• 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

Try asking it for the outcome on an upcoming fight over consecutive days.

It changes its probability wildy every time for me.

Ironically one of the ones that it has maintained the 3 times I've asked it is Imavov (60% chance to win) vs Dolidze, but I see it is telling you it has Dolidze at 60% 😂.

It told me Thomas Peterson was a 65% chance to win and likely by SUB in the 2nd round, on the first day; then Thomas Peterson was a 75% chance to win and likely by KO in the 2nd round, on the second day; and then yesterday it said his opponent, Pogues was a 55% chance to win and likely by TKO in the 1st round.

Doesn't seem like something I would be putting a lot of confidence in at this stage, although I will say it seems incredibly intelligent when asking it to explain its reasoning on why it expects a fighter to win, etc.

Yeah, this tool is definitely in its early stages, but I reckon this could be something big once it's ironed out properly.
 
Yeah, this tool is definitely in its early stages, but I reckon this could be something big once it's ironed out properly.
Just odd that OP had picks from the previous event and some from the one before and they all were super accurate, but then when a random user tries to use it it spits out inconsistent predictions 🤷‍♂️.
 
Well I went in and paid for the plus.

Been having a lot of fun with it.

I also side with these results.


Fight Analyst Advanced

Fight Analysis: Chris Barnett vs. Mohammed Usman​

Scheduled for an upcoming UFC Fight Night, this heavyweight bout features Chris Barnett (23-8, 18 KOs) against Mohammed Usman (20-4, 9 KOs).

Chris Barnett Profile:​

  • Strengths: Barnett, also known as "Beastboy," is known for his striking power. He has a 49% striking accuracy with a significant strikes landed per minute (SLpM) of 4.41 and absorbs 5.87 significant strikes per minute (SApM). Barnett has an 18 wins by KO/TKO, showcasing his finishing ability.
  • Recent Performances: Barnett has shown his power in the heavyweight division with a recent victory over Jake Collier and a notable finish against Gian Villante.

Mohammed Usman Profile:​

  • Strengths: Usman, "The Motor," has a balanced skill set with a focus on striking. He has a 40% striking accuracy with an SLpM of 2.92 and absorbs 3.96 significant strikes per minute. Usman also has a 19% takedown accuracy and has shown knockout power with 4 wins by KO/TKO.
  • Recent Performances: Usman has been on a winning streak, with recent victories over Jake Collier and Junior Tafa, as well as a knockout win in the TUF 30 heavyweight final.

Analysis and Prediction:​

  • Striking: Barnett has a higher output in striking and more knockout finishes, suggesting an advantage in striking power. However, Usman’s striking accuracy and defense are notable.
  • Grappling: While neither fighter heavily focuses on grappling, Usman’s slightly better takedown accuracy could play a role if the fight goes to the ground.
  • Physicality and Stamina: Both fighters have shown they can handle fights going the distance, but Barnett's higher rate of first-round finishes indicates explosive power early in the fight.
  • X-Factors: Barnett's experience and unique striking style could surprise Usman, while Usman's recent form and balanced approach make him a formidable opponent.

Final Prediction:​

  • Predicted Winner: Chris Barnett
  • Confidence Level: 55%
  • Predicted Round: Round 2
  • Method of Victory: KO/TKO

Additional Considerations:​

  • Weight Class: Heavyweight. In this division, knockout power is often a decisive factor.
  • Scoring System: The emphasis is on effective striking and grappling. Barnett’s striking could be decisive if he maintains his high output.
  • External Factors: Training camp quality, recent injuries, and weight cut impacts are unknown but can significantly affect performance.


Fight Analysis: Brad Tavares vs. Gregory Rodrigues​

Scheduled for February 10, 2024, at UFC Fight Night, this middleweight bout features Brad Tavares (20-8, 5 KOs) against Gregory Rodrigues (14-5, 9 KOs).

Brad Tavares Profile:​

  • Strengths: Tavares is an experienced fighter with a well-rounded skill set. He has a 43% striking accuracy and a 27% takedown accuracy. His significant strikes landed per minute (SLpM) is 3.38, and he absorbs 3.03 significant strikes per minute (SApM). Tavares also has a 57% significant strike defense and an impressive 82% takedown defense.
  • Recent Performances: Tavares has had a mixed record in his recent fights, with notable victories over Chris Weidman and Omari Akhmedov, but losses to the likes of Robert Whittaker and Israel Adesanya.

Gregory Rodrigues Profile:​

  • Strengths: Rodrigues, known as "RoboCop," is a powerful striker with a grappling background. He has a 56% striking accuracy and a 50% takedown accuracy. His SLpM is an impressive 5.99, and he absorbs 5.67 significant strikes per minute. Rodrigues has a 51% significant strike defense and a 100% takedown defense.
  • Recent Performances: Rodrigues has shown his striking prowess with recent knockout victories over Denis Tiuliulin and Brunno Ferreira, demonstrating his ability to finish fights.

Analysis and Prediction:​

  • Striking: Rodrigues has an edge in striking power and output, which could be significant if he can maintain his high pace against Tavares.
  • Grappling: Tavares’ experience and solid takedown defense might help him neutralize Rodrigues' grappling threats.
  • Physicality and Stamina: Tavares' experience in longer fights might give him an edge in strategy and endurance, while Rodrigues' recent first-round finishes suggest explosive power.
  • X-Factors: Rodrigues' knockout power and recent form make him a dangerous opponent. Tavares' experience against top-level competition could play a key role in a longer fight.

Final Prediction:​

  • Predicted Winner: Gregory Rodrigues
  • Confidence Level: 60%
  • Predicted Round: Round 2
  • Method of Victory: KO/TKO

Additional Considerations:​

  • Weight Class: Middleweight. In this division, striking power and speed are often decisive factors.
  • Scoring System: The emphasis is on effective striking and grappling. Rodrigues’ striking could be decisive if he can keep up his high output.
  • External Factors: Training camp quality, recent injuries, and weight cut impacts are unknown but can significantly affect performance.




Fight Analysis: Roman Dolidze vs. Nassourdine Imavov​

Scheduled for February 3, 2024, at UFC Vegas 85, this middleweight bout features Roman Dolidze (12-2, 7 KOs) against Nassourdine Imavov (12-4, 5 KOs). Both fighters have demonstrated strong performances in their recent fights, showing a blend of knockout power and technical skills.

Roman Dolidze Profile:​

  • Strengths: Dolidze showcases a balanced fighting style with a strong grappling background and striking ability. He has a 44% striking accuracy and a 50% takedown accuracy. His significant strikes landed per minute (SLpM) is 3.02, and his significant strikes absorbed per minute (SApM) is 3.01. Dolidze has a knockdown average of 0.76 and a 55% significant strike defense.
  • Recent Performances: Dolidze's recent track record includes wins over notable opponents like Phil Hawes and Jack Hermansson, with a loss to Marvin Vettori. He's shown resilience and adaptability in his fights, often finishing opponents early.

Nassourdine Imavov Profile:​

  • Strengths: Imavov, known for his striking, has a 53% striking accuracy with a SLpM of 4.55 and a SApm of 3.76. He possesses a solid takedown defense of 73% but a lower takedown accuracy of 32%. His striking is diverse, focusing on head and body shots.
  • Recent Performances: Imavov has demonstrated his prowess with wins against Edmen Shahbazyan and Ian Heinisch, showcasing his knockout ability. His loss to Sean Strickland, however, exposed some limitations in dealing with high-level strikers.

Analysis and Prediction:​

  • Striking: Imavov has the edge in striking, both in accuracy and diversity. His higher SLpM indicates a more active striking game.
  • Grappling: Dolidze’s grappling is superior, with better takedown accuracy and a varied submission arsenal.
  • Physicality and Stamina: Dolidze has shown a tendency to finish fights early, which could be advantageous in a shorter bout. Imavov's longer average fight time suggests better endurance for a prolonged fight.
  • X-Factors: Dolidze’s experience with top-level opponents and his balanced approach might give him an edge in adapting to Imavov's striking.

Final Prediction:​

  • Predicted Winner: Roman Dolidze
  • Confidence Level: 60%
  • Predicted Round: Round 3
  • Method of Victory: Decision

Additional Considerations:​

  • Weight Class: Middleweight. The power and speed dynamics are crucial at this weight.
  • Scoring System: Emphasis on effective striking/grappling and aggression can influence the judges in a close fight.
  • External Factors: Training camp quality, recent injuries, and weight cut impacts are unknown but can significantly affect performance.
Man I just don't see Dolidze winning this.
 
Just odd that OP had picks from the previous event and some from the one before and they all were super accurate, but then when a random user tries to use it it spits out inconsistent predictions 🤷‍♂️.

Hundred percent. I mean the results of Ank KO R2, Magny KO R3, Miller sub R3, Pennington winning, etc... insane.

I thought I'd pay for the plus, but have had completely mixed results. Albeit, the fights haven't happened yet, but 50/50 every fight on the AI. I have since cancelled my plus.
 
Just odd that OP had picks from the previous event and some from the one before and they all were super accurate, but then when a random user tries to use it it spits out inconsistent predictions 🤷‍♂️.
Chat GPT-4 doesn't know anythiny beyond April 2023. So it's use for prediction of new events is limited.

Edit: apparently this is not true and it can look stuff up. That is weird because I've used the paid version to assist writing code and it would never write code for the latest versions of software despite being explicitely asked to. So I assumed it was the same with info on MMA.
 
Last edited:
Fight Prediction: Robert Whittaker vs. Sean Strickland

Fighter Profiles and Analysis:


  1. Robert Whittaker:
    • Nickname: The Reaper / Bobby Knuckles
    • Strengths: Robust striking skills, high fight IQ, strong takedown defense
    • Notable Victories: Wins over Yoel Romero (twice), Ronaldo Souza, Derek Brunson
    • Losses: Stephen Thompson, Court McGee
    • Recent Performance: Consistent with striking, experienced in 5-round fights
  2. Sean Strickland:
    • Nickname: Tarzan
    • Strengths: Aggressive striking, good chin, solid ground game
    • Notable Victories: Wins over Brendan Allen, Krzysztof Jotko, Uriah Hall
    • Losses: Kamaru Usman, Elizeu Zaleski dos Santos
    • Recent Performance: Return after a layoff, shows good striking and durability
Fight Metrics Comparison:

  • Striking: Whittaker's striking is more versatile and technical. He lands significant strikes effectively, maintaining distance control. Strickland is aggressive but can be predictable.
  • Grappling: Both fighters have good grappling defense. Whittaker has shown stronger takedown defense in past fights.
  • Physical Traits: Similar in height and reach. Whittaker has a slight edge in speed and agility.
  • Experience: Whittaker has more experience in high-stakes fights, especially 5-round bouts.
  • Recent Activity: Strickland has been more active recently, but Whittaker's experience in top-level fights is significant.
Prediction:

  • Robert Whittaker vs. Sean Strickland
  • Prediction Confidence: 70%
  • Predicted Winner: Robert Whittaker
  • Predicted Round: Round 4
  • Method of Victory: Decision
Additional Considerations:

  • Weight Class: Middleweight
  • Scheduled Rounds: Likely a 5-round fight given their standings
  • External Factors: Whittaker's experience in main events and title fights could play a key role. Strickland's recent activity provides him with momentum, but he may struggle against Whittaker's fight IQ and technical striking.

Lol it predicted a 4th round decision. Classic chatGTP.
 
I would like to clarify @TSaL that most of the fights I ran through on both different tabs confirmed the same results (winner, round and method).

But a few came up differently, and I ran those fights through again to the same results.
Yes seeing this also. I have been working on better models but need more fights to happen. The gpt is supposed to go out and pull updated information then score the metrics it find. If it can't find a metric for a fighter it will give the fighter an average score and move on. This is what it sopppsed to do. I was thinking about changing it to only print out the prediction to save on tokens and data.
 
Chat GPT-4 doesn't know anythiny beyond April 2023. So it's use for prediction of new events is limited.

Edit: apparently this is not true and it can look stuff up. That is weird because I've used the paid version to assist writing code and it would never write code for the latest versions of software despite being explicitely asked to. So I assumed it was the same with info on MMA.
Not true with mygpts and plugins. It will go out and pull data.
 
Last edited:
Fight Prediction: Robert Whittaker vs. Sean Strickland

Fighter Profiles and Analysis:


  1. Robert Whittaker:
    • Nickname: The Reaper / Bobby Knuckles
    • Strengths: Robust striking skills, high fight IQ, strong takedown defense
    • Notable Victories: Wins over Yoel Romero (twice), Ronaldo Souza, Derek Brunson
    • Losses: Stephen Thompson, Court McGee
    • Recent Performance: Consistent with striking, experienced in 5-round fights
  2. Sean Strickland:
    • Nickname: Tarzan
    • Strengths: Aggressive striking, good chin, solid ground game
    • Notable Victories: Wins over Brendan Allen, Krzysztof Jotko, Uriah Hall
    • Losses: Kamaru Usman, Elizeu Zaleski dos Santos
    • Recent Performance: Return after a layoff, shows good striking and durability
Fight Metrics Comparison:

  • Striking: Whittaker's striking is more versatile and technical. He lands significant strikes effectively, maintaining distance control. Strickland is aggressive but can be predictable.
  • Grappling: Both fighters have good grappling defense. Whittaker has shown stronger takedown defense in past fights.
  • Physical Traits: Similar in height and reach. Whittaker has a slight edge in speed and agility.
  • Experience: Whittaker has more experience in high-stakes fights, especially 5-round bouts.
  • Recent Activity: Strickland has been more active recently, but Whittaker's experience in top-level fights is significant.
Prediction:

  • Robert Whittaker vs. Sean Strickland
  • Prediction Confidence: 70%
  • Predicted Winner: Robert Whittaker
  • Predicted Round: Round 4
  • Method of Victory: Decision
Additional Considerations:

  • Weight Class: Middleweight
  • Scheduled Rounds: Likely a 5-round fight given their standings
  • External Factors: Whittaker's experience in main events and title fights could play a key role. Strickland's recent activity provides him with momentum, but he may struggle against Whittaker's fight IQ and technical striking.

Lol it predicted a 4th round decision. Classic chatGTP.
Yep this is a bug round 4 decision. Helps to specify if its a title fight
 
I'm currently working on improving the consistency of my fight predictions, especially when you ask me about the same fight multiple times. To address this, I've been proactively asking the GPT model why it gives different prediction percentages for the same fight and what I can do to provide more accurate and consistent results. Here's its response:

The difference in prediction confidence between the two analyses likely arises from variations in the evaluation of the fighters' recent performances and overall career statistics during the research phase. To enhance accuracy and consistency in future predictions, especially when analyzing the same fight more than once, you could add instructions emphasizing the importance of:

  1. Consistent Data Sources: Prioritize certain reliable sources for fighter statistics and recent performance analysis.
  2. Standardized Scoring Methodology: Define a fixed scoring system for evaluating fighters' skills, records, and recent performances.
  3. Repeat Analysis Protocol: Establish a procedure for handling repeated fight analyses to ensure consistent assessment and prediction outcomes.

So I have added this into its instructions and also a explicit the it should never have a fight ending in the 4th round decision lol.

This has not yet been published to the gpt. I will test more then publish later today if results are good.
 
I have updated the build: https://chat.openai.com/g/g-a4l5RVxni-fight-analyst-advanced

Feel free to offer more suggestions
It's not working for me at all anymore unfortunately.

I'm unable to get it to give me a breakdown on a fight - it either says "given the constraints of our current setup, I'm unable to perform this detailed analysis in real-time or access up-to-date fighter statistics and recent performance data directly.", or it will just give me a nonsensical response like information about the date of the card and when it starts, or one of the fighter's records and their bio.
 
It's not working for me at all anymore unfortunately.

I'm unable to get it to give me a breakdown on a fight - it either says "given the constraints of our current setup, I'm unable to perform this detailed analysis in real-time or access up-to-date fighter statistics and recent performance data directly.", or it will just give me a nonsensical response like information about the date of the card and when it starts, or one of the fighter's records and their bio.
Thanks I'll take a look
 
It's not working for me at all anymore unfortunately.

I'm unable to get it to give me a breakdown on a fight - it either says "given the constraints of our current setup, I'm unable to perform this detailed analysis in real-time or access up-to-date fighter statistics and recent performance data directly.", or it will just give me a nonsensical response like information about the date of the card and when it starts, or one of the fighter's records and their bio.
Working for me. Not loving the current format though:

Fight Analysis: Aljamain Sterling vs Calvin Kattar

Aljamain Sterling's Profile:


  • Record: Sterling has demonstrated versatility in his victories, including submissions and decisions. Notably, he has achieved multiple rear-naked choke wins and unanimous decision victories. His skill set is diverse, with notable submission skills.
  • Recent Performance: His recent fights show a trend of winning by submission and decisions, with a significant win over Pedro Munhoz and a notable KO against Marlon Moraes.
  • Physical and Technical Stats: Sterling's striking accuracy stands at 52%, with a significant strikes landed rate of 4.73 per minute. His takedown accuracy is 24%, and he lands about 1.97 takedowns per 15 minutes. He also has a strong takedown defense at 45%.
Calvin Kattar's Profile:

  • Record: Kattar has a solid record with a high percentage of KO/TKO wins. His striking ability is evident in his fight history, with many first-round finishes.
  • Recent Performance: Kattar has shown strong performance in recent fights, including a decisive victory over Dan Ige and an impressive KO against Jeremy Stephens. However, he also has decision losses against Zabit Magomedsharipov and Max Holloway.
  • Physical and Technical Stats: Kattar's striking accuracy is 40%, and he lands an impressive 5.12 significant strikes per minute. His takedown defense is notable at 91%, but his takedown accuracy is relatively lower at 29%.
Fight Prediction:

  • Scheduled Length: This fight is scheduled for three rounds.
  • Striking and Grappling Comparison: Sterling's grappling and submission skills might give him an edge in ground combat, while Kattar's striking and knockout power are significant factors in stand-up exchanges.
  • Physical Traits: Both fighters have comparable physical attributes, but Kattar's higher rate of significant strikes landed per minute could be a decisive factor.
Prediction Outcome:

  • Prediction: Calvin Kattar to win.
  • Confidence Percentage: 60%.
  • Predicted Winner: Calvin Kattar.
  • Round: 3rd.
  • Method of Victory: Decision.
This prediction factors in both fighters' recent performances, striking and grappling skills, and physical attributes. Kattar's striking advantage, especially in longer fights, and his high takedown defense rate, could neutralize Sterling's grappling threat and lead to a decision victory in his favor.
 
What do you mean by GPT?
You ought to define uncommon acronyms.
 
Here are my picks for tonight's fights:View attachment 1027551

Seem like a lot of subs and not many decisions. I guess we will see how it goes.

This is interesting haha my picks from the GPT tool are mainly all decisions lol.

Mine had Carolina and Petersen winning too, interested to see how this turns out for us both.

106801-3882851.png
 
This is interesting haha my picks from the GPT tool are mainly all decisions lol.

Mine had Carolina and Petersen winning too, interested to see how this turns out for us both.

106801-3882851.png
When did you generate them?
 
Back
Top