Hey frens

How can you predict this accurate in 2009? Show the whole time frame from 2009 to now for reference, fren.
Also, what are the different colors? Guessing its scenarios with slighty different hypotheses?

Attached: 1546285647643s.jpg (125x125, 3K)

interesting chart. what is the basis of your predictions? what are you measuring and extrapolating?

sorry its from feb 9th of this year. it takes the pump that happen today into account. been posting for a while but no one believed me

its based on periodics instead of price estimates. it regresses on the assumption that different market participant groups trade based on different time scales from short term to long term. the different colors are regressions on several found clusters within the long term groups.

dude you need a trip, or a particular writing style

You are why I come here.

Good. Clusters being what exactly? Institutions, groups and so on? What constitutes these groups or sets them apart?

This may sound very in depth questioning, but Im just trying to get the general picture. If its clusters = groups of people then this sounds like alot of insider info is involved.

Whats your background in terms of work experience or studies?

Attached: 1546287356612s.jpg (125x122, 2K)

sorry im no good with writing, concepts being especially hard

thanks fren ^_^

clusters of 'time scales'. sorry if this is vague but lemme try. some look at 1m charts and trade at high freqs. some at 1h. some at 1d, 1m, 1y and so on. these are different groups/clusters in itself, but i found we can increase the granularity and find more clusters within these 'groups' that have significant differences, or profile 'gaps' between them. OP pic is that of those i found in the 1y/1m groups. higher granularity leads to more 'specific' waves, but the error also increases more quickly the farther i predict into the future. lower granularity is more vague, but error std stays constant further out into the future.

what i did was map out the cycles, or periodics i.e. ups and down, instead of trying to estimate exact prices.

my background is in machine learning, but i went to grad school for bioinformatics. data science with crypto is just a hobby.

Uhhh so, buy now or stay poor?

depends, fren. how long are you willing to hold?
if you have ADHD, dont buy now.
if you have the fortitude of a buddha, there will never a better time time to buy than now.
everything in between applies.

I have 1 bitcoin and 5000 link. Am I foolish for neglecting my retirement in the belief I'm going to make it?

I evenly distribute my bi-monthly DCA into shitcoins, BTC and ETH and will hold all of them to 0$ If nexesaary

Based user fren

How high is your iq? What do you do for a living? UPVOTED

what kind of regression is this?

Not even close to bottom yet.

when do you retire? if its past 2030, you can sell at even at the lowest point of the bearest bear market then and still live a comfortable life till death. you might not even need to sell, if you know what i mean.

dunno about link, but if the memes are true, you should be golden. alts usually have a way of dying out according to paretos rule though. dunno.

DCA is the way to go. btc and eth will not go to 0. bearish market in Q3 though.

wow OP you are a golden nugget

Anything I should know about ETH? Is a "flippening" something to take seriously?

Is it good or bad to be in ALTS vs BTC?

Still kinda confusing for someone with no technical or economics background, but I like you gave it a shot. Maybe I'm just too stupid.

Did you do well in crypto so far? I, as many others, failed to sell (more - I sold a bit through Nov - Jan) during high times... bothers me a little but I wouldnt have made it by any standards so meh.

Attached: 1546293843459s.jpg (125x125, 3K)

I do agree with this cyclical nature of BTC, but I can't help but feel like it burned all of it's fuel in this last "crash" with regards to mass adoption and general populace interest. What makes you think we still haven't bottomed, and that it can generate levels of normie interest higher than these past few years?

This I dont worry about. People will always be greedy and once a hint of bull season appears, there will be fomo just like 2017.

Attached: 1546286674629s.jpg (82x125, 2K)

Write me an email srsly prohibitdw (at) gmail.com... im in serious need for smart lads interested in ai/ edge/ deep machine learnin ) neural network mates niche

thanks fren. i do data science on bio data, mostly epigenetics / histone mods. seems its the trend among geneticists these days.

I can't say much about alts, ive only worked with btc. but what i find is that alts follow paretos rule. try applying something like zipfs law to normalized btc dominance data and youll get a picture of the alts that will survive/die out.

i have a tech background and ive been in btc since 2013 because of it, so ive done quite well even without algorithms and such. but i discovered the s curves that eventually became metcalfes law (the lowest res / granularity trend) back in 2014 on the bitcointalk forums which gave me some ideas on algos.

if you have trouble selling, make sure you keep it that way long term. otherwise, try to take profit now and then in small amounts. i had the same problem, i never sell, which is why ive been building algorithms to tell me when and what to do.

So basically ETH will do good? Thanks

Second question OP unrelated

How hard is it to self-learn machine learning? Like studying R and Python and shit. Is it possible to learn these things on my own and apply it to things like crypto like you do? Where would one even begin?

Also OP I would pay you for your analysis

by comparing the past to present. back in around 2015, i was using btc on places like alphabay. now we have legit etfs and insts. like bakkt trying as well as wallets preloaded on phones, etc. it doesn't matter if they succeed or not, they will come and go with time. but just the fact that they are trying get a piece of the pie tells a lot.

Actual code itself isn't complicated if you know programming and there are libraries that do the dirty work for you. The real difficulty lies within knowing what you're looking for, how to look for it and how to interpret the resulting data. (so it's more of knowing the area you're generating data for than actual code)

I'm a programmer but I'm not really into ML so I could be wrong tho.

theres no legend for this op the fuck how am i supposed to read the other lines that arent the yellow one

ETH is good, the hard part is seeing other small cap alts massively pumping (then dumping) and not getting tempted.

whats your math and coding background like?
machine learning is all math. with coding, you can easily apply prebuilt libraries and such, but theres a limit to how far you can get because out of the box libs are usually generalized to the max. i find that playing around with these does eventually lead you down to the maths though.

im thinking of monetizing my analysis, but it in itself already makes profit so im in a bit of a dilema. ill probably be posing the results, even the high res predictions, on Jow Forums from time to time, ive already been testing and doing so for a few months now.

sorry, this thread was supposed to be a 'ha i told you so' thread that i explained in a previous one. more detail here: the lines are clusters of long term cycles and have a high error standard deviation but remain constant for a long time out into the future. so think of them more like a guide line.
i wouldnt trust it too much past 6-7 months out from prediction date.

last question before bed time.
How good of a short term prognosis can you make with your algos given u want relatively good accuracy asserted to the forecast? I guess my question is: Do your algos help with daytrading? If not, whats the shorted time frame you see them working in?

Night night and thanks for the thread.

Attached: 1509546580414.png (750x775, 930K)

>ETH is good, the hard part is seeing other small cap alts massively pumping (then dumping) and not getting tempted.

I plan to hold some good alts long term, like OMG LINK REQ RLC KMD ZEC

>whats your math and coding background like?
6 months of JAVA in high school, kek.

>relatively good accuracy
this itself is a bit relative. if i use daily data, its good up to about a week, with errors (both in price and time) slowly increasing afterwards. anything smaller, such as 1m, 20m, 1hr etc has less error std, but is only good for up to a day or two, then the errors increase exponentially. Also, data with higher resolution is harder to come by. its really a gradation then a hard set error depending on the time scales i use.
im really working on weekly data because it seems really stable for a long periods of time (albeit the high error std on the daily jitters as you can see in the OP pic) and i can use it as a guide line for the finer predictions. i also need more time validating the results due to its scale, which is why ive been posting on biz but so far the peaks and troughs are on target.

hope that helps, night night fren

If that was around the higher low, then I believe I was getting laughed at too.

>OMG LINK REQ RLC KMD ZEC
again, fren, i haven't done much analysis work with alts so i dunno, but hope it works out for you. I do hold OMG and LINK with a bit of RLC though, but its more just fun gambling for me.
The best method i found is to analyze btc, then distribute investments in a pareto fashion across alts based on the btc analysis results since theyre highly correlated at this point.

>6 months of JAVA
try out python, it should be much easier than java. it will eventually lead to using ML libs, then digging into maths if it fits you. if youre good with stats, start with R.

Attached: hugs.png (288x175, 7K)

...