collective intelligence in sports gambling

[big0mar]

Oh. Didn't see you add this either.

But, of course, it did. That goes without saying.

Obviously.

Oddly not anymore? PROVIDED. Keyword.

Yes. Surely you couldn't expect a particular model to last forever. Unfortunately a pretty widely known gambler outted the particular data I'm referring to. Either way there are always others to find. But those with more maneuverability will always be the first to find out.

[strixee]

It is done once and it's over. Not sure what you fail to comprehend. I can code it faster than finding someone else to do it, explain to them exactly how I want it, and test it later.

Yes, sometimes that's true. The biggest advantage of coding it yourself is you can spot errors in the data much better etc.

[allin1]

Just like any BLOATED operation you mean.

Would you consider Buffet, Gates, Soros and others similar to them running bloated operations? Maybe when one reaches the top of the food chain starts to think different, act different.

[MonkeyF0cker]

Yes. Surely you couldn't expect a particular model to last forever. Unfortunately a pretty widely known gambler outted the particular data I'm referring to. Either way there are always others to find. But those with more maneuverability will always be the first to find out.

Not following. So, now you don't incorporate shot selection data into your model? Are you saying that your model was solely comprised of shot charting data?

[MonkeyF0cker]

Would you consider Buffet, Gates, Soros and others similar to them running bloated operations? Maybe when one reaches the top of the food chain starts to think different, act different.

Let's compare apples to apples, shall we? How many trading or investment firms have you worked for?

[MonkeyF0cker]

Probably none. So, let me just state this: There's a very good reason that every successful quantitative trading firm works in SMALL teams of analysts and programmers.

[allin1]

Let's compare apples to apples, shall we? How many trading or investment firms have you worked for?

Just 1.

I was not comparing, it was like a metaphore. I think that as your capital grows beyond the 1Mil mark the game is different and most likely your life is different and your priorities and approach change

[allin1]

Probably none. So, let me just state this: There's a very good reason that every successful quantitative trading firm works in SMALL teams of analysts and programmers.

I was a trader for five years. I knew programming but there were others that were far better at programming. They knew nothing about trading but they made my work easier because they did it faster and better while I worried about other things. Working in a team where each is good at what he does could make things easier and better SOMETIMES.

[MonkeyF0cker]

Just 1.

I was not comparing, it was like a metaphore. I think that as your capital grows beyond the 1Mil mark the game is different and most likely your life is different and your priorities and approach change

Not necessarily. There are more people than you might think with seven figure bankrolls that work almost exclusively alone and/or with the employ of a few runners. Perhaps, some lose interest in it at that point and wean themselves of responsibilities and workload for a reduced ROI. However, that is another point in preference. The discussion began as an assertion by Justin that "You (alone), a single person, or even a small group cannot include everything." and "The larger groups can include a lot more information, and are the ones that define the closing price." There has been no evidence offered that even adumbrates any fragment of those assertions to be true.

We were also talking about Justin. Do you honestly think he's dealing with a $1 million haircut? If so, why would he still be working here?

[big0mar]

Not following. So, now you don't incorporate shot selection data into your model? Are you saying that your model was solely comprised of shot charting data?

No I'm saying that many others now use it.

[allin1]

Not necessarily. There are more people than you might think with seven figure bankrolls that work almost exclusively alone and/or with the employ of a few runners. Perhaps, some lose interest in it at that point and wean themselves of responsibilities and workload for a reduced ROI. However, that is another point in preference. The discussion began as an assertion by Justin that "You (alone), a single person, or even a small group cannot include everything." and "The larger groups can include a lot more information, and are the ones that define the closing price." There has been no evidence offered that even adumbrates any fragment of those assertions to be true.

We were also talking about Justin. Do you honestly think he's dealing with a $1 million haircut? If so, why would he still be working here?

Yes I think J7 is at the top of the food chain. I think that beacause of what I have read in his book and on the forum. I don't think he "works" here just like I don't think he wrote his book for money. I believe he is in it for other reasons... But I might be wrong and too naive. Maybe you know more about him, I am kind of new to the forum

[strixee]

Yes I think J7 is at the top of the food chain. I think that beacause of what I have read in his book and on the forum. I don't think he "works" here just like I don't think he wrote his book for money. I believe he is in it for other reasons... But I might be wrong and too naive. Maybe you know more about him, I am kind of new to the forum

Yes, I also think money wasn't really the reason why he wrote the book. It's quite a good book BTW. The only thing I'd like to briefly explain by J7 is how he uses ZunZun for complex models with many inputs

[mathdotcom]

what a battle . com

[MonkeyF0cker]

Yes I think J7 is at the top of the food chain. I think that beacause of what I have read in his book and on the forum. I don't think he "works" here just like I don't think he wrote his book for money. I believe he is in it for other reasons... But I might be wrong and too naive. Maybe you know more about him, I am kind of new to the forum

If you know anything about statistics, it shouldn't be too hard to pick up on his (lack of) statistical aptitude.

He certainly has quite a few people fooled though.

[probettor1]

Next tell the guy to used 100 jars. In science you need to use more than one sample.

[probettor1]

any way if you want collective intelligence go to covers and check consensus picks. There you will know where is the public bet.Dont be too exited. I've been there. Collective stupidity aplies too.

[probettor1]

This is an example of collective intelligence: 15 millions americas do bet in sports, 285 millions americans don't bet in sports because they know they are going to lose. This is the best thing you are going to be able to use in collective intelligence about gambling. Just dont do it.

[probettor1]

Having most americans buying old houses 6 times the original price making 10 dollars per hour at work is an example of collective intelligence when it comes to investment. Maddoff could probably talk to you about it. Or Warren Buffet. How do you think buffet makes his billions? Answer: using the collective intelligence. Buying cheap crap and selling it to the "collective intelligents" for twice the price. once they are broke, they work, they save and they buy again. Not one guy, it has to be millions so that Warren Buffets has been able to make ove 50 billions over time, just buying and selling the same crap over and over.

[allin1]

If you know anything about statistics, it shouldn't be too hard to pick up on his (lack of) statistical aptitude.

He certainly has quite a few people fooled though.

shame on me if I have been fooled.. I haven't picked up on such a "lack", but I am a complete novice...

[Inspirited]

If I didn't mess up this should show that collective intelligence is not so simple. I used squared error. There is diversity in the predictions. No two people make the same predictions. One group consists of squares. Another group consists of sharps. The third group consists of the two squarest squares and the sharpest sharp. The last group consists of two sharps and a super sharp.

All of the squares benefited by using their average predictions rather then their individual predictions. All of the sharps benefited by using their average predictions.
The mixed case is kind of unexpected. The sharpest sharp benefited by going with the average predictions with the two squarest squares rather than using his own prediction. Finally, the super sharp with the two other sharps did better on his own.



Actual Results
5, 7, 10

Predictions

Squares
a: 11, 2, 5
b: 3, 17, 15
c: 7, 12, 16

squared error
a: 86
b: 129
c: 65

average predictions
7, 10.33, 12

average squared error
19.11

Sharps
a: 4, 9, 12
b: 2, 8, 11
c: 5, 9, 8

squared error
a: 9
b: 11
c: 8

average predictions
3.67, 8.67, 10.33

average squared error
4.67

two scrubs, one sharp
a: 11, 2, 5
b: 3, 17, 15
c: 5, 9, 8

average predictions
6.33, 9.33, 9.33

average squared error
7.67

two sharps, one super sharp
a: 4, 9, 12
b: 2, 8, 11
c: 6, 8, 9

squared error
a: 9
b: 11
c: 3

average predictions
4, 8.33, 10.67

average squared error
3.23

[Inspirited]

I'm guessing this is 'kind of like' the idea behind ensemble learning.

http://en.wikipedia.org/wiki/Ensemble_learning

[MonkeyF0cker]

That's an interesting approach, Inspirited. How do you arrive at those estimations though? They seem somewhat arbitrary. Although, so are the terms sharp and square I suppose. I wonder if randomizing the estimations within a targeted range of error for each subset and simulating over a large sample would yield similar results.

[Inkwell77]

good thread!

[GunShard]

This won't work.

To be good at sports gambling, you have to see what works and doesn't work. Which factor in the particular sport effects and what factor has no effect.

To fade the public, you would have to bet against what the public is confident on.

[Inspirited]

That's an interesting approach, Inspirited. How do you arrive at those estimations though? They seem somewhat arbitrary. Although, so are the terms sharp and square I suppose. I wonder if randomizing the estimations within a targeted range of error for each subset and simulating over a large sample would yield similar results.

Yeah, these are arbitrary. I was just trying to show the different possibilities. A simulation is a good idea.

[HUY]

But the only way you can know what the public thinks is through the market odds, which has already used this information and is thus ahead of you.

[Dark Horse]

Collective intelligence can cover more bases. But collective intelligence rarely goes deeper than individual intelligence. In some cases it may. See Manhattan project.

[Sportsguy_USA]

Harvard Business Review
"Decisions 2.0: the Power of Collective Intelligence"

http://hbr.org/product/decisions-2-0-the-power-of-collective-intelligence/an/SMR302-PDF-ENG

MIT
Centre for Collective Intelligence

http://cci.mit.edu/

Like any application of methodology, you can go weak and get unquantifiable results, or go hard and get those deep individual and collective quantifiable results.

Its used in the markets today making the rich get richer, and highly popularized with large brands in social media marketing for consumer insights to name two.

In the end - its how the data is harnessed and tested. It won't always be 100%, but 'over time', a safer bet over one individual.

Damn - didn't anyone just watch Star Trek - 'the borg' - damn!

[357vegas]

Ask Billy Walters About Collective intelligence? Has worked for him for over 30 years.