[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance

View post   
File: 83 KB, 500x750, butt_dump.jpg [View same] [iqdb] [saucenao] [google]
875168 No.875168 [Reply] [Original]

I have a developed a mathematical model for NFL betting that makes long term profits (about 56-58% win rate). I have painstakingly developed it over the course of the past few years, using play-by-play data and state of the art machine learning algorithms.

The problem is I have no capital. How do I make bank off this system? It seems really hard to find big money that would be willing to 'invest' with me.

My best idea so far is to go directly to the bookmakers and try to either sell it to them or get a job with them.

>> No.875169

How to succeed in life.

Write down on paper in order of their importance the things and conditions you really want.

Do not be afraid of wanting too much. Go the limit in writing down your wants.

Remember to right in the PRESENT TENSE (example - I have a girlfriend, I own a Honda Civic 2016)

Change the list daily, adding to or taking from it, until you have it about right. Do not be discouraged on account of changes, as this is natural. There will always be changes and additions with accomplishments and increasing desires.

Three Positive Rules of Accomplishment
1. Read the list of what you want three times each day: morning, noon and night.
2. Think of what you want as often as possible.
3. Do not talk to anyone about your plan except to the Great Power within you which will unfold to your Objective Mind the method of
accomplishment.

It is obvious that you can not acquire faith at the start. Some of your desires, from all practical reasoning, may seem positively unattainable; but, nevertheless, write them down on your list in their proper place of importance to you.

There is no need to analyze how this Power within you is going to accomplish your desires. Such a procedure is as unnecessary as trying to
figure out why a grain of corn placed in fertile soil shoots up a green stalk, blossoms and produces an ear of corn containing hundreds of
grains, each capable of doing what the one grain did.

If you will follow this definite plan and carry out the three simple rules, the method of accomplishment will unfold quite as mysteriously as the ear of corn appears on the stalk, and in most cases much sooner than you expect.

>> No.875172

>>875168

Ass is nice

>> No.875177
File: 221 KB, 267x200, 3772501+_0f6cad9cae91cc06a8d2eb643fa0987b.gif [View same] [iqdb] [saucenao] [google]
875177

10/10

>> No.875181

>>875169
>remember to right

>> No.875183
File: 806 KB, 320x180, mmm.gif [View same] [iqdb] [saucenao] [google]
875183

>>875177

Tattoos are a no for me though

>> No.875213

No proof didn't happen. Post picks. Everyone has a 60% win rate with their algorithm. Some even have 90%. But something everyone has is no evidence.

>> No.875215

>>875213

Well, I plan on posting picks over the course of this season in real time.

>> No.875216

>>875168
HNNNNNNNNG

If you have a system that makes long term profits why don't you simply start using it to make money?

>> No.875224

>>875216

Not enough capital to invest. With my win rate, I only see maybe an 800% return on the bet size for a season.

Meaning, if I bet $100 a game over the course of the season, I will only get $800 profit. Now, if I can bet $10,000 a game, that is a profit of $80,000. The problem is bankroll. Safely, one doesn't want to bet more than 1-2% per game of their total bankroll. So, to bet $10,000 a game, I would need somewhere between a $500,000 and $1,000,000 bankroll.

>> No.875234

>>875224

Also, note that my $100 a game bet would require between a $5,000 - $10,000 investment based on the 1-2% bankroll rule.

>> No.875239

>>875224
>>875234
If you need to bet only 1-2% of your bankroll that means there's a chance it will fail. And that means nobody will want to invest. If you indeed have found a system that works, use a bigger percentage of your bankroll and use your profits to increase your bankroll.

>> No.875241
File: 136 KB, 713x1080, images.jpg [View same] [iqdb] [saucenao] [google]
875241

>>875168
Read this for some color on the way secret betting groups run their business.

>> No.875242

>>875239

This is what every investor does? Do you dump all of your cash into one stock? You diversify.

Professional handicappers use the bankroll rule to ensure they do not lose all their money. The reason for this is due to losing streaks. If you know anything about statistics, there is always a positive probability of losing, say 10 (or any number) times in a row, even if your model is a winning model. So, it's to mitigate risk. If you bet half your money every time, the probability of losing two in a row is quite high. But if you only bet 1%, the probability of losing 100 times in a row is almost zero.

>> No.875244

>>875241

Yes, I know this guy. And 'The Computer Group' in the 80s. The problem is, where do you find the people to back you?

>> No.875278

>thinking you can use an algo for sports betting

Never mind if you could, why would you try? There are so many variables that are laid out for you in the media regarding squad selections, players attitudes etc. that I'm willing to bet (lol) you would make a better profit paying attention to that sort of thing, rather than trying to codify it (unsuccessfully) in some algo.

The edge in sports betting comes from being able to take these things into consideration.

>> No.875288

>>875278

Okay, but you're not answering my question. Let's assume that I can't win doing 'algo'. But now let's assume I can win using this subjective type of betting.

How do I make bank doing that?

>> No.875289

>>875278

Also, my 'algo' takes into account many of these factors - injuries, weather, etc.

>> No.875294

>>875168
>56-58%
So its like you pick one of two teams and then its 50:50? Lol no, your algorithm is shite

>> No.875326

>>875294
that is a pretty retarded oversimplification. just because there are 2 options doesn't mean probability is 50/50...

>> No.875333

>>875294
To break even your average winrate has to be around 52% to overcome what's known as the vig or overround. Anything above that has an edge which means long term profit. Very difficult if not impossible to have a winrate significantly higher than 60% in the long run.

>> No.875340

>>875234
>>875239
I think he means he needs 1-2% per game in the entire league?

>> No.875341

>>875333

This guy gets it.

>> No.875344

>>875340

What I mean is that, if you are going to place a $100 bet, you better have 50-100 times that in your account.

It doesn't really matter when you place those bets. Could all be in one night. Could be over the course of the season. The bankroll insures you don't lose all your money in case of a losing streak. Even if you do lose 10 bets in a row, with this bankroll you can continue betting the same amount as before (assuming your system wins in the long run). If you didn't have the bankroll, you would have to lower your bet amount in the event of an unlucky streak.

>> No.875345

>>875168
The problem is that you're betting the NFL. It's probably the hardest sport to beat, and even if you can, there are only 256 games during the regular season. Could your model be applied to college football? That would add a lot of potential bets.

>> No.875348

>>875345

I am going to start working on systems for all four major sports. NBA is next. The problem with college and other less professional leagues is 1) lack of data and 2) the trends within the league can change drastically in a short period of time (i.e. the model assumptions may be incorrect).

>> No.875350

>>875348
what does your model use for data

>> No.875352

>>875348
Well, college sports are harder on an absolute level to handicap, because of SoS differences and stuff like that, but that doesn't really matter. What matters is how good your model is relative to Vegas--even if your model is worse at predicting college than NFL, if Vegas is even worse, then your model is relatively better, and thus more profitable.

>> No.875360 [DELETED] 

OP here again.

I really don't care if people don't believe my model or not. But it's pretty depressing to know you can make money, but just don't have the capital for it.

It's comparable to having your house sit above a massive pile of gold, but you don't have the means to dig it up.

>> No.875365

>>875360

That's a misguided analogy. You're imaging your model to 100% produce something of tangible value, like finding physical gold. A more accurate statement of the analogy would be having your house sit above a massive pile of something; the material might be worth something or it might make you lose your house after you started digging.

>> No.875367

>>875169
>Think & Grow Rich

Might look silly, but once u applied it u get to know how it works on yourself

>> No.875371

> Oh look this thread again

>> No.875376

>>875168
>(about 56-58% win rate). I have painstakingly developed it over the course of the past few years, using play-by-play data and state of the art machine learning algorithms.

Your system is barely better than coin flipping. Actually, if you take into account the fact that your system probably got lucky streaks, it's equal to coin flipping.

No one is going to buy into this

>> No.875382

>>875367

You sayin' this works?

>> No.875387

>>875376

Well, you need at least 52.4% win rate to be profitable. You can show (statistically, using a simple test of proportions), that the probability of a 56% win rate over the course of >100 matches, due to random luck is extremely small.

Explained in more detail here:

<<http://isites.harvard.edu/fs/docs/icb.topic116457.files/Review_Session_Slides_-_Part_2_-_Yves.pdf

>> No.875429

>>875382
Yeah, you gonna believe what your reading and take it seriously, cause it really is. It helps on following and reaching goals, setting objectives etc... In some way it makes u stay focus constantly, reducing distractions as well as at the same time you invest that time on this instead of other less "productive " things.

Also it improves your determination and confidence.

I know it could sound ridiculous etc, I even feel uncomfortable talking about it here (I'm not a moronic self esteem magician prophet) ... But yes, it works while applying it day after day, you can give it a try

>> No.875434

>>875429
>You're *

>> No.875445

>>875376
If he genuinely has a 56-58% win rate against NFL lines, that is huge.

If I knew and trusted someone with these results, I would certainly invest.

>> No.875452

>>875445

I think my best plan of action at this point is to reach out to the bookies. They are the ones with capital and would benefit the most from having this knowledge.

>> No.875468

>>875429
>sounds ridiculous
dont even, your original post is one of the most intelligent wisdom filled posts ive read in years. You dont belong on 4chan anon get out of here. Go make a change in the world.

>> No.875477

>>875168

POST YOUR WEEK ONE PICKS.

>> No.875486

>>875452
Keep in mind that few bookies actually set lines themselves, most just copy another sportsbook's lines. betCRIS is the only online sportsbook that sets NFL opening lines I think. There are Vegas sportsbooks that do too, but I don't remember which ones.

>> No.875504

OP,
From what it appears, you have two opportunities, you can game bookies with your system and get rich or you can run your own books and once it gets big enough you can absorbed into a larger suit.

I work a reasonable job and have money I want to diversify.
If you can prove you have something viable here, I'd be very willing to put thousands of dollars into your operation. You can take the management cut and let the money ride and if things are working out well, I have many coworkers who would be willing to let you handle their money too.

If things get that far, I would be very open to putting more into your model and helping you find people willing to bet with you. You get a management cut of it all, you give me a finders collection and then you run your own book or betting mutual fund.

If running a crowd sourced book sounds interesting. Lets talk more.

>> No.875508

>>875504

This sounds good to me. Shoot me an email here:

supergeekman@mail.com [I'd give my normal email, but it is 4chan after all.]

>> No.875512

>>875504

Also, back to what I was saying before, option #1 won't get me rich because I don't have the capital to invest. I would need between $500,000 - $1 Million to see a decent return.

It is similar to stocks - if you have money, you can make money. If you don't, well, then you're shit out of luck.

>> No.875516

>>875512

Option 1 can get you rich. http://www.cnbc.com/2015/06/23/nevada-okays-mutual-fund-for-sports-gambling.html

Having a track record and not quitting your day job is all you need to get big finance interested.

>> No.875523
File: 13 KB, 430x365, bionominal.png [View same] [iqdb] [saucenao] [google]
875523

>>875387
Getting a 56% winrate or better on 100 matches from just flipping a coin is 13,56%.

>> No.875799

>>875523

bump

>> No.875816

>>875523
wow I do not miss stats

>> No.875850

>>875523

Is there no possibility of draws in NFL?

>> No.875858

>>875242
In commodity trading that is called linear regression analysis.

>> No.876082

Congratulations, you've successfully back tested a system for predicting outcomes in a league, that if anything, is good at reacting to prior trends.

Prepare to get bent over by your bookie.

>> No.876206

>>875168
the problem with this is the algos would pick you up, unless you're on a bottom end site.

watch the movie runner runner to see what happens to people who try to push their ways into an established dirty business.

>> No.876226

>>875486

It's far more complicated than that now. This isn't the 60s or even the 90s when thing were done by phone and word of mouth - betting is a global business. I don't know where you are but in the UK and most of Europe, betting is legal. Bookmaking companies like Ladbrokes, William Hill and Bet365 earn millions per year - Ladbrokes is listed on the stock market. The salient point is they employ a vast network of contacts and experts - people paid to sit and watch games and do statistical analysis - and this information is what factors into their oddsmaking. The idea that OP is gaming this enormous system, all the money they throw at it and the well-paid mathematical brains behind it AND beating the vig at the same time, is ludicrous. Profiting at gambling involves picking your battles, being able to spot where you're getting good value for a bet and taking your chances. It's not "setting up an algo" and letting it ride.

>> No.876285

>>876226

OP here again. Well, this is the idea of my model. It doesn't bet on every game obviously. I guess I should have clarified.

Most of the time my model agrees with the Vegas line. It's when my model disagrees with the Vegas line that is interested. Essentially, I am finding market inefficiencies. In general, the Vegas line will be dictated by the public (i.e. if the public favors a team heavily, the line will move). The idea is that the public is usually good at picking winners, but sometimes they are not. An example is winning streaks - the public puts too much emphasis on win/lose streaks, when in reality these are very poor indicators of future outcomes.

>> No.876288

>>875168
nice

>> No.876302

>>876285

The public are fucking awful at picking winners - this is why bookmaking exists as a business at all.

>> No.876303

Signs someone might be a stupid sack of shit who you shouldn't trust with absolute basics, nevermind your personal money:
>typing "u" instead of "you"
>getting right/write wrong
>getting too/to/two wrong
>getting advice/advise wrong
>suggesting you should buy/sell according to emotions rather than info

>> No.876310

>>876302

Well, then you are only confirming my point. Typically, bookmakers will set the spread to get 50/50 money on each side of the line. In this way, they make the ~10% vig no matter what.

So, really, it doesn't matter to the bookie if the line is 'true', they only care about getting even money. So if the public is wrong, my model can take advantage of this inefficiency in the market.

>> No.876363

>>875169
We know how corn grows.

>> No.876382

Just use the Inverse-Gambler's-Ruin formula to work out how much capital you would need to have a 99% chance of not going bankrupt and get a loan for that value saying it's for home improvements.

>> No.876401

>>875850
there is but it is very rare.
eagles fan here... inb4 mcnabb debacle...

>> No.877189

>>875168
Having a 57% winrate on algorith of past evens is not as difficult as it seems. I can analyze data of 100 football matches and say as darkest is the t-shirt of a team more chances have to win, I run the model on past events and give me a high ROI, but the next match my expected ROI will be 0 before rake anyways.

>> No.877353

>about 56-58% win rate
OP, YOU HAVE REVOLUTIONIZED THE FIELD OF MACHINE LEARNING.

THIS BY FAR BEATS MY PREVIOUS MODELING ATTEMPTS, FEATURED IN CHAPTER ONE OF MY NEW BOOK, "INTRODUCTION TO STATISTICS".

I CALL THIS MODEL THE "FAIR COIN"

>> No.877596

>>877189

Yes, that is called over-fitting. I am using cross validation on my entire test dataset, so stuff like this won't happen.

>> No.877602

>>877353

This is against the spread, most professional handicappers barely achieve this rate.

>> No.877628

>>875168
Sounds like some one is trying to jew some greedy gambling addict out of what little monies they have left

Home boy is clearly trying to extort one of you pathetic gamblers.

Don't give home money; get some help.

http://www.gamblersanonymous.org/ga/content/recovery-program

>> No.878123

>>875168
>a mathematical model for NFL betting that makes long term profits (about 56-58% win rate)
So, it's fractionally more profitable than flipping a coin.

>> No.878166

>>875168
Bookie here. I doubt you. My best wise guys win 55% of the time. Unless you've have 20+ tickets of you winning that good why would anyone believe that.

>> No.878168

>>878123
That's really good for sports betting.

>> No.878284

>>877596
HAHAHAHAHAHAHAHAHAHAHAHAHAHA
HAHAHAHAHAHAHAHAHAHAHAHAHAHA

I wrote 20 pages in my dissertation on why that assumption has NO BASIS in reality. None.

Please do yourself a favor and show your results to a bookie and try to sell the model-making software/algorithm for (whatever you can get) and move on.

>> No.878318

>>878284
Not OP, but why is he wrong? Do you have a link to any articles/papers that could explain it?

>> No.878381

>>878318
Here's a paper on one of the failures (cross-validation won't necessarily find a good model): http://www.jstor.org/stable/2290328?seq=1#page_scan_tab_contents

Here is a paper where someone fell into this fallacy:
http://link.springer.com/chapter/10.1007%2F978-3-642-13470-8_14

But the problem is the underlying assumptions:
1 - get data
2 - make a bunch of models, using cross-validation
3 - best performing model is now the model with "best generalization"
-----
4 - get NEW data from NEW time and NEW distribution. Assume previous model will be good!

If you are working on a problem like "Netflix movie rating prediction", this is totally cool for two reasons:
1 - you are using your 'best' model predictions on the existing set (you trained on a dataset with Tom Jones and you are predicting his stuff now, pretty legit.).
2 - any new customer that shows up is probably remarkably similar to some other person in your dataset (Joe Tomas has similar ratings to Tom Jones)

However, for things with changing underlying models, and ESPECIALLY in winner-take-all systems (betting pools, stock market, etc.), there isn't a good reason to think that people will invest in 2017 like they did in (all of recorded history prior to then). There is especially no reason to think that the automated trading (/betting) algorithms (against which you are competing) will get 0 update and perform just like they did 12 months ago.

>> No.878395

>>878318
>>875168
This is why I advise that you just sell the software to someone with the ability to make market.

"Help! I am an entrepreneur who invented something but do not have an ability to take it into the marketplace"
"This sounds like you are an _inventor_, rather than a business owner. You should sell your invention for a price which the market will sustain to a customer who will take it to market."

If there is evidence that the invention/product will make someone $1,000,000 over the course of 6 months (a year is a long-distance prediction in this climate), you might settle for $250K and start working on the next invention.

More likely, some sucker will buy it for $20K, not realize your flawed assumptions, and lose all their money. But hey! $20K for you.

>> No.878442

>>878395

I guess then my question is why would anyone be willing to buy my model. I'm not sure the bookmakers would be interested.

Even if I can prove that my model can predict NFL outcomes, how does this help the bookmakers?

>> No.878445

>>878381

What is your background in?

Also, the first paper you show doesn't contradict my analysis. That paper shows that there is a problem with leave ONE out cross-validation, not k-fold cross validation. They are also using automated variable selection at each step, which is in general bad practice.

>> No.878449

>>878442
bookmakers probably aren't interested in it. A professional gambler or syndicate might be, but I have no idea how you'd contact someone like that.

>> No.878452

>>878449
There are forums for sports betters, like sportsbookreview and covers. But they all seem like degenerates, not real pro bettors.

>> No.878462

>>878452
>>878449

My other idea is to start a website, post my picks for a while and then turn it into a subscription based service. I have seen a bunch of those, but who actually subscribes to them???

>> No.878464

>>878462
Only degenerates subscribe to them, since almost none of them actually make money. They're scams, and post fake records.