Photo by Igor Omilaev on Unsplash
The same data-over-reputation shift moving through NFL front offices is reshaping the daily-fantasy market that feeds off the season.
Caleb Downs never ran the 40-yard dash at the combine or his pro day, which left front offices guessing at a basic question about the Ohio State safety: how fast is he? This spring, some teams answered it with artificial intelligence, tapping the league’s Microsoft partnership and firms like Teamworks to pull speed and efficiency metrics straight from game film. The verdict on Downs was blunt: his measured game speed graded below the other top safeties in the class. AI made a similar call on edge rushers Arvell Reese and David Bailey, flagging that Reese dropped into coverage on roughly half his snaps and rushed less efficiently than his peers, the kind of read that changes how a team plans to use a player rather than whether it drafts him.
That shift, from gut and reputation to a measured number, is the same instinct that now drives a fast-growing corner of the football economy. Daily-fantasy operators built their whole product on the idea that a projection can beat a posted line, and the players who win are the ones who trust the data over the famous name.
Speaking to RotoWire, which publishes the best PrizePicksplays alongside refreshed projections for every NFL slate, one data analyst framed it this way: “When a model tells you Caleb Downs grades slower than the field even though he skipped the 40, you learn to trust the measurement over the highlight reel, and that’s exactly the muscle a sharp pick’emplayer uses when a projection disagrees with the number on the board.”
The math rewards that discipline. On PrizePicks, a two-pick Power Play pays 3x, so each leg has to hit about 58 percent of the time just to break even, which makes the popular two-pick entry the worst price on the menu. The value sits in the five- and six-pick flex plays, where insurance on a missed leg pulls the effective price down to roughly -119 and -118 per pick. “The name doesn’t pay you,” one observer noted. “A projection that beats the posted number does, whether that projection comes from a scouting model or a Sunday player-prop board.”
The business scale underneath all of it is why operators are racing to build it out. PrizePicks, bought by Allwyn for $1.6 billion in January, moved its U.S. pick’em to a peer-to-peer model in August 2025 to keep the format running nationwide, and analysts at Jefferies pegged prediction-market volume around Super Bowl LX at a record $20 billion. It sits on top of a league whose own revenue keeps compounding: as Sportico reported, each of the 32 NFL teams cleared more than $450 million in gross revenue last season, with the national payout up roughly 5 percent year over year to about $14.5 billion. The same analytics wave that front offices spent two decades absorbing is now compressing into AI, and every layer of the sport that trades on player performance is adjusting to it.
Readers who want the infrastructure side of that build-out can dig into more of our sports-betting coverage in the B2B race running underneath the betting boom. With the season opening Sept. 9, the teams and the platforms are chasing the same edge from opposite ends of the same data.
Anyone playing daily fantasy or betting on the season should do so responsibly; confidential help is available around the clock through the National Council on Problem Gambling at 1-800-GAMBLER.
