/sci/ - Science & Math

[StatsKid !!gYPVTJz4rL0 No.2928075]

sup ladies

I'm a majoring in stats and data modelling atm. Although I'm pretty good at math, what I really enjoy at uni is applied problem solving and forecasting (to the point where I can spend hours on the same assignment and still get a real kick out of it).

What I want to do is find a practical way to apply some modelling/problem solving that I can make money off (doesn't have to be much). Atm the two main areas I'm looking are sports or the stock market. So:

Creating a model to forecast the outcome of games and then using it to bet on certain games.

Or using the google finance api to manage a portfolio of stocks that I can buy and sell quickly depending on market events.

Problem is that I guess these have already been done to death and I'll be in a market filled with much bigger fish. So main questions I have are has anyone done this? Do you have any tips? Does it take much to break even or do you have to fucking number crunching wizard?

Also, if anyone has ideas for anything else that could do with some modelling/forecasting that could be either interesting or profitable, please say. I'm very interested in this kind of thing so would love to hear all sorts of ideas.

[Anonymous No.2928087]

meteorology. show up just after the freak earthquake with 200 gallons of water at $30 bucks a bottle

[StatsKid !!gYPVTJz4rL0 No.2928104]

>>2928087

[StatsKid !!gYPVTJz4rL0 No.2928176]

Another idea was betting strategies for various casino games (always been a roulette fan), but seems to be lots of problems with doing this. Half don't work, those that do get you kicked out, they've all been worked out and done a thousand times over and it isn't nearly as exciting since your just applying the same rules over and over.

Preferably I'd like something with a bit of forecasting and time series analysis involved. Problem is I'm still a fledgling programmer so struggle with developing web apps to suck data out of a bunch of different rss feeds.

[StatsKidd !!gYPVTJz4rL0 No.2928363]

nothing /sci/? Guess I'll try again later

[Anonymous No.2928384]

Mathematician here. My adviser is a stats guy and he has done the sports thing to death. There are too many variables in most games to predict the winners with any sort of precision, and if you could no one would take any bets off you after the first couple times (or you would get murdered). Casinos are a no go, they have analyzed their games to death (along with many many other statisticians). The stock market is similar, too many variables to model accurately, the best you can hope for is a stochastic model that is sort of kind of accurate some of the time.

"You cannot beat a roulette table unless you steal money from it." -Einstein

[Anonymous No.2928407]

>>2928176
There is no way to beat a Roulette table dude. The expectation value of each $1 bet is - $0.05. Nothing changes that. No Martingale, etcetera. So unless you bring a pc into the casino and analyze the physics of each spin you are hooped.

tldr; indicated poster fails at statistics. Get out of \sci\

[StatsKidd !!gYPVTJz4rL0 No.2928448]

>>2928384
Sucks, it's the impression I'm getting as well. All I really want is something fun to model/forecast that doesn't require access to some super expensive database.

If anyone has ideas for interesting things to forecast I'd like to hear about them. I'm getting to the point where I just want to do something, regardless of what.

[StatsKidd !!gYPVTJz4rL0 No.2928474]

>>2928407
No shit. I wasn't saying it was viable, just that I enjoy the game. Have a funny martingale story to post as well.

Friend of mine spent a week or two obsessing over martingale and how it was a good way to win while betting on the same colour. Tried to tell him it would be a bust but he wasn't listening. Spent ages talking it up and we finally all went to a casino (he wanted to test it out, rest of us were just going to bet whatever).

He goes off to another roulette table because he didn't like the set we were at. Came back about a minute later, we weren't sure he'd even had time to start betting. Turns out he had decided to bet repeatedly on red, doubling his bet each time he lost. Starting from the very first spin, casino gave him 11 or so blacks in a row. Never seen someone get shut down so hard. He only had $50 so he got knocked out at 3rd/4th spin, but would've needed thousands to float himself through to the 12th spin.

[Anonymous No.2928487]

>>2928448
try to analyze forex (stock on steroids). go for very short term (scalping) technical analysis using price action and support/resistance levels. It might be more artificial intelligence than data crunching but hey it's just my 2 cents.

[Anonymous No.2928492]

>>2928407
>>There is no way to beat a Roulette table dude.

does anybody have the explanation for why that "bet $100 on red, bet $200 if you lose, $400 if you lose again... eventually guaranteed to make $100 profit" approach fails? intuitively I feel it should, but I have yet to understand why...

[StatsKidd !!gYPVTJz4rL0 No.2928500]

>>2928492
Because of odds of it being red or black aren't 50/50. Since there is also the zero, the odds of it being red or black is in fact 48.7% each way (i forget exact number, but it is something like that).

[Anonymous No.2928506]

>>2928474
same from >>2928487 i heard you can also use martingale as a money management system. by that i mean going to play normally but having different series of black/red betting in mind that you altern, and doubling the bet each time you loose on a serie. when your capital don't allow you to bet anymore just start a new serie with the minimal amount permitted. don't know if it's ev+ tho, again just my 2 cents heard that a while ago

[StatsKid !!gYPVTJz4rL0 No.2928521]

>>2928487
Appreciated, I'll look into it.

>>2928492
Also, it is pretty dangerous and can spiral out of control. If your unlucky enough to get to the 5th bet without a win, your looking at having to float yourself for a $3200 bet. You can see how this gets worse very quickly, the 10th bet is $100k. For a $100 profit? No thanks.

[Anonymous No.2928532]

>>2928492

All of the spins are independent, over the long-haul you'll make the same amount of money (read: go broke) regardless of how you bet. No matter how much money you have, eventually you will hit a streak of losses that will bankrupt you, and as someone already said, it doesn't take long until you are betting hundreds of thousands of dollars for a $100 profit.

[Anonymous No.2928538]

>>2928075
you could just forecast useful stuff for a company and they'd pay you for it. Most viable chance at getting money off of forecasting anyway.

Also, best way to make money off of people using stats is scams (like casinos).

[Anonymous No.2928541]

>>2928521
ah, good point... so effectively it is still based on luck... you're just betting against the chance of losing several times in a row

which is a lot less likely than losing just once or losing and winning intermittently, but then again the payoff isn't great and the cost if you do happen to lose is high

[StatsKid !!gYPVTJz4rL0 No.2928559]

>>2928538
Oh yeah definitely, I intend to do this (or something similar, management consulting perhaps) when I finish uni. Problem is, at the moment I don't have time to work full time. I've been trying (still am) to find something remotely relevant that I can do part time but it's proving pretty tough since there isn't much on offer part time and I have no relevant work experience (i work in a cafe).

So I thought if I could create an effective model or forecast of something on my own, it'd be fun, keep me interested/entertained with the field and also look good on my resume (show self motivation etc) at the end of the year when I finish.

[StatsKid !!gYPVTJz4rL0 No.2928590]

>>2928541
Pretty much. Thing is, even if it wasn't going to screw you over, if you could afford to float yourself for long loss streaks then the tiny profit you are making from it is probably a waste of your time.