CMONEY BETTING SECTION ON PFA - INTERESTED? COMPUTER BASED PICKS

[cmoney]

So I have been working with some nerds the last few months and worked out a bunch of computer simulations for different sports. We are getting into the 60 percent win rate with a large sample size. I think it would be cool to have an invite only section of the site where I posted the picks for the day and people can bet them if they want. We can also come up with theories and test them. I fully realize programs online exist that do this very thing, but I almost certain I figured out an edge with certain systems. This wont be Cmoney lock of the week where one year I luck boxed 20 in a row. As I said, we are getting, at best, 60-63 percent historically on some systems. The idea is to grind it out and not fire 10k on the first pick that posted. It could very well lose 5 or more in a row.

Anyway, thought I would throw it out there. If not enough people interested no point. I am mostly interested in people throwing out theories to test, plugging them in, and see what we get. More degen heads the better.

[sonatine]

this is funny; im literally working on a predictive alg for mma fights.

very interested.

[JimmyG_415]

I'd be up for this.
60-63 is an excellent win percentage.
57 % is the making $$$ number.

(52.5 ish to break even, about 57% to make money per @andrewssports )

[cmoney]

this is funny; im literally working on a predictive alg for mma fights.

very interested.

That is one we have not touched yet but would to dive into for sure. I always thought there was an edge there but never had a good idea for a system.

[handicapme]

I'd be interested in seeing this. Count me in.

[Charham]

I am interested. I had a few algos a number of years ago developed, thought about raising serious $ to bet them, but the returns started dropping and frankly its such a pain in the ass to manage the bets.

[MumblesBadly]

So I have been working with some nerds the last few months and worked out a bunch of computer simulations for different sports. We are getting into the 60 percent win rate with a large sample size. I think it would be cool to have an invite only section of the site where I posted the picks for the day and people can bet them if they want. We can also come up with theories and test them. I fully realize programs online exist that do this very thing, but I almost certain I figured out an edge with certain systems. This wont be Cmoney lock of the week where one year I luck boxed 20 in a row. As I said, we are getting, at best, 60-63 percent historically on some systems. The idea is to grind it out and not fire 10k on the first pick that posted. It could very well lose 5 or more in a row.

Anyway, thought I would throw it out there. If not enough people interested no point. I am mostly interested in people throwing out theories to test, plugging them in, and see what we get. More degen heads the better.

I have a background in statistical testing of predictive models, and have a question: Was the model/models to predict good bets tested using hold-out samples? This article discusses how to do so.

http://www.plottingsuccess.com/3-pre...cy-tests-0114/

Without such verification of a model, the predictive power could easily be overstated by overfitting the model to the data. And that would mean that the model will could easily have poor predictive power with events going forward.

[Brittney Griner's Clit]

Call it the Nerd Hole.

[WillieMcFML]

would obv like to hear more about this

bet game and half

[cmoney]

So I have been working with some nerds the last few months and worked out a bunch of computer simulations for different sports. We are getting into the 60 percent win rate with a large sample size. I think it would be cool to have an invite only section of the site where I posted the picks for the day and people can bet them if they want. We can also come up with theories and test them. I fully realize programs online exist that do this very thing, but I almost certain I figured out an edge with certain systems. This wont be Cmoney lock of the week where one year I luck boxed 20 in a row. As I said, we are getting, at best, 60-63 percent historically on some systems. The idea is to grind it out and not fire 10k on the first pick that posted. It could very well lose 5 or more in a row.

Anyway, thought I would throw it out there. If not enough people interested no point. I am mostly interested in people throwing out theories to test, plugging them in, and see what we get. More degen heads the better.

I have a background in statistical testing of predictive models, and have a question: Was the model/models to predict good bets tested using hold-out samples? This article discusses how to do so.

http://www.plottingsuccess.com/3-pre...cy-tests-0114/

Without such verification of a model, the predictive power could easily be overstated by overfitting the model to the data. And that would mean that the model will could easily have poor predictive power with events going forward.

Can you PM me an email address and I will send you some samples to take a look at?

[Shizzmoney]

I've assisted on making hockey/baseball algos in the past, and currently use one for football. I'd be interested in hearing more.

[Dan Druff]

I will be creating a "hidden" forum for this shortly. Cmoney will have the say as to who can access it and who can't.

[Corrigan]

So I have been working with some nerds the last few months and worked out a bunch of computer simulations for different sports. We are getting into the 60 percent win rate with a large sample size. I think it would be cool to have an invite only section of the site where I posted the picks for the day and people can bet them if they want. We can also come up with theories and test them. I fully realize programs online exist that do this very thing, but I almost certain I figured out an edge with certain systems. This wont be Cmoney lock of the week where one year I luck boxed 20 in a row. As I said, we are getting, at best, 60-63 percent historically on some systems. The idea is to grind it out and not fire 10k on the first pick that posted. It could very well lose 5 or more in a row.

Anyway, thought I would throw it out there. If not enough people interested no point. I am mostly interested in people throwing out theories to test, plugging them in, and see what we get. More degen heads the better.

I have a background in statistical testing of predictive models, and have a question: Was the model/models to predict good bets tested using hold-out samples? This article discusses how to do so.

http://www.plottingsuccess.com/3-pre...cy-tests-0114/

Without such verification of a model, the predictive power could easily be overstated by overfitting the model to the data. And that would mean that the model will could easily have poor predictive power with events going forward.

Mumbles actually said something useful.

I also have a background in this stuff (aka do this for a living) if you need any additional input C $

[GambleBotsChafedPenis]

im definitely interested in this stuff, trying to figure out how to do this kinda crap for DFS...

[thesparten]

I'd like to hang out with u guys..

[unchainedghost]

im definitely interested in this stuff, trying to figure out how to do this kinda crap for DFS...

Same, except for the dfs part

[MumblesBadly]

I have a background in statistical testing of predictive models, and have a question: Was the model/models to predict good bets tested using hold-out samples? This article discusses how to do so.

http://www.plottingsuccess.com/3-pre...cy-tests-0114/

Without such verification of a model, the predictive power could easily be overstated by overfitting the model to the data. And that would mean that the model will could easily have poor predictive power with events going forward.

Can you PM me an email address and I will send you some samples to take a look at?

I no longer have the software to analyze the data. And I see from a message above that Corrigan has experience in this area, and is currently doing such work. Perhaps he can directly help verify the robustness of the model.

[ThreeBet]

:whynot

[DaGreek23]

I'm in the with the Nerds. C$ is God.

[cmoney]

I have a background in statistical testing of predictive models, and have a question: Was the model/models to predict good bets tested using hold-out samples? This article discusses how to do so.

http://www.plottingsuccess.com/3-pre...cy-tests-0114/

Without such verification of a model, the predictive power could easily be overstated by overfitting the model to the data. And that would mean that the model will could easily have poor predictive power with events going forward.

Mumbles actually said something useful.

I also have a background in this stuff (aka do this for a living) if you need any additional input C $

Awesome can you PM me your email and I will send you some samples? I want full transparency, so if i send you something you think isnt good, please let me and the forum know and why. Without full transparency this whole experiment is pointless. I want to work with as many people as possible and figure out how to win and also have fresh ideas to adjust. I have some ridiculous math guys and programmers at my disposal so lets figure this out