Crypto modeling

Has anyone given serious crypto modeling a shot? Maybe something like a random walk with varying drift based on sentiment analysis?

Other urls found in this thread:

leaprate.com/cryptocurrency/bitcoin/yales-robert-shiller-calls-bitcoin-gold-fads/
twitter.com/SFWRedditGifs

>random walk
>sentiment analysis

OP, are you fucking retard?

EMH Crypto version?

You are clearly not familiar with any of the words he listed. The cryptomarkets are literally almost only sentiment driven, I don't think we can apply the random walk, since that would imply semi-strong efficiency.

fucking kek

Get real OP, trying to study crypto with that knowledge you picked up in academia is like trying to apply some advanced ML model to the random betting of drunkards on the outcome of games of darts or some shit.

Prices in this market are purely a function of the number of stupid people who have been shilled a coin and are willing to throw their fiat at it.

Shiller's fads model
>literally means market is driven by sentiment and not rational behavior.
This shit is text book behavioral finance

There's better ways to model crypto but they're not really price driven

Yes, it is working as intended. If certain parameters shoot up in the next day or so, we will see a btc run. It has been right over 80% in terms of pointing to a run within a 7 day period. Last time it pointed at Nov 11.

easier to just buy cheap stuff while the total crypto market cap is trending up

>Shiller's fads model
this guy?

leaprate.com/cryptocurrency/bitcoin/yales-robert-shiller-calls-bitcoin-gold-fads/

This.

good points. OLS with sentiment index, got it

Exactly.

I’ve got an evaluation model I have been running simulations through and they’ve been fairly accurate at picking up moon coins. Give me a coin and I’ll run it through user — will output “BUY” or “AVOID”.

Haha I want to try this shit
iExec pls

When btc is sleeping you can choose almost at random and make gains.

LINK

CAN

QRL

repeat after me: the floor is at 0

not that guy, I'm his AI posting for him....
my calculations say AVOID ALL

AVOID

AVOID

BUY

AVOID


— closest to being “BUY” other than CAN was QRL.

The full list of simulations I ran today:

ICX 0.08 BUY
VEN 0.10 BUY
XRB 0.30 AVOID
DRGN 0.03 BUY
OPT 0.10 BUY
COLX 0.38 AVOID
LRC 0.08 BUY
LTC 0.48 AVOID
NEO 0.20 BUY
ZCL 0.40 AVOID
BTDX 0.33 AVOID
TRX 0.38 AVOID
ETC 0.40 AVOID
DGB 0.55 AVOID
NEBL 0.25 BUY
SNOV 0.25 BUY
HPB 0.15 BUY
NCT 0.13 BUY
KMD 0.25 BUY
BAT 0.23 BUY
NXT 0.43 AVOID
COB 0.23 BUY
POE 0.25 BUY
ELF 0.30 AVOID
RLC 0.38 AVOID
LINK 0.38 AVOID
CAN 0.13 BUY
QRL 0.35 AVOID

POE

Lmao look at this guys can bags

#’s in the middle indicate risk level — closer to 0 the lower risk it is!

Merry Christmas ya filthy animals.

What does the number represent? Probability of expected returns based on historical data?

>tfw CAN might actually have a floor

...

...

>avoiding link
Your model is broken. I-it has to be. Reeeee

any tips on coding a trading bot user?

Crypto market doesn’t follow standard principles. Work on building a risk matrix for evaluating a rule set — asking questions of each crypto and grading on a scale

>tfw don't understand any of that
where do I start if im a brainlet that wants to not stay poor

uhh sentimental analysis, yes. most algo traders use it. Google trends, tweepy, reddit scraping, etc.

random walks? no. how would that help? (yes i know youre talking about the stochastic process)

I use a neural network classifier myself.
>t. ML engineer

Make a list of coins and start ranking them subjectively first.

Then find data to rank them further with less subjectiveness (example: $10M market caps).

Then build a preference set — $10M market cap, 50M circulating coins, within top X hundred... etc...

Soon you start building this out, in excel, and you have a table that progresses from subjective / opinionated sorting/ranking to scientific analysis/sorting based upon criteria that you developed.

Then once you have criteria build out in excel / google sheets, use python and a coin api to pull in data and filter through your criteria via simple code and you can now run through the full 1400 coins.

Does that help?

www google com

thank you user, I guess I need to buy a python book first

lol no.. i'm not an attention whore or have tits, so noone will give me crypto for modelling

No XLM?

XLM is a BUY

cool approach, what are your inputs?