Who will win the 2025 Men’s NCAA Gymnastics Championships?

By SORA AI.

I could make this short and say the podium will be some combination of Michigan, Stanford, and Oklahoma, with Nebraska having an outside shot, as that seems to be the conventional wisdom. But I won’t because sometimes the conventional wisdom is wrong, and time and chance happen to them all. So, let's find out with some numbers.

I used this data Spreadsheet of all team D and E scores this season . See the first sheet for an explanation of the color code. I used ChatGPT to help me with the data analysis and visualizations. I did have to use judgment to discard some of the improbable (Army beating Michigan) and the impossible (Cal being the seventh team in the final). Something to keep in mind when AI judging comes along. Judges won’t be able to just sit back with a cold one and let the AI do all the work, because it does hallucinate.

The first question is who will make to the final. I ran a Monte Carlo simulation using every score from throughout the season to answer the question “How often would each team place 1st, 2nd, or 3rd if the meet was held 10,000 times?”

The simulation predicts Oklahoma, Nebraska, and Penn State having by far the most probability of advancing to the final out of session one, and Michigan, Stanford, and Illinois advancing to the final out of session 2. Because of their seeding, Oklahoma and Nebraska are almost guaranteed a final spot.

California has a far outside chance of advancing. They almost did last year, but that was with a team of veterans. This year’s California is a young team, so advancing to the final is a very long shot.

The interesting case (as predicted by most people) is the Illinois-Ohio State matchup. They are so close together that the simulations I ran on other days have Ohio advancing over Illinois.

The final prediction looks, well, interesting. Who among us really thinks Stanford is going to finish fourth? A lot of data went into this simulation, but not all of it is equally meaningful. It doesn’t account for injuries, coaches resting or judiciously competing gymnasts, stars being away at FIG World Cups, and coaches experimenting with lineups. These factors will result in lower regular season scores for teams with more depth or National Team members. For example, Illinois and Air Force got lower scores on pommel horse when Patrick Hoopes and Brandon Dang were away at World Cups. It also doesn’t account for variability in judging. Check out this excellent article by Kathi Sue Rupp of Flipped Decisions for one possible source of variability (and to understand part of the judging process).

So I attempted to correct for that. I asked for a simulation where half the scores came from conference championships over 5000 runs, weighted away meets 1.2 x home meets and added a parameter that gave some weight to teams with higher D scores. I did this for both Ohio State and Illinois making the final.

I did a simulation with conference championships counted as half the the total meets as conference championships are normally when a team will put out a line up that is closest to what they will use at Nationals. This really helps out Michigan, as they brought out their two Olympians (who if you are reading this you probably know are Paul Juda and Frederick Richard). I would expect Michigan to bring them on for the Finals.

I asked ChatGPT why Stanford was so low despite having a score close to Michigan’s. Its reply was “Stanford is more likely to either crush it or fall out of the top 3.” Stanford also did not beat both Michigan and Oklahoma at the same time in many of the simulations. My gut feeling is this is not a probable podium, but it isn’t impossible. Stanford fans can just blame the computer, and they wouldn’t be wrong, because when it’s this tight, small difference in the intial assumptions can change the results.

Also, it is giving Illinois a 1.1 % to win. That is not impossible. Remember this is based on holding a meet thousands of times. All of the teams ahead of Illinois have had bad meets, except for maybe Oklahoma and it could even happen to them on the day. So if Illinois has the meet of its life.

Ohio State has enough difficulty so that if they go 24 for 24, they could make things interesting.

I tried another way - just counting the conference championships twice as often, but not excluding earlier meets half the time, and keeping all the other assumptions the same.

In this one, Oklahoma and Michigan are identical as far as win probability. These latter two models really hurt Ohio State, as they had one of their worst meets at conference championships. Ohio State does have a lot of difficulty, and if they go 24 for 24 they could end up on the podium.

It still seems to be undervaluing Stanford. I asked for a simulation which took into account that Stanford had won the last 5 National Championships, but it barely changed the results. Keep in mind Stanford won those last 5 Championships under a different code and a different format. The 4 up 4 count/8 skill regime decreases the role of team depth and increases the role of chance. On the other hand, the simulations don’t know Khoi Young has been seen sticking vaults, or that Stanford tends to save the best for last. Also, since prelims are technically regionals, and therefore a separate meet from Finals, teams are not limited to using the same 12 gymnasts in prelims and finals.

More interesting charts!

Note Illinois pommel horse and Michigan parallel bars. Both are capable of high scores, but sometimes not. Keep an eye on those events. The units are in points.

What events are teams most likely to bomb (rotation score less than 50)? Unsurprisingly, high bar is at the top. There have been several times this season where teams have gone into the final with a lead only to lose it with mistakes on high bar. Surprisingly, the pommel horse doesn’t look too bad.

So who does AI think will win?

So, pretty much what most men’s NCAA gymnastics would say, except they might add something about Michigan having home field advantage (which exists in all sports), and Nebraska possibly being a spoiler.

This is the most uncertain championship I can remember, where 4 teams have a real shot at the title!