The Field Vision Model is alive and improving every week
Tony Romo recently coined the term “Pat GPT.” He was talking about Kansas City Chiefs quarterback Patrick Mahomes, and his Ai-like ability to take in defensive information as the game is playing out, quickly decipher and learn from that information, and adjust his approach on the fly in whatever way necessary for his team to win.
Ai – artificial intelligence – is defined as the following: the development of computer systems that can perform tasks typically requiring human intelligence, such as learning, problem-solving, decision-making, and perception, often by analyzing large datasets to simulate these cognitive functions.
In 2025, everyone’s got an Ai-model for sports. Every week there’s a new model for fantasy projections, win probabilities, betting outcomes. Most of them don’t work, and they definitely don’t learn. At Field Vision we take pride in our data science.
We track our performance relentlessly, and we’re here to tell you that a year-and-a-half into our predictive Ai-journey, our model undoubtedly works, and is getting better each passing NFL week that we ingest data.
The proof is in the numbers. Field Vision’s Matchup Index quantifies a player or teams' matchup advantage over their competition, and predicts outcomes against market expectations (i.e. Vegas odds). To beat Las Vegas – or, to make money - you need to be right 53% of the time. Below are our Field Vision Results through a season and a half:
2024 NFL regular season: 63% correct
2024 NFL Playoffs: 65% correct
2025 NFL Week 1: 55% correct
2025 NFL Week 2: 70% correct
2025 NFL Week 3: 69% correct
2025 NFL Week 4: 75% correct
That’s now three consecutive weeks where our model has accurately predicted player and team outcomes at nearly 70% or greater. 70%! That’s unheard of. But how are we doing it? We’ll, our model is learning. Let me give some examples just from Week 4 of the 2025 Season:
The Chiefs-Ravens OVER 48.5 pts. Result – Chiefs 37, Ravens 20 = Total of 57 pts. Chiefs games have consistently been going under the total for two years now. So why would our model love the OVER in a Chiefs game? Well, let me count the ways. The Ravens defense is easier to break through than a wet napkin. Through the first three weeks, they were last in the league in points per game allowed, at over 33 points per game. The totals in the first three Ravens games? 81 against Buffalo, 58 against Cleveland, and 68 against Detroit. They haven’t even sniffed a total in the 40s yet. This defense is bad. Couple that with the Chiefs returning Xavier Worthy, Mahomes getting comfortable with Tyquan Thornton, and Kelce moving better than he has in a couple years, and our model had a feeling there would be points. Turns out there was. A lot of them.
The Seahawks -1.5 vs. the Arizona Cardinals. Result – Seahawks 23, Cardinals 20. Our model’s pre-game prediction was actually exactly on the mark for this one, predicting a 23-20 outcome in favor of the Seahawks. In reality, what we knew is that the Cardinals numbers (and performance) were inflated by a very poor level of competition through the first three weeks – they had beat the two worst teams in the league (the New Orleans Saints, the Carolina Panthers), and lost to a 49ers team that was without Nick Bosa, George Kittle, and Brock Purdy. Meanwhile, Mike MacDonald’s defense has been one of the best in the league - 2nd in points allowed per game, 7th in EPA per game, and 4th in explosive plays allowed. Mike MacDonald is one of the elite defensive play callers in the league, and he now has the talent in Seattle to pair along with his scheme. We knew the Arizona offense would struggle to move the ball, and the reality is they had 3 points midway through the 4th quarter before two garbage time touchdowns. In retrospect, we all should have seen this outcome coming. Our model did.
Bo Nix OVER 211.5 Passing yards vs. Cincinnati: Result – 29 completions for 326 yards. In retrospect, this was obvious (which our model knew beforehand). Cincinnati has one of the worst pass defenses in the league. In Week 1 they gave up 290 yards passing to 40-year-old Joe Flacco. Week 2, 270 yards to Trevor Lawrence. Week 3? 48 points to Carson Wentz and the Vikings. We knew Sean Payton would exploit the short and intermediate areas of the field against an extremely week Bengals secondary. He absolutely did.
Ladd McConkey UNDER 81.5 receiving yards against the Giants. Result – 1 reception, 11 yards. Another easy win. McConkey’s numbers through 4 games of 2025? 6 for 74 against the Chiefs, 5 for 48 against Las Vegas, 4 for 41 against Denver. He’s gotten worse every week, and that has coincided with the emergence of Quinton Johnston as a legitimate #2 threat, and the reemergence of Keenan Allen as a preferred target for Justin Herbert. Markets are still valuing Ladd McConkey like it’s 2024. It is not. And we told you on Friday morning.
Saquon Barkley UNDER 78.5 rush yds against Tampa Bay. Result – 19 carries for 43 yards. If you’ve been following Saquon Barkley so far in 2025, he’s not the same threat as he was in 2024. Neither is the entire Eagles offense. Saquon is averaging 3.1 yards per carry in 2025, which is good for 141st best in league! Saquon has gone over this number one time all season – but the kicker here from model is that the Eagles were playing the Tampa Bay Bucs. Todd Bowles defense has historically been VERY strong against the run with Vita Vea manning the middle. You can beat their defense, but you have to do through the air. In 2025, the Bucs are giving up the 5th fewest rush yards per game. Add these two together, and you get a 43 yard performance for Saquon. Getting into the 70’s was always a stretch, and our model knew it.
Last year, we believed our model was learning. We improved our performance weekly, and that culminated in a 70% correct performance in the Super Bowl, with our model giving one of its strongest matchup indicators all season - a dominant win for the Eagles. This year? We know it's working. The model is ingesting data, and getting more accurate week-to-week. If you haven't been tailing, you've been missing out.
We'll keep it rolling in week 5. Look out for this week's best bets, as well as our predictions for all your primetime matchups. We look forward to the conversation, debate, and your feedback. Hit us up at @fieldvisionmi on X, @fieldvisionsports on Instagram, or on our website www.fieldvisionsports.com.