Ohio State Won Nearly a Quarter of a Simulated Season. Here's What EA's Model Actually Sees

A hundred simulated seasons in College Football 27 gave Ohio State the crown 22 times, and the number says as much about how roster ratings drive outcomes as it does about who will actually win in January.

There's a specific kind of superstition that takes over a fan base every July, when actual football is still six weeks away and the only games being played happen on a screen. USA TODAY leaned into it this month by running the newly released College Football 27 through 100 full simulated seasons, treating EA Orlando's roster ratings and dynasty engine as a crude oracle for the 2026 College Football Playoff. Ohio State won the whole thing 22 times. Nobody else got out of the teens. That is either a meaningful signal about the sport's power structure or a reminder that video game algorithms and human polling committees are answering slightly different questions, and probably both.

Why this exercise matters

College football fans have always hunted for shortcuts to certainty, from way-too-early top 25s in January to betting market power ratings in June. What makes the College Football 27 simulation different is that it isn't really a prediction in the traditional sense. It's a byproduct of how EA rates 138 rosters, weighs returning production, and lets a game engine play out matchups thousands of times over. When Ohio State reaches the national title game in 57 of 100 simulated seasons, that's the game's internal math saying the Buckeyes' roster grade separates them from the field by a wide margin, not that a committee of humans would agree in December.

That distinction matters because last year's version of this same exercise pointed at Penn State as the overwhelming favorite, and the Nittany Lions never got within a mile of the actual playoff. Instead, James Franklin was fired midseason and resurfaced at Virginia Tech, while Indiana, a team the game apparently didn't see coming at all, finished a perfect 16-0 and beat Miami for the national championship. The simulation isn't just for fun. It's also a running experiment in how much roster talent alone can tell you about a 12-team playoff field that keeps producing outcomes nobody projected.

Where the contenders actually stand entering 2026

Ohio State's dominance in the sim tracks with where actual college football voters have the Buckeyes: at or near preseason No. 1 in early polls and prediction models, largely because Jeremiah Smith returns as the best wide receiver in the country and Julian Sayin, a 2025 Heisman finalist who completed better than 77 percent of his passes last season, is back at quarterback. Ohio State went unbeaten in the regular season a year ago before dropping its final two games, and Ryan Day now hands his offense to a new coordinator in Arthur Smith while second-year defensive coordinator Matt Patricia tries to rebuild a defense that lost significant pieces.

Miami's 13 championships in the sim line up with a program that made the actual national title game last season under Mario Cristobal before falling to Indiana, then went out and added Duke transfer quarterback Darian Mensah, running back Mark Fletcher and wide receiver Malachi Toney to a roster that was already stacked. If the model likes continuity plus an obvious quarterback upgrade, Miami checks every box.

Then there's Oklahoma, which the simulation had winning it all 17 times, more than every team except Ohio State. That's the most interesting result in the entire exercise, because it doesn't fully match how humans are talking about the Sooners right now. Brent Venables has a defense that finished third nationally against the run and led the country in sacks last season, and quarterback John Mateer is back after a season interrupted by a thumb injury. But Oklahoma's offense ranked 92nd nationally in total yards, and the program's floor and ceiling in most human previews are described in the same breath, from an SEC title run to an eight-win disappointment if the offense stalls again. The game may be crediting Oklahoma's defensive talent and Mateer's ceiling more heavily than it's discounting last year's offensive inconsistency.

Indiana, the actual defending national champion off a 16-0 season, won zero of the 100 simulated titles, which is its own small data point on how EA's ratings can lag behind a program that just outperformed every projection model that existed twelve months ago.

What the model seems to reward, and what it might be missing

Looking at the full list of winners, a pattern emerges. Ohio State, Oklahoma, Miami, Ole Miss and Southern California combined for 71 of the 100 titles, and every one of those programs enters 2026 with either a returning star quarterback, a major transfer addition at the position, or both. That's not a coincidence. Dynasty-mode engines lean heavily on individual player ratings, and quarterback play tends to be the single most heavily weighted variable in any football simulation, video game or otherwise. Programs that made a marquee quarterback move in the portal, like Miami with Mensah, are going to overperform in a sim relative to how cautious human beta writers are about unproven fits.

What the model can't fully account for is exactly what tripped it up last year with Penn State: locker room chemistry, coaching seat pressure, injury luck, and the randomness of an expanded 12-team playoff bracket that has already produced two straight years of results nobody called in July. Indiana's run to the title happened with a coach in Curt Cignetti who the sim likely undervalued relative to his actual program-building, and a roster that outperformed its component-level ratings as a group. Software is good at grading talent. It's still catching up on grading teams.

Players to watch across the favorites

Jeremiah Smith is the flashpoint for Ohio State's title odds. He enters his junior year as arguably the best receiver in the country, and if he stays that dominant, he covers for a lot of defensive uncertainty. Julian Sayin's development as a full-time starter matters just as much, given his 2025 Heisman finalist run came with a completion percentage that led the sport. On the Oklahoma side, Mateer's health is the swing factor. He threw for over 3,100 yards and 29 touchdowns in his first season as a starter before the injury changed the offense's trajectory down the stretch, and a fully healthy version of him is the difference between the Sooners' floor and ceiling. Miami's addition of Mensah gives Cristobal a quarterback with real starting experience behind an offense that already had explosive playmakers in Fletcher and Toney.

Questions the sim can't answer

Can Ohio State's rebuilt defense hold up against a schedule that includes road trips to Texas, Indiana, Iowa, Nebraska and USC? Can Oklahoma's offense finally match its defense across a full season instead of in flashes? And does Indiana's actual returning production do a better job proving the doubters wrong a second time than any 100-simulation run gave it credit for? Those are the questions that will decide the real season, and none of them show up cleanly in a roster rating.

What it means heading into August

None of this settles anything, and it isn't supposed to. What the simulation actually offers is a rough sketch of where the sport's talent is concentrated heading into fall camp, filtered through an engine that rewards quarterback play and returning production above almost everything else. Ohio State's 22 titles and Miami's 13 both track with where actual college football opinion sits right now. Oklahoma's 17 is the outlier worth watching, either an early tell on a defense-driven title run or a reminder that video games and voters read offensive inconsistency very differently.

The bottom line

Simulations are entertainment dressed up as analysis, and this one is no exception. But buried in 100 randomly generated seasons is a genuinely useful exercise in spotting what a rating system values, and just as importantly, what got left out entirely. Last year that blind spot was an undefeated Indiana team. This year, keep an eye on whether it's Oklahoma's offense, or something nobody's simulated yet.