[ 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: 156 KB, 1746x868, 10293810237418.jpg [View same] [iqdb] [saucenao] [google]
10968976 No.10968976 [Reply] [Original]

>> No.10969008

Statistical models say there is a 80-95% chance of this happening. Volume profile confirms. Can't give you my code b/c I'll lose my edge. Just a minor deviation from the last model. Resistance faced in the past, and now will become support on the way up. Just like how 5-6k become support where as before it was resistance.

>> No.10969041

>>10969008
what model?

>> No.10969103

>>10969041
Seconding

>> No.10969106

New ATH before EOY? Lol

>> No.10969113

>>10969041
Carhart four factor modeling on Bitcoin. Modified heavily from some research papers back in 2010. Works really well over long periods, there is some calibration involved that takes a long time to perfect (machine learning helps). The one I use with my team is was accurate to predict 12k Bitcoin last year. It actually went to 19k. Making it much more bullish than expected.

>> No.10969130

>>10969113
did it accurately or somewhate predict the fall from 19k to where we are today as well? if so good for you, bullish for the rest of us

>> No.10969180

>>10969130
It predicted 12k as the peak, and that there would be a decline in price movement resulting in the fall. The peak was off (mostly due to human factors like FOMO), but the bottom was accurate in the 5-6k range.

>> No.10969193

>>10968976

Still haven't solved the fee problem.

>> No.10969200

>>10969180
not bad fren, not bad at all....

>> No.10969203
File: 119 KB, 377x659, PSX_20180825_230054.jpg [View same] [iqdb] [saucenao] [google]
10969203

>>10969193
Behold your shit covered salvation.

>> No.10969242

Less than 4 months remain, how exactly are you planning to push btc over 100k in a bear market, low volume market, improper regulations and bad sentiments? Meme magic and coordinated mass manipulation? Rhetoric question btw, saging this low quality "analysis".

>> No.10969254

>>10969242
We don't have to do anything. It'll do it on it's own. There is no reason to look for reasons when there are none to look for. It is just a model that has been accurate this far.

>> No.10969280
File: 44 KB, 720x663, 1536165389287.jpg [View same] [iqdb] [saucenao] [google]
10969280

>>10969254
What about altcoins?

>> No.10969296

>>10969280
>>10969254
TELL ME ABOUT REQ

>> No.10969321
File: 26 KB, 498x499, 1532749980085.png [View same] [iqdb] [saucenao] [google]
10969321

>>10969254

i dont even care if this is legit or not thank you im so fucking depressed its not even funny and its not even like ive lost a lot of money i just didnt sell the top and have been bag holding for half a year

>> No.10969325

>>10969254
RT is shilling crypto. It can happen through whale wars.

>> No.10969328

>>10969321
op how can I contact you I will literally pay you if you get this right

>> No.10969349

>>10969280
We have only run it on Bitcoin, and got successful results. This is due to the large data set of Bitcoin. Alt coins do not have large enough data sets. However, since Bitcoin is the leading indicator of direction in crypto. I think it is safe to say alts will likely benefit from the capital flow from a Bitcoin peak.

>> No.10969353

>>10969008
>Statistical models say there is a 80-95% chance of this happening.

Hello Marius

>> No.10969371

>>10969113
Wrong and absolutely false.

Artificial neural nets are correct AT FUCKING BEST 60% of the time.

Stop coping and making up these false stories.

>> No.10969381

>>10969353
Marius only makes predictions with 100% certainty lol.

>> No.10969419
File: 29 KB, 507x507, 6s8bmosz3lvz.jpg [View same] [iqdb] [saucenao] [google]
10969419

>>10969254
I'll believe it when I see it. It took 3 years for BTC to fully recover after Mt.Gox caused the 2013 bubble to pop. I bet it takes at least that long for BTC to recover after this bubble finishes deflating. We won't see 20k again until after the price stabilizes and the normies collectively forget the pain from this bear market. Since the total market cap was higher this time than it was back in 2013-2014, it will take longer for the normies to forget the 2018 crash than it took for them to forget the 2014 crash.

My best guess is we will see 20k BTC again in 2023 and 100k BTC in 2024.

>> No.10969423

>>10969353
No, not Marius. He runs similar models though. We don't like the attention, nor do we sell the information for profit. Many of us come from MIT. We are scholars.

>> No.10969428

>>10969349
How big a portfolio do you need in order to afford this software?

>> No.10969437

>>10969193
bitmain fixed it for us by effing themselves so now they cant afford to spam the mempool anymore.

>> No.10969496

> statistical model
turns out, bullfags are really this retarded

>> No.10969505

>>10969371
The whole model isn't based off neural nets. We have a formula that works based off the data set. Machine learning is used to calibrate several coefficients that are applied to the derivatives of the formula.

>> No.10969523

>>10969349
I'm willing to bet that you guys didn't cross-validate on a dataset that wasn't used for training the model and are now bragging about how "well" you overfitted.

If my speculation is correct then your predictions are utter garbage. If my speculation is wrong, then what sort of training/validation split did you use?

>> No.10969524

>>10969423
You ARE Marius.

Marius is the only fucker in this entire universe who could believe bullshit like this:

>Statistical models say there is a 80-95% chance of this happening.

It's time to stop Marius.

>> No.10969530

Anon, I.... I don't know how else I can tell you

It's over. You need to move on. The money is gone and it's not coming back

>> No.10969535

>>10969505
I believe in you Marius.

>> No.10969571

>>10969505
>Open excel
>Do linear regression
>Bitcoin is higher now than it was back in 2008, so the model predicts that bitcoin will go up forever
>Brag on 4chan about your fancy statistical model

I know you probably aren't using linear regression, but you still sound like a fucking moron.

>> No.10969584

>>10969423
>Pretends to be a scholar
>Knows who Marius is
OP is Marius confirmed. AHAHAHAHAHAHAHAHHA

>> No.10969597

>>10969505
GANs? I've been trying to GAN an order book

>> No.10969606

>>10969571
but dude hes from MIT and a scholar. He used some fancy five factor ASSet pricing model. Heavily modified at that. Straight from a research paper too. Who are you to question anything?

>> No.10969610

>>10969584
Fucking larpers today man. Price action brings them out of their faeces-covered caves.

>> No.10969613

>>10968976
so we are finishing an enormous wave 2?

>> No.10969634

>>10969571
Linear regression is better than meme lines at least.

>> No.10969644

>>10969571
Hey, that's just your opinion. If you are so smart why don't you have a model? It's easy to shit on ideas. The algo has come a long way since 2014. Some of our mentors where mentioned in these articles.

http://news.mit.edu/2014/mit-computer-scientists-can-predict-price-bitcoin

https://www.technologyreview.com/s/610614/how-network-theory-predicts-the-value-of-bitcoin/

>> No.10969665

>>10969644

"Where mentioned"

Into the trash it goes....

>> No.10969675

>>10969644
Marius, you said we don't have enough data to predict altcoins, right?

The truth is we don't have enough data to predict Bitcoin either. Ten years is not a lot of time in the grand scheme of things.

>> No.10969699
File: 30 KB, 450x450, 1535989087489.jpg [View same] [iqdb] [saucenao] [google]
10969699

OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN
OP IS MARIUS LANDMAN

>> No.10969701

>>10969665
Do as you please. We just wanted to post this here so we can claim that it was on /biz/ first.

>> No.10969705

tfw you just want a glimmer of hope and this OP is crushing your soul with his LARPing

Been on 4chan since 2006 and you have successfully hurt my feefees via shitposting for the first time, OP

>> No.10969709

>>10969701

Yea learn to spell first ya brown skinned dothead

>> No.10969727
File: 39 KB, 604x569, 1434683403038.jpg [View same] [iqdb] [saucenao] [google]
10969727

>>10969106
We are still higher today than exactly one year ago, and the ATH was in December of last year, so it's not impossible, however unlikely it actually is.

>> No.10969752
File: 31 KB, 400x400, 1531586169195.jpg [View same] [iqdb] [saucenao] [google]
10969752

>>10969705
Marius can sell you hope, you just have to believe in the super advanced models designed by his 'team' of MIT phds.

>> No.10969753

>>10969106
>crypto never produces surprising and seemingly impossible events within the market
until it does; hell, DROP might go to $4000 EOY, never say never, right?

>> No.10969850

>>10969571
>>10969644
HAHAHAHAHA, you responded to a post in which I made fun linear regression by sending me an article in which you admit to using BAYESIAN linear regression.

Well, excuse me Mr. Fancy Pants. I still think you're a fucking retard.

> why don't you have a model?
I use RNNs to trade stock options. I haven't put any effort into crypto.

>> No.10970007

>>10969850
Well, excusez-moi monsieur candyass.
I still think you're a fucking retard.

>> No.10970020

>>10968976
the absolute fucking state.

>> No.10970047
File: 567 KB, 731x1172, 1523158528962.png [View same] [iqdb] [saucenao] [google]
10970047

>j-just hold a couple of years more

>> No.10970058

>>10969644
Thank you for your dedication OP
>ignore uneducated low IQ pigs and don’t throw pearls at them

>> No.10970069

>>10969850
Nope. Not linear regression. Those articles where meant to give you insight on some of our mentors. Not our methodology. We use RNN coupled with LSTM and MIST (superior version of NARX).

>> No.10970070
File: 11 KB, 456x285, desctri-conti-dd.png [View same] [iqdb] [saucenao] [google]
10970070

>>10968976
That is the most obvious monthly descending triangle in the world. Anyone with a brain unironically knows we are at least falling to 4000$ before moon.

>> No.10970137

>>10968976
Thank you for this, even though I find it hard to believe.

>> No.10970161

>>10970069
Ok, now we're talking.
>MIST (superior version of NARX)
Do you have links to any good papers about MIST? I haven't heard of it. I briefly tried searching google and arxiv but not much came up.

>> No.10970275

>>10970161
https://arxiv.org/pdf/1702.07805.pdf

https://pdfs.semanticscholar.org/5fa7/39bb5a06f9ef1f17a51959eb8b326560b99c.pdf

https://arxiv.org/pdf/1806.06384.pdf

>> No.10970279

>>10969644
OP is such a dumb faggot damn... Mixing several terms, mentioning factor models and then linking to an HFT strat, linking to absolute fluff pieces saying shit like
> doubled their investment in 50 days of trading

and then here's the paper:
> https://arxiv.org/pdf/1410.1231v1.pdf

guess how they doubled their fucking investment? Thats right by running a simulation without
> spread
> fees
> slippage
this is why financial academia is such a fucking joke...

>> No.10970407

>>10969419
the difference is there hasn't been a scandal of that proportion since, just waning enthusiasm

>> No.10970684
File: 2.13 MB, 4032x3024, C93A7807-2C4C-4FA1-A6B8-EC8EC0FD335D.jpg [View same] [iqdb] [saucenao] [google]
10970684

>>10970279
You clearly too low of IQ to comprehend, drink your milk and go to bed

>> No.10970796
File: 74 KB, 623x713, 1meta_203992_290.jpg [View same] [iqdb] [saucenao] [google]
10970796

I mean, can we just, for a second, ask a simple question, may I pose, I will, is there any reason Bitcoin CANNOT be $100,000? Because I'm certainly convinced now from this thread it might actually be happening by eoy.

>> No.10970820

>>10968976
I'm going to wake up tomorrow and everything is going to be even more miserable than today

>> No.10971018

>>10970820
October will be a good month. I promise.

>> No.10971097

>>10970275
How can you use RNN to predict the price?
What are the inputs except for price history?

>> No.10971198

>>10968976
can confirm, i work at jp and were about to release crypto trading to the public on friday

>> No.10971451

>>10970796
Please get off /biz/

>> No.10971633

>>10969419
>best guess
based on nothing but "I think"?

Retard. Atleast think about what the next bubble will be about. Smart Contracts are due for a bubble and they'll be widely used by 2020.

>> No.10971743

>>10969254
>There is no reason to look for reasons when there are none to look for.
brainlets here can't trade without a narrative to guide them

>> No.10972548

>>10969008
hey look, the first actual correct graph on 4chins. Unironcally.

>> No.10973013

>>10968976
So next stop is 9300?

>> No.10973034
File: 96 KB, 1425x1450, btc-prediction-eoy-06-09-18.png [View same] [iqdb] [saucenao] [google]
10973034

>>10968976

>> No.10973059

>>10968976
>that crash from 120k to 60k in a month

>> No.10973107

>>10968976
OP don't be a cuck, come back here and tell me how to fucking swing trade to increase my stack.

>> No.10973150

>>10973059
Short 5x at the very top from $120k (on the downside of a head and shoulders figure) and you'll be rich by the time the trend switches back to bull after 60k

>> No.10973245

>>10973150
The thing is how can you be sur 120k is the top, maybe it will go further. In 2017 everyone was saying 12k is the top, it will crash soon, but it went up to 19k

>> No.10973261

>>10970275
Dude, https://arxiv.org/pdf/1409.7193.pdf and https://arxiv.org/pdf/1702.07805.pdf are using "MIST" to describe totally different things.

https://arxiv.org/pdf/1702.07805.pdf discusses MIxed hiSTory RNNs, which are truly similar to NARX.
https://arxiv.org/pdf/1409.7193.pdf discusses "a Momentumized Iterative Shrinkage Thresholding algorithm" which as far as I can tell is just a fancy way to apply an l0 regularization penalty to high-dimensional linear regression.

Did you even bother skimming https://arxiv.org/pdf/1409.7193.pdf before you linked to it, or did you just assume it was about mixed history RNNs because MIST was in the title?

Also, that paper by Guo isn't even about MIST. He used NARX.

>> No.10973286

>>10973245
The top will be the average price of the head and shoulders. It will look like /\/\/\ or maybe just two tops. Either way it'll look something like that in daily candles (don't go for short duration tops and bottoms, always zoom out) and even if you sell the bottom of these fluctuations it will still have a long way down to correct.

120 was used as an arbitrary figure, nobody knows how much and for how long it will go up or down, except for the market makers

>> No.10973295

>>10973286
>don't go for short duration tops and bottoms, always zoom out
What timeframe is best? 1H, 3H, 1D, 1W?

>> No.10973307
File: 200 KB, 1120x840, 1527178384976.jpg [View same] [iqdb] [saucenao] [google]
10973307

>>10969371
Im on /biz/ and you cant make me think rationally and make my own research. So Ill believe what appears to be good for me. Checkmate, this is not delusion, its CHOOSING A TIMELINE, holy shit I shouldve known this earlier, quantum financial freedom here I come!

>> No.10973336

>>10973295
1w for determining trends and 1d for daytrading. Hourly candles are a bitch to get good info out of. For example, by using 1w candles I get a feeling like LINK hasn't bottomed out yet and that it is currently on a bulltrap for a month. If you switch to hourly candles it looks like a perfect buy on the way up.

Guess I'll find out soon enough if I fucked up on that swing or not.

>> No.10973410

>>10973336
Makes sense. Thanks for sharing friend.

>> No.10973425

>>10973261
The papers are influenced from one another. MIST = Mixed history recurrent neural networks it works for Bitcoin b/c of the network effect.

https://github.com/rdipietro/mist-rnns

>> No.10973601

>>10969699
keked and checked

>> No.10973603

Marius is the Nate Silver of 2018

>> No.10973627
File: 37 KB, 800x450, brainlettttt.jpg [View same] [iqdb] [saucenao] [google]
10973627

>>10969437

>It was bitmain, not the thousands of normies trying to use bitcoin.

I'll just say good luck with you being so gullible... you'll need it