/biz/ - Business & Finance

based camille poster

>>53072705
Yes, but only 50% accuracy

>>53072705
Doesn’t make sense to me. For example: How would an A.I. factor in the existence of other A.I. Price predictions? How can it know how accurate the other A.I.s are and how their prediction will influence the price.

>>53072709
Thanks

>>53072752
People say it wouldn't be possible for AI to create art, yet here we are

Were gonna use AIs to predict the girl we like.
It will read her mind and get her to say yes to dates, and anything else we desire ;)

>>53072705
no

>>53072705
Bros I just want a cute gf like this but all I end up finding are vapid whores at the bar

>>53072747
kek

>>53072705
>Will it eventually be possible to predict the lottery with AI?

AI is a meme and if you fall for the meme you are a retard

>>53072705
You don't need AI to predict the market. Don't fight the Fed, the rest is noise.

>>53072705
wetware already does this fairly well.

>>53072705
kek
you do know professional trading (as in: people who actually earn money trading) is already automated, right ?

>>53072705
Read up on Renaissance Technologies. It is not as attractive as you thing

>>53072705
>>53072766
haha stop posting my gf lol thats weird haha

>>53072705
Can't be done because if it would turn into the halting problem.

AI says the market is going down so everyone ajust their positions due to it. The machine then needs to take into account this to make a new prediction and people will act with that knowledge, modifying the AI prediction, in a loop.

Would be like making a machine that could tell you with 10 seconds of notice if you will rise your left or right hand. Since the machine tells you, you can always lift the other, and hence the machine needs to take into account that for whatever answer it gives you, you will do the opposite or something different, becoming an endless task to give a final answer.

>>53072709
>>53072766
you take credit for collecting and reposting pictures of women? like are you roleplaying as them or something?

>>53072705
How do I get a silly little goose gf?

>>53072709
what's based about being a simping faggot

>>53072933
Its not fair is it bro I settled for a 6/10 because shes kind of fun. Imagine a 9/10 like this who is also cute and funny and speaks with a sweet french accent

>>53072747
Lmfao

>>53072705
Was asking myself this aswell whe nthinking about AI Art. Then everyone will have an Auto daytrader

>>53074904
The market is not mysterious, it's just data.

>>53074494
idk

>>53072705

I am 28 years old and I never even kissed a girl

Neural Networks Prediction Models For Trading In Python

CodeTrading
15.4K subscribers

>>53072766
I tried to do it this year. Tried feeding price movements of about 600 stocks into all possible kinds of prediction models. Regressions, XGBoost, Catboost, all possible neural Networks. Always took a week to train them. Nothing worked. Sometimes it would seem like a model makes bank but in the end its just random luck. Also those models which had seemingly good results would yield maybe 3% in a month. So if you invested say 50k and the model is right 30 months in a row that would give you about 121k before taxes and trading fees. 50000*1.03^30=121k

>>53075091
3% a month is incredibly good though if you can get that relatively consistently

>>53075175
Thing is you can't and its only incredible if you trade shittons of other peoples money or if you're leveraged to the tits. If you're a poorfag you might just as well DCA into Indexfonds. Will take about the same 60 years to make any substantial cash.

>it's time for another episode of "can't tell if I'm on /biz/ or /r9k/"
time to take my meds

>>53075028
Same but I just turned 29

>>53072705
It's prediction you mentioned and not precision. The recent bags I loaded ORE, SAUCE and PHA were all based on research but we can only predict since we have no crystal balls

I absolutely seethe at this phenotype
My sister is very similar but hotter
Why do I hate it so much

>>53072705
This is the stupidest fucking thread, the fuck do you guys think quants do / HFT is under the hood

>>53075965
They have so much beauty and power but their lives revolve around being as nonchalant about it as possible. Their only desire is to make everyone seethe with jealousy at them. Its a banal way of life.

>>53072705
>>53072705
No. Simulations will always be a standard deviation away from normal which will offset everything. It might be able to predict an upward trend or downward trend but not the severity.

>>53072705
>>53072766
holy shit she is so beautiful bros. my knees are weak just from looking at her. imagine someone like that just smiling at you once. would make your day so happy. i could probably die happy if that occurred

No. But it’s already possible with my brain

>>53072705
cute

>>53072779
ok but turn it around
this random girl shows up at your house telling you you're her match
you don't even like her but she gives such good head you end up marrying her

>>53075965
It's how God let's you know there are far more cruel things than that you can imagine

>>53074095
Stupid analogy.
In the era of information, all legit data on a stock will be kept under proper security and disinformation will be fed to the opposition.
Then we'll have specific people to try and break the security of said neural network to gain access to the legit data or learn the source data for themselves, leaking is a net loss to everyone due to the halting problem, but even then the halting problem will exist only if we perfect such an A.I. which is never going to be the case.
Additionally, neural networks don't learn ONLY from the training period, they're almost always set to learn constantly, thus changing the source code and requiring an additional break-in and so on.

But A.I. will only be able properly predict the stocks if it will be able/trained to compile data from News Sites regarding that specific company, Insider Info, Patents and Costumer Feedback. Basically it NEEDS to somehow mathematically quantify abstract information and make a decision based on some constantly changing formula of importance and priority.

For example, you can analyze what happened to 600 other stocks when some negative information was posted on News sites, then arrive at a number of correlation, 200 of those stocks dropped X% when Y total number of News sites publicized posts containing Z% negative attributed words. Define the negative and positive words (negative: fired, closed studio, etc - positive: hired, opened, new, etc) and measure them inside a news media outlet regarding that stock.
This can be done with Insider Info, this can be done with Patents, this can be done at the level of Competition or level of Industry, etc.
If you can quantify abstract ideas into constantly changing numbers, you can arrive at a formula that would be able to predict what is in essence human behavior in regards to finance/resources.

It's a long-shot, but it can be done and I'm pretty sure it is already being done, I'm a moron, this isn't a novel idea.

>>53072705
>>53075091
>metalabelling
>use ML to predict the success of your proprietary trading algorithms and choose parameters
Combined with the Kelly criterion you basically have an automated system for finding alpha.

My futures trading system went from a Sharpe ratio of 0.9 to 2.6 with this method.

>>53072705
90% of Jews can beat the market, why can’t an AI? Oh that’s right, insider trading data can’t be trained on.

>>53075965
It's what I think a beautiful French girl would look like

>>53072705
No. Price action is just supply and demand. There are too many reasons to either buy or sell that go beyond financials and sentiment. It's too random

>>53072705
dont care about your question but she is pretty and very beautiful 10/10

>>53072766
I dont understand, Im anglo-celtic mix but have similar mouth, nose, eyes, and eyebrows as her. Why? Or is she larping as french

>>53072705
There is an ai driven etf, last I checked it wasn't doing great.