higher stakes
Survivor pools ask for one straight-up winner a week, no spreads, with elimination for a single miss. It is its own discipline: burning a heavy favourite early can cost you the team you need in Week 11, so future value matters as much as this week’s matchup. The model’s highest-probability winner each week — the top of the table above — is a natural survivor starting point, but weigh the weeks ahead before you spend it.
NFL picks FAQ
Yes, all of them. No paywall, no premium tier, no picks sold by text message. The business model is the same as the rest of the site: if you use a sportsbook link here, we may earn a commission; the picks themselves cost nothing.
Beating the sportsbook’s margin rather than just winning the game. A -6.5 favorite must win by 7 or more; a +6.5 underdog covers by winning outright or losing by 6 or fewer.
Anyone publishing honestly will tell you the same thing: 52.4 percent against the spread is break-even at standard -110 pricing, 55 percent over a season is strong, and 60 percent sustained is essentially unheard of. Judge any picks source, including this one, against those numbers.
The picks travel; the prices do not. A side we like is the same side at any book, but always check your own book’s line before betting, because it can differ from the one referenced in the pick. In states without sportsbooks, prediction market contracts price the same games differently again.
Through the week as lines and injury reports settle, ahead of each Sunday slate. Prime-time games are updated closer to kickoff when late news warrants it.
The Squawka betting team, with every pick attributed and every record published. No guest touts, no syndicated filler.
21+ in most states (18+ in some) | Gambling problem? Call or text 1-800-GAMBLER | Picks are opinions, not guarantees; odds referenced at time of writing and subject to change | More support resources on our gambling awareness page.