• Xenforo Cloud is upgrading us to version 2.3.8 on Monday February 16th, 2026 at 12:00 AM PST. Expect a temporary downtime during this process. More info here

Created a MMA fight Picker GPT

Ok new reasoning model for Open AI was released. I just just some simple instructions analyze fights using "reasoning" and "search". Super easy to do and it free.

Here is what it came up with for me:
I’ve fucked around with a few other AI models guys have developed to predict fights and yours is still more consistently correct
 
Ok new reasoning model for Open AI was released. I just just some simple instructions analyze fights using "reasoning" and "search". Super easy to do and it free.

Here is what it came up with for me:

Awesome.

Very interesting to see how those predictions are correct.

Here's mine:
Du Plessis by KO R3.
Suarez by Sub R3.
Teixeira by KO R2.
Bellato by KO R2.
Matthews by UD.
Jenkins by KO R2.
Nolan by UD.
Wang by UD.
Topuria by KO R2.
Steele by KO R1.
Jousset by UD.
Salkilld by KO R1.
Park by UD.
 
I would like to see someone stand up a Deepseek r1 locally with fightmatrix.com statistics. Or would like to see what Open AI Deep Research could produce (currently $200 for 30 requests I think).
 
  • Like
Reactions: TCE
what would really be interesting is to match the system's predictions with betting lines.

see if it's any better at predicting the winner than the bookies.
 
what would really be interesting is to match the system's predictions with betting lines.

see if it's any better at predicting the winner than the bookies.
This is what a guy does for about 6 months then live on the beach for the next 20 years.
 
58th Bitches on topology. Hope I don't eat my words :)
 
That's great picks
I am surprised the AI can pick up new-coming outsiders like Micallef winning
 
That's great picks
I am surprised the AI can pick up new-coming outsiders like Micallef winning
It has a hard time when there is not a lot of data on the fighter.
 
Ok with this fight card I basically did the same as last. Use generic chatgpt (with search and reasoning selected).


This time the model seemed way more confident with labeling maybe fight high confidence. see below. Ussaly never see % in the 90's

Jared Cannonier vs. Gregory Rodrigues
Winner: Gregory Rodrigues
Round: 5
Via: Decision
Confidence: 72%


Calvin Kattar vs. Youssef Zalal
Winner: Youssef Zalal
Round: 1
Via: TKO
Confidence: 83%

Edmen Shahbazyan vs. Dylan Budka
Winner: Edmen Shahbazyan
Round: 2
Via: TKO
Confidence: 95%

Ismael Bonfim vs. Nazim Sadykhov
Winner: Ismael Bonfim
Round: 2
Via: TKO
Confidence: 66%

Rodolfo Vieira vs. Andre Petroski
Winner: Rodolfo Vieira
Round: 2
Via: Submission
Confidence: 84%

Jose Delgado vs. Connor Matthews
Winner: Jose Delgado
Round: 2
Via: TKO
Confidence: 93%

Angela Hill vs. Ketlen Souza
Winner: Ketlen Souza
Round: 3
Via: Decision
Confidence: 59%

Jared Gordon vs. Mashrabjon Ruziboev
Winner: Jared Gordon
Round: 2
Via: TKO
Confidence: 71%

Rafael Estevam vs. Jesus Aguilar
Winner: Rafael Estevam
Round: 1
Via: TKO
Confidence: 83%

Gabriel Bonfim vs. Khaos Williams
Winner: Gabriel Bonfim
Round: 3
Via: Decision
Confidence: 58%

Vince Morales vs. Elijah Smith
Winner: Elijah Smith
Round: 2
Via: TKO
Confidence: 81%

Valter Walker vs. Don'Tale Mayes
Winner: Valter Walker
Round: 1
Via: TKO
Confidence: 84%

Julia Avila vs. Jacqueline Cavalcanti
Winner: Jacqueline Cavalcanti
Round: 3
Via: Decision
Confidence: 94%
 
Where this could really take off is when/if the AI can get a lot of specific information about the fighters and could make connections between events around winners and losers that a normal brain can't make.

An AI might be able to put together complex combinations of factors that shift the probabilities of wins or losses, and to make accurate predictions based on those how those factors sit.

The thing is, someone would have to get current, specific info about the fighters, and the AI would have to be more powerful, that could be coming sooner than later though.
 
Last edited:
Interesting though I think it will just give confirmation to people with very basic analytic skills. You know the type of people who love to look at stats rather than observe how the actual fights play out. The latter of which gives more detail into how fighters matchup in strengths, weaknesses, movements etc and factors in more heavily into fight prediction
 
Back
Top