Welp, there it is guys

Attached: btc.png (1362x723, 186K)

AAAAAHHHHHHHHHHH

autism

its really embarrassing how bad technical analysis actually is.

it appears to be a graph of the historical btc/usd price compared to somebody's (presumably OP's) prediction of how the BTC price should be (presumably based on """TA""")

haha. Can I see the code that leads to a 0 dollar 2019 prediction?

That website is absolute hogwash. According to this website, all cryptocurrencies are heading straight to 0.

I already posted what this is. Scroll up

>its a forecast for next year based on generalized additive regressive models created and used by facebook data scientists

they are though

>dubs confirm
so is this good or bad for crypto?

Attached: eth.png (1311x657, 130K)

your dubs double confirm. The forecast says that we will land somewhere between 5k and 23k at EOY 2019. It is not the 100k or million delusional people are looking for. However it is higher than todays prices..

nevermind I can't argue with quads

>quads confirm what i've been saying all year
kek
fuck yeah. thanks user

and why would you put any stock in it?

firstly, the range it predicts appears to be "somewhere between $4,800 and $22,000 by 2020"

secondly their prediction line has not really followed the btc price at all in the entire model

lastly the data facebook is crunching (demographics for advertisers) is a lot more predictable than shitcoin gambling

that's ETH, it's a shitcoin. King Daddy will be fine

If it hits 23k in 12 months the normies will start to lap it up and maybe we can moon the year after

2020 moon race

Is TA a self fulfilling prophecy?

some kinds of technical analysis can be, but not this kind. this is just people not understanding the math theyre using to make pretty graphs.

TA is a complete nonsense. The market is random

Inb4 4 years to next bull

brb, going to run the simulations for up to 2023..

this is what more people need to understand. TA can help, but crypto is entirely speculative so the markets do whatever they feel like most of the time. And most of the time what the market wants is just code for what whales want.

are you retarded? he posted it to show how wrong it is. look at the fucking price it's already 20 sigmas outside their 95% confident region. you idiot

>bragging about a model without showing how it does on a cross-validation set

Attached: 1512786520508.jpg (645x729, 49K)

>babby's first boom/bust cycle

>Optimization terminated normally:
>Convergence detected: relative gradient magnitude is below tolerance

Attached: bitcoinForecast.png (1000x600, 47K)

yup, maybe 2024 (after great recession of 2019)

>Predicts negative prices for some of the times in the training data
Dude, even brain-dead models like linear regression or Monte-Carlo simulations would get you better results than whatever the fuck you're doing. At least they predict valid prices at all timesteps.

that's just drawn scaled up to see fluctuations.

Really? Your axis doesn't indicate that. It is also weird that you'd do that to the predicted prices but not the actual prices. It genuinely looks like your model returned a lot of negative prices in the early timeframes. seems to confirm that it is capable of returning negative values, which is pretty fucking dumb. If your goal was to accentuate predicted price changes in the early timesteps you would have been better off using a log scale.

This does nothing to address the criticism in the post you were replying to. Try training it on a subset of the historical data and seeing how it does on the rest. Then compare how well it did on that second dataset to something like Monte-Carlo simulations to see if your model is actually returning anything useful. Ideally, you would perform this sort of testing with a variety of currencies and timeframes to see if your result from the first test was just a fluke.

From what I can see in the first graph you posted, it looks like one of the models learned that "BitcOiN ALwAYS GOes uP" and the others just oscillate.

Attached: Screenshot from 2018-12-11 23-02-31.png (312x39, 6K)

A model done with the prophet library facebook?

>according to our data scientists, the price is either going to go up or the price is going to go down

Attached: xs.png (486x515, 20K)

Crypto is even better than a Ponzi scheme.
Usually when you take the money and run, the Ponzi Scheme is done and over.
With Crypto, you can just make up new "co(i)ns and people come back thinking it's THEIR turn to rake in the big money. It's a never-ending cycle of pump-and-dump grift.
Anyone who hasn't cashed out months ago is so terminally stupid that they genuinely deserve to be taken for everything.

Attached: Alan_Keyes.jpg (416x431, 57K)

why is why alts will never succeed long term, because the only people that care about them are late adopters, and as soon as a new hot alt comes out, every late adopter jumps to that one, leaving idiots behind like those bagholding chainlink.

>linear charts

Attached: 9.png (456x359, 60K)