Created a MMA fight Picker GPT

[austinblair]

Welp that sucked. I think I will hold off on posting more predictions till I think I have a model that works better. Sorry guys I'll keep trying.

I think it was closer early on

 

[GSP_Iceman]

Welp that sucked. I think I will hold off on posting more predictions till I think I have a model that works better. Sorry guys I'll keep trying.

Your model will never be 100% right, the goal is to even be 65% correct and you will achieve this through iteration and tweaking. I've stopped and started multiple times with data collection/fight odds tracking. You need tons of data and time. Don't stress about accuracy right now the idea is what matters and how to operationalize it

 

[theincognito]

It works/worked great for fights that have known fighters. It can't ever be on that level for these cards that have a lot of unknowns. There's just nothing to go by. That's where most of the losses are coming from.

Maybe start excluding fights if the fighters don't have at least 3+ fights in major promotions or something.

 

[TSaL]

Your model will never be 100% right, the goal is to even be 65% correct and you will achieve this through iteration and tweaking. I've stopped and started multiple times with data collection/fight odds tracking. You need tons of data and time. Don't stress about accuracy right now the idea is what matters and how to operationalize it

I'm shotting for 90%

 

[TCE]

Since subscribing to this again, it's done well for me the last two events.

This event it nailed Fiorot by decision, Njokuani and Jandiroba winning, who were underdogs.

The most crazy one though it had Weidman winning. Put it through 5 times, and it always said Weidman.

 

[DoctorTaco]

Since subscribing to this again, it's done well for me the last two events.

This event it nailed Fiorot by decision, Njokuani and Jandiroba winning, who were underdogs.

The most crazy one though it had Weidman winning. Put it through 5 times, and it always said Weidman.

Keep’em coming. I’m invested

 

[Georges Hefner]

It should work for Ufc 300. Considering how many established fighters are competing

 

[Tweak896]

Blanchfield/Fiorot wasn't posted? Was curious how accurate it was, the card was kind of wild with result from fouls which probably throw a wrench in predictions.

 

[GSP_Iceman]

I'm shotting for 90%

Good on you. The challenge is momentum, these cards can swing sometimes and around the 100, 200, 300 events the cards get paper thin with anonymous fighters. Stick with it and see where it goes, I'm personally working on just taking notes of every fighter and every fight I watch, lots of work to get started but I feel like I've identified a few fighters that hopefully pay off in gambling down the line. At minimum your bot has been a good benchmark to identify value bets.

 

[cburm]

Predicted Winner: Georges St-Pierre (GSP)
Expected Round: 3
Manner of Victory: Decision

Detailed Breakdown for a 3-Round Fight:​

Georges St-Pierre (GSP):​

1. Striking Accuracy: 75 - Maintains precise striking, effective in shorter bouts.
2. Wrestling Ability: 95 - Dominant wrestling, can control opponents effectively in a 3-round fight.
3. BJJ Skills: 88 - Strong defensive skills, less submission risk in a shorter fight.
4. Cardio and Stamina: 90 - High energy levels, crucial for a 3-round high-paced fight.
5. Fight IQ: 98 - Excellent adaptability and strategy formulation for shorter fights.
6. Weight Cut: 82 - Efficient weight management, advantageous in a non-championship fight.
7. Reach: 76 - Reach advantage remains constant, beneficial in striking exchanges.
8. Training Center/Team: 90 - Consistent training quality, ensures well-rounded preparation.
9. Recent Performance: 85 - Stable performance, with an emphasis on strategy in shorter fights.
10. Mental Strength: 94 - High resilience and focus, key in a high-pressure, shorter bout.

Anderson Silva:​

1. Striking Accuracy: 92 - Exceptional striker, but less effective against GSP's style.
2. Wrestling Defense: 75 - Competent, but historically struggled against top-level wrestlers.
3. BJJ Skills: 90 - Excellent offensive BJJ, but less likely to be a factor against GSP.
4. Cardio and Stamina: 85 - Good, but slightly less impactful in a 3-round fight.
5. Fight IQ: 90 - Highly experienced, but sometimes less adaptable in the face of GSP's strategy.
6. Weight Cut: 78 - Generally well-managed, but less critical in a non-title bout.
7. Reach: 84 - Advantageous, but mitigated by GSP's fighting style.
8. Training Center/Team: 87 - Strong team, but slightly less effective in preparing for GSP's style.
9. Recent Performance: 80 - Solid performances, though age and activity level are factors.
10. Mental Strength: 89 - Highly experienced, but potentially less adaptable in this specific matchup.

In a 3-round non-championship fight, GSP's wrestling and strategic approach give him a significant advantage, especially in controlling the pace and position of the fight. Silva's striking and BJJ skills are formidable, but GSP's ability to dictate where the fight takes place is likely to neutralize these strengths.

If they fought today*****

Silva struggled vs wrestlers? He's beaten everyone he's faced .. finished.. subbed chael off his back..

 

[Yep]

Just need to create one that you sell and it tells half the people one person wins and the other half the other person. You win 100% of the time. Genius

 

[WoozyFailGuy]

Silva struggled vs wrestlers? He's beaten everyone he's faced .. finished.. subbed chael off his back..

After getting the shit kicked out of him for 4.5 rounds. It was two minutes away from the MW GOAT losing the most lopsided decision in title fight history. The judges had the fight scored 40-36, 40-35, and 40-34 through the fourth round. Two judges had round 1 as 10-8, and one had Round 3 as 10-8 also.

 

[RockyLockridge]

I'm shotting for 90%

That’s too extreme in a sport where anything can happen

 

[TCE]

This is what I got from AI on tonight's card:

Allen by Sub R3.
Jackson by Sub R3.
Charrière by UD.
Bahamondes by UD.
Walker by KO R1.
Peek by KO R2.
Morono by UD
de Randamie by UD.
Hugo by Sub R1.
Rodriguez by UD.
Matsumoto by UD.
Almeida by UD.
Mullins by UD.

 

[TSaL]

I'll leave what I have created out there. I'm waiting for GPT4.5/5 to drop before working on new models. Experimenting with other LLM but nothing solid yet.

 

[TCE]

AI UFC 300 predictions

What's really interesting about this, I ran the Holloway/Gaethje fight through 5 times, and every single time it said Max by decision.

I'm not so sure, I think Justin gets the late TKO. But AI has proven me wrong many times before. I'm highly interested in this fight.

 

[High Test With da Best]

AI UFC 300 predictions

What's really interesting about this, I ran the Holloway/Gaethje fight through 5 times, and every single time it said Max by decision.

I'm not so sure, I think Justin gets the late TKO. But AI has proven me wrong many times before. I'm highly interested in this fight.

Thanks for posting this! With the model's success with known fighters (as has been mentioned) I'm going to check out how it does Saturday. I don't do any betting though so it's just out of curiosity

 

[DoctorTaco]

@TSaL and his magic robot putting some money in my pocket!

 

[TCE]

AI UFC 300 predictions

What's really interesting about this, I ran the Holloway/Gaethje fight through 5 times, and every single time it said Max by decision.

I'm not so sure, I think Justin gets the late TKO. But AI has proven me wrong many times before. I'm highly interested in this fight.

Not bad at all.

 

[randomg1t]

lol that's extremely impressive, and it came within one second to get the max fight perfectly, too.