Looking partner with someone who understand A.I. to build the ultimate NBA model.

[Goat Milk]

I am a guy who understands matchups, down to the player. I know their moves, I know their weaknesses, I understand the coaching aspect of the game. For many years I've been hitting 55%-60% of my picks, although I've never really bet 1 unit a play and I often get carried away. I win money after every NBA season, and usually give it back on football and other things I know less about.

Here is an example of 100 plays I made from this year: https://www.sportsbookreview.com/for...ys-thread.html

I hit roughly 60% in that thread, and there was room for improvement. Now people might be saying, you should be making tons of money with those picks. I'm not going to bet 5,000 on every pick. I simply don't have that kind of money. As you can see, I started this thread like 0-4 or something. Had I bet 5,000, I'd be down 20,000 off the bat.

I would like to team up with someone who understand the next gen of data mining and AI. There are tons of models out there, but a next gen model that has my pre-game analysis put into its algorithm, along with all the other statistical factors, has never been done.

I bring my expertise about the game, you bring your expertise about A.I.

I believe we can be profitable. I don't want to hit 55 or 60%. I want to hit 65% or 70%. I want to win 7 for every 3 losses. I believe that is doable.

Message me if you're interested.

[Roger T. Bannon]

You are completely underestimating your own level of genius here. The most AI could give you is maybe 54%. You are looking to go backwards. Keep betting away at 56-60% and you will get rich. Forget AI.

[rake922]

[yak merchant]

You are completely underestimating your own level of genius here. The most AI could give you is maybe 54%. You are looking to go backwards. Keep betting away at 56-60% and you will get rich. Forget AI.

What he said. If you truly can hit 60% with more than 100 plays a year you need to start working on getting beards and finding outs. With a 10,000 starting bankroll and a true 60% win rate, just bet full Kelly and within 2 to 3 years you'll be a millionaire worried about how to get 50K down a game.

[Goat Milk]

What he said. If you truly can hit 60% with more than 100 plays a year you need to start working on getting beards and finding outs. With a 10,000 starting bankroll and a true 60% win rate, just bet full Kelly and within 2 to 3 years you'll be a millionaire worried about how to get 50K down a game.

I didn't understand what "working on getting beards and finding out" and "fully kelly".

For an even money bet, the Kelly criterion computes the wager size percentage by multiplying the percent chance to win by two, then subtracting one. So, for a bet with a 70% chance to win (or 0.7 probability), doubling 0.7 equates 1.4, from which you subtract 1, leaving 0.4 as your optimal wager size: 40% of available funds.

How would I know if a -110 pick has a 70% chance of winning? I'm not good at math.

Thank you.

[Goat Milk]

You are completely underestimating your own level of genius here. The most AI could give you is maybe 54%. You are looking to go backwards. Keep betting away at 56-60% and you will get rich. Forget AI.

Thank you. What I was hoping to do was create something AI has never done, which would make those normal %s irrelevant. It's hard to say exactly how it would work or what would happen, as I don't have the first clue about math or python.

If you have a simple strategy for how I should be betting, let me know.

How you doing Rake. Nothing like our old days chat on this place.

[Roger T. Bannon]

Thank you. What I was hoping to do was create something AI has never done, which would make those normal %s irrelevant. It's hard to say exactly how it would work or what would happen, as I don't have the first clue about math or python.

If you have a simple strategy for how I should be betting, let me know.

How you doing Rake. Nothing like our old days chat on this place.

For starters, you want to stop betting on all the sports you are losing on. This will eliminate the problem of ruining your 60% betting percentage with losers. To the extent you are unable to, you should bet the minimum amount possible.

Secondly, it is highly likely that you are overestimating your edge. So you want to start this enterprise with the goal of winning but the expectation until proven otherwise that you will lose.

So now, you will bet 1% of bankroll per bet against the spread. Should you double your money doing this, you will then have some confirmation that this plan is working and can increase to 2% of BR with some level of confidence going forward.

In all likelihood, this winning percentage will mysteriously disappear fairly soon after starting so proceed with caution.

In any case, AI is only good at projecting outcomes that can be calculated which have certain outcomes. It is not good at forecasting uncertain outcomes. So you are better off using your own abilities if you have any.

[Goat Milk]

For starters, you want to stop betting on all the sports you are losing on. This will eliminate the problem of ruining your 60% betting percentage with losers. To the extent you are unable to, you should bet the minimum amount possible.

Secondly, it is highly likely that you are overestimating your edge. So you want to start this enterprise with the goal of winning but the expectation until proven otherwise that you will lose.

So now, you will bet 1% of bankroll per bet against the spread. Should you double your money doing this, you will then have some confirmation that this plan is working and can increase to 2% of BR with some level of confidence going forward.

In all likelihood, this winning percentage will mysteriously disappear fairly soon after starting so proceed with caution.

In any case, AI is only good at projecting outcomes that can be calculated which have certain outcomes. It is not good at forecasting uncertain outcomes. So you are better off using your own abilities if you have any.

Ok, thank you. I know I can hit between 55-60 as I've been doing so for many years. I will try and stay away from all other sports and go into each game betting the same amount. I'll start with 2% next season, and will give you my results.

Thanks.

[Roger T. Bannon]

Ok, thank you. I know I can hit between 55-60 as I've been doing so for many years. I will try and stay away from all other sports and go into each game betting the same amount. I'll start with 2% next season, and will give you my results.

Thanks.

That's highly doubtful. If you have not actually tracked your bets, you do not know what your winning percentage has been. But you can fairly easily determine your likelihood of winning going forward with your CLV. If the line moves in your favor on 60% of your first 100 bets, you are likely to win going forward. If not, it is a red flag warning.

I would start with 1% bets on your first 100 bets as 2% bets are going to be throwing your money in a furnace if you are not as good as you believe.

[PAULYPOKER]

Finally moved it to the right forum I see.....

Lol,Bro what made you think you'd find a partner in PT of all places?

[clockwise1965]

[Goat Milk]

Finally moved it to the right forum I see.....

Lol,Bro what made you think you'd find a partner in PT of all places?

What you mean? I put this thread in think tank not PT.

[JoeCool20]

Nobody knows who is going to win any game when you add in the pointspread.

Just put your dollars out there and hope you have won when the game is over.

Until your "Algo" and "model" realizes that EVERY (pointspread) game you bet is 50-50 before the game starts,

then you need to keep plugging in different stats until your "model" realizes that, and comes to that conclusion.

[Goat Milk]

Nobody knows who is going to win any game when you add in the pointspread.

Just put your dollars out there and hope you have won when the game is over.

Until your "Algo" and "model" realizes that EVERY (pointspread) game you bet is 50-50 before the game starts,

then you need to keep plugging in different stats until your "model" realizes that, and comes to that conclusion.

Why is it 50-50 if you have information that vegas doesn't?

[Goat Milk]

That's highly doubtful. If you have not actually tracked your bets, you do not know what your winning percentage has been. But you can fairly easily determine your likelihood of winning going forward with your CLV. If the line moves in your favor on 60% of your first 100 bets, you are likely to win going forward. If not, it is a red flag warning.

I would start with 1% bets on your first 100 bets as 2% bets are going to be throwing your money in a furnace if you are not as good as you believe.

I've tracked bets on this site in the past. The most recent was the link that was posted. That was all tracked for 100 bets, so it's not like I'm trying this for the first time. I've been on this site for 10 yrs.

[Roger T. Bannon]

I've tracked bets on this site in the past. The most recent was the link that was posted. That was all tracked for 100 bets, so it's not like I'm trying this for the first time. I've been on this site for 10 yrs.

Yes. Well, that's great. But probably not all that many bets. If you have not won overall, it is unlikely you are sharp at NBA. If you are going to get serious about your betting, I'd just start off on the right foot.

[RedApples]

Yes. Well, that's great. But probably not all that many bets. If you have not won overall, it is unlikely you are sharp at NBA. If you are going to get serious about your betting, I'd just start off on the right foot.

+1. lol @ 100 bets. talk about small sample size. dumb thread.

[cashin81]

" although I've never really bet 1 unit a play and I often get carried away. I win money after every NBA season, and usually give it back on football and other things I know less about."

You could 80% but still if you are prone to get carried away, whats the point?
Alarm bells for anyone wanting to help you build a spaceship

[Goat Milk]

" although I've never really bet 1 unit a play and I often get carried away. I win money after every NBA season, and usually give it back on football and other things I know less about."

You could 80% but still if you are prone to get carried away, whats the point?
Alarm bells for anyone wanting to help you build a spaceship

That's the point. To build a mathematical model that keeps me more grounded.

[Goat Milk]

Yes. Well, that's great. But probably not all that many bets. If you have not won overall, it is unlikely you are sharp at NBA. If you are going to get serious about your betting, I'd just start off on the right foot.

+1. lol @ 100 bets. talk about small sample size. dumb thread.

Not anywhere near 100 bets. Like I said 10 years on this site.

Vegas oddsmakers do not know the intricacies of the game of basketball like I do, and neither does anyone in this think tank. And I don't know the intricacies of stats like anyone in here. That's was my purpose for coming to this thread.

Keep an eye on my work going forward and you might learn something about the league. If you decide to reconsider, you know where to find me.

[Vyasports]

Finally moved it to the right forum I see.....

Lol,Bro what made you think you'd find a partner in PT of all places?

thats what i wast thinking the whole time. Trust is important in partnership like this...

[Roger T. Bannon]

Not anywhere near 100 bets. Like I said 10 years on this site.

Vegas oddsmakers do not know the intricacies of the game of basketball like I do, and neither does anyone in this think tank. And I don't know the intricacies of stats like anyone in here. That's was my purpose for coming to this thread.

Keep an eye on my work going forward and you might learn something about the league. If you decide to reconsider, you know where to find me.

So the trouble here is that the guys that model and would have an ability to have an AI-based program (which does not exist) do not believe that the intricacies that you believe you are able to cash in on exist. They reject this concept entirely. So you would not have anyone team up with you anyhow. And you are far better off with your own intricacies if you are half as good as you think.

So good luck with your bets going forward.

[VinagreSantos]

Hello;

1) 100 positions/bets are not enough to sustain model´s performance. Although position size is fundamental, i still don´t trust in a "normal distribution way" to analyze returns. So, you need way more "bets".

2) You speak about "matchups", weakness, coaching.etc. One of the true challenges on AI, is to transform concepts into variables. From your knowledge to a well defined variable to use as an input in a Neural Net (for example), it takes a lot of work and research. Things look easy and similar on paper, but then you go into their nature "in the numbers" and things change and they usually end up being quite complicated.

3) AI is very competitive. There´s always new ideas. Years a go everybody was thrilled with LSTM´s, now there´s people using other tools with a lot of success. Next Gen today could be weak or insufficient in the future.

4)The idea of creating a "next gen model" goes into 2 directions. Or you create a new architecture (which is a monumental task) or you end up using one that´s being used.

5) If you really have concepts/constructs that can be perform very well, be careful with who you are sharing information. Information is key to create "value".

6) I think you should first, navigate a little bit in more simple "solutions" Alone. For example try to study about different types of variables, then start to learn about correlation and regression, for example. Both are insufficient, but they will start to show you the "way of thinking".

Good luck!