Testing "the luck" of AI in sports betting #4 update (February 23, 2025)

[Strongkored]

Given the results, the AI achieved 3 out of 6 correct predictions, with a success rate of 50% for this prediction.

This confirms that AI does not increase our winning rate, for example betting on a single bet with $10 each bet, we have lost $30 on 3 losing bets and 3 winning bets do not cover the losses suffered, because the odds of Arsenal around @1.35, Manchester City @2.00 and Fulham @ 2.27
Bettors may still use AI but still choose which matches to choose not all of them are chosen to bet on, for example only on the Manchester City match with fairly good odds because their chances of winning are still open even though their game is unstable.

[mcdouglasx]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

[Ziskinberg]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

50% for a start,  that should keep you going since the results are okay, even if they don't bring profit. Since this is just an experiment with no real money at risk, I think one month of testing is enough to prove whether this will work or not. (what's your plan about the timeframe of your experiment by the way?)

[DaNNy001]

3/6 isn't really impressive enough because you probably would have done better if you predicted it by yourself... A lot of people need to realize that Ai cannot be accurate with some certain things most especially football predictions... Believing that you have an edge over the bookies with this ai strategy means that you think that ai can predict the future Which is absolutely impossible..Ai was programmed to give out informations that it's capable of giving but it doesn't mean that all are authentic..when it comes to football predictions you must learn to rely on your knowledge.

[EarnOnVictor]

I asked Copilot for predictions for tomorrow's Premier League matches acting as an expert. Just sharing out of curiosity to see if tomorrow will be a lucky day for the AI, and also a bit because of the topic of luck in sports betting.
And this is what it replied:

copilot:

Sure! Let's provide expert predictions for the matches scheduled on February 15, 2025. While I don't have access to events beyond my last update in October 2023, we can speculate based on historical performances and current trends.

Leicester City FC vs. Arsenal
Prediction: Arsenal Victory
-snip-

My friend, my advice: don't rely on this, the AI will always give you the same answer as you would have concluded upon by yourself if you have the right information, and this will also align with the ones we see in popular places online in 1x2 betting.

But this doesn't always work unless luck is on your side. And the teams with the draw possible outcomes are those the AI believe have the same feat judging by their past performances, but they are all not a good indication of what will happen in the future.

Therefore, even if the AI wins more this week, it may fail next week, there's no guarantee in such predictions.

[mcdouglasx]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

50% for a start,  that should keep you going since the results are okay, even if they don't bring profit. Since this is just an experiment with no real money at risk, I think one month of testing is enough to prove whether this will work or not. (what's your plan about the timeframe of your experiment by the way?)

Just out of curiosity, I'll do it for about 5 weeks, on Saturdays and Sundays, to better see and evaluate its 'luck.' I still don't know whether to call it precision or performance. So far, it got the first game right with the draw and the score. So today we can say that the prediction started on the right foot. We'll see how the round ends.

[Patikno]

I think the AI ​​prediction for each upcoming match is not very accurate, and I think the AI ​​does not have specific data about the latest data that has occurred, such as players who have just been injured, the latest data from each player, and others that are so specific. I only see a general explanation of the AI ​​analysis. Therefore, I think it is better to have a personal analysis than to use AI to predict in a match. Even if it is just to try, it is better to be careful and not to involve such a large amount in betting.

That's good, using general prediction on matches will advance you on prediction analysis instead of using AI, because AI is a programm which it can not know everything and it can predict its matches base on the previous performance of the clubs, it doesn't really give the data analysis of any matches as you said,  for the injured players, so instead of me to use AI I rather use my own understanding of the team to prediction match against their opponent, already if you want to get close range prediction, I think we have check from their previous performance and also the players of the team that's evolved.

Yeah that is what I am talking about, it should be better for us to analyze any match by ourselves and not using AI for it. And personally I prefer to see news pages about matches or data that occurs in each team to do analysis, because this is very updated and can be used to predict upcoming matches. The pages that I usually use are usually sites that often provide quite complete information. Some of them are skysports, and bbcsport.

[swogerino]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

50% for a start,  that should keep you going since the results are okay, even if they don't bring profit. Since this is just an experiment with no real money at risk, I think one month of testing is enough to prove whether this will work or not. (what's your plan about the timeframe of your experiment by the way?)

Just out of curiosity, I'll do it for about 5 weeks, on Saturdays and Sundays, to better see and evaluate its 'luck.' I still don't know whether to call it precision or performance. So far, it got the first game right with the draw and the score. So today we can say that the prediction started on the right foot. We'll see how the round ends.

It can be a very fruitful experiment and it can give us the insights needed to evaluate if this AI "luck" is better than most of these tipping sites that are even asking for money for their Premium services at least some of them. The fact that AI is doing relatively well in finding the good results is a good sign that we can improve our "predicting" power. We can also try with very little amount of money like 0.10 dollars bet on each of the games that we want to try as everyone of us can afford to spen 1-5 dollars for such an experiment which we all need to see the results, I know we can check the results for free yet it is even better when we place a small bet amount on such games as it adds the adrenaline while doing so. Still it is good that AI has started giving out results.

[fullhdpixel]

As long as gambling is concerned no too ways about it, is just a game of luck. nobody is perfect when it comes to gambling, we only do the predictions and live the rest for luck, because no matter how you do what would happen would still going to happen. So relying on AI prediction might not likely be the best because gambling is not something AI can predict the final outcome, although is whatt trying, since you have already made up your mind.

Nobody is perfect in all aspects of life, so that includes gambling. And as you said, gambling is mostly about luck, so that makes the majority of the gamblers to make a mistake here. As long as we already started betting, yeah that any outcomes can already happen to us but if we don't want a bad thing in gambling, we still can prevent it from happening if we never started it in the first place.

I can feel that AI's are being improved from time to time. That being said, maybe they are now or soon to become better than us in some aspects like on sports betting for example, because we may not know what teams to pick or we are simply lazy of doing it but them (AI's) can do it immediately.

[Kemarit]

3/6 isn't really impressive enough because you probably would have done better if you predicted it by yourself... A lot of people need to realize that Ai cannot be accurate with some certain things most especially football predictions... Believing that you have an edge over the bookies with this ai strategy means that you think that ai can predict the future Which is absolutely impossible..Ai was programmed to give out informations that it's capable of giving but it doesn't mean that all are authentic..when it comes to football predictions you must learn to rely on your knowledge.

In paper it looks good as it is 50/50, but if you consider if he bet on an average of $10.00, he could have lost because those AI predictions that won might not cover the lost that he got so it's a bad bet for him.

And again, it shows that AI is still very far from predicting the result in any sports. I don't know if there will be a time that it will have the analysis of a human to see some intangibles in the game. Or even our gut feeling as a sports fan, not looking on all the data and numbers, that something might happen in the game or that a huge underdog might pull a upset that day.

So I do agree that it's better to rely on our knowledge and experience in sports betting rather than AI.

[Odusko]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

I have seen a lot of forum users trying to prove some fact for fun, just like the trading competition and now AI-aided game's analysis and selection, I haven't read through your previous update entirely but quite sure of one fact AI predictions will not be accurate at all time and for that you shouldn't depend on it so much, and newbies also should be warned to avoid over-dependency on AI in gambling.
Clarity needs to be given all the time and sound warnings need to be served out too, gambling has its unique ways and loss is one of the prominent experiences in gambling since the casinos are out for business and putting their interest first, but then also for the help of tools like the AI one can manage to have a little advantage sometimes, so let see how this fun ends to keep it up, mate.

[o48o]

3/6 isn't really impressive enough because you probably would have done better if you predicted it by yourself... A lot of people need to realize that Ai cannot be accurate with some certain things most especially football predictions... Believing that you have an edge over the bookies with this ai strategy means that you think that ai can predict the future Which is absolutely impossible..Ai was programmed to give out informations that it's capable of giving but it doesn't mean that all are authentic..when it comes to football predictions you must learn to rely on your knowledge.

Imho it's even worse then that, because you are counting winners, and AI predicted results with goal scores, and every single one of those were wrong.
But frankly, if it would get 3/6 games correctly with right goal results, i would have called it impressive, now it's below average of what i usually get by placing bets.

I have tried this same technique with UFC, and while it was interesting and fun experience, i can't underline enough that whole "AI" is just a language model, and even the data it's getting for comparing isn't going to be accurate in many cases.

[bitterguy28]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

50% for a start,  that should keep you going since the results are okay, even if they don't bring profit. Since this is just an experiment with no real money at risk, I think one month of testing is enough to prove whether this will work or not. (what's your plan about the timeframe of your experiment by the way?)

in a pure experiment stand point, yes 50% is okay but i doubt it will always be 50% there is a possibility that it decreases sometimes or sometimes it increases i do not think this is stable and reliable at all

for an ai though this is not that impressive since many people hold ai in a higher regard but good experiment to show that ai is not a know it all

[taufik123]

-snip-
I have tried this same technique with UFC, and while it was interesting and fun experience, i can't underline enough that whole "AI" is just a language model, and even the data it's getting for comparing isn't going to be accurate in many cases.

This is because some AIs only take limited data, such as in free chatGPT which only gets data until June 2024 and for real-time data that requires the help of web searches or is only limited to news on some websites and is also not very accurate.

Like OPs who use ChatAI Copilot, their data is always updated following developments from various available sources.

I tried to use several ChatAIs such as ChatGPT, Deepseek, Gemini and Copilot to predict the score of the Spanish League Barcelona vs Rayo Vallecano match which will be held on Tuesday, 18/02/2025 at 03:00 WIB, each AI has their own data and predictions, depending on where they take the data.

ChatGPT


Gemini


DeepSeek


Copilot

[freedomgo]

in a pure experiment stand point, yes 50% is okay but i doubt it will always be 50% there is a possibility that it decreases sometimes or sometimes it increases i do not think this is stable and reliable at all

for an ai though this is not that impressive since many people hold ai in a higher regard but good experiment to show that ai is not a know it all

If you want to win, you don't need just 50%, you need more than that, but since it was only the first day; we'll have to wait for more days to see its average winning rate. So eventually if it hits more than 50% (which I doubt, sorry) then I'm sure many gamblers will start using AI and might even ask the OP what specific AI he is using and what kind of subscription he has.

[Ziskinberg]

I've updated the post to continue putting the AI predictions to the test; let's see what happens today. Just to clarify, this is purely for fun—I don't bet money on these predictions. No matter how attractive it may seem, I don't recommend anyone to follow these predictions as absolute truths.

50% for a start,  that should keep you going since the results are okay, even if they don't bring profit. Since this is just an experiment with no real money at risk, I think one month of testing is enough to prove whether this will work or not. (what's your plan about the timeframe of your experiment by the way?)

Just out of curiosity, I'll do it for about 5 weeks, on Saturdays and Sundays, to better see and evaluate its 'luck.' I still don't know whether to call it precision or performance. So far, it got the first game right with the draw and the score. So today we can say that the prediction started on the right foot. We'll see how the round ends.

That's great! I'm as excited as you are to see how this experiment turns out, because I think none of us here have tried it before. Although I get why most of us believe AI won't work in our favor in sports betting, for a curious mind like OP, he's here to prove it for himself and share his findings with us. I guess, good luck for now.

[mcdouglasx]

The following matches have been updated for the matchday on February 22, 2025, in LaLiga.

So far, Copilot's AI predictions have left much to be desired. We will see how it evolves as the matchday progresses:

1-Deportivo Alavés 1 - 1 RCD Espanyol.

2-Rayo Vallecano 1 - 2 Villarreal CF.

3-Valencia CF 0 - 1 Atlético Madrid.

4-Las Palmas 0 - 3 FC Barcelona.

[mcdouglasx]

The following matches have been updated for the matchday on February 22, 2025, in LaLiga.

So far, Copilot's AI predictions have left much to be desired. We will see how it evolves as the matchday progresses:

1-Deportivo Alavés 1 - 1 RCD Espanyol.

2-Rayo Vallecano 1 - 2 Villarreal CF.

3-Valencia CF 0 - 1 Atlético Madrid.

4-Las Palmas 0 - 3 FC Barcelona.

LaLiga matches  for February 22, 2025 result: 3/4 75%

1-Deportivo Alavés 0- 1 RCD Espanyol

2-Rayo Vallecano 0 - 1 Villarreal CF.

3-Valencia CF 0 - 3 Atlético Madrid.

4-Las Palmas 0 - 2 FC Barcelona.


Here are the updated predictions for today's LaLiga matches on February 23, 2025


1- Real Madrid CF 3 - 0 Girona FC

2- Getafe CF 1 - 1 Real Betis

3- Real Sociedad 2 - 0 CD Leganés