How I use machine learning to pick my crypto portfolio

[crabby]

Machine learning has become a hot topic. With the release of GPT-3 we are in quickly changing times. As we continue this trend, we will begin to see more investigation into how we can apply machine learning to crypto portfolio management. In this article, I briefly discuss how I use machine learning to select my portfolio.

Selecting my portfolio with machine learning.

Check it out and let me know if you have any questions!

[1714115732]

1714115732

[nelson4lov]

I've skimmed through the article. It seems like an interesting read so I've added it to my reading list which I would check out later. Machine Learning is a really innovative technology. I've made some models in the past but for picking portfolio assets? That's another level on its own. The things one can accomplish with machine learning these days is truly amazing. Btw, I wanted to know if there's any way to make some configurations etc.

[Kelvinid]

It is really interesting to learn and much more interesting to see its effectivity as we are living in a volatile market. We are still unpredictable and you've mentioned 7D prediction, it will pretty hard to rely on it.

With this machine or any tools innovation, it is still not warrant anything but it is still all in the possibility. I believe it closely works with that but not it gives assurance to point out where we go in the coming days. Maybe this could work into a stable market but not in crypto. I hope I'm wrong and you can prove its effectiveness once fully operated.

[Twinkledoe]

It is really interesting to learn and much more interesting to see its effectivity as we are living in a volatile market. We are still unpredictable and you've mentioned 7D prediction, it will pretty hard to rely on it.

With this machine or any tools innovation, it is still not warrant anything but it is still all in the possibility. I believe it closely works with that but not it gives assurance to point out where we go in the coming days. Maybe this could work into a stable market but not in crypto. I hope I'm wrong and you can prove its effectiveness once fully operated.

Maybe just to test the efficiency of this tool but this is still not reliable when you actually implement to crypto. Especially those coins that have short lifespan owed to their dev teams, with no long term vision but just to rip money from the gullible investors. You can also check it also in other established coins, try starting in bitcoin or eth, and check how far or accurate you can get.

[stompix]

Check it out and let me know if you have any questions!

Why would we ask questions when all you do around here is posting topics advertising your blog?

Basically the whole "article" is an ad for nomics(dot)com and their so-called machine learning predictions.
Well, let's tests their machines, this is their 7 days best buy



Just for fun

Bitcoin      $9,116.18
Ethereum       $233.98
Bitcoin Cash   $216.66
Litecoin        $40.27

LE:
I'm really bad with tables as it seems so I'm sticking to screenshots.

[pooya87]

you can only computerize things that can already be done by a human and most importantly have patterns to work with specially when it comes to machine learning. altcoins have no pattern and are mostly unpredictable thanks to their pump and dumps which is by nature unpredictable. this means having a portfolio of altcoins is the wrong step on its own let alone wanting to computerize the process of managing it!

[leea-1334]

I do not really know if machine learning will be good enough for usage today, but I guess we have to accept that a lot of use cases already show machine learning success (like in predicting what we want to see on google search for example) but in crypto it has not done very well yet. I remember many AI projects using NLP even since 2016 and I even joined a couple. Not very good results (2017 everything did well so that screwed the results).

[beerlover]

I hate going through clickbait links. If you really care about this community then you must include your article as an optional for those who do not mind visiting external links. This is not the right place for trying to monetize your contents. Intentionally driving traffic from this forum is a neg-able offence as per DT level community guidelines if I remember correctly what I learned lately.

Machine learning is what we are already enjoying in the name of artificial intelligence based bots and its based trading. When people are already moved into a level of making trading decision with the help of AI bots, I guess managing a portfolio with the help of those bots will not be a big thing. (I have seen many people are claiming that this human race has not achieved up to the level of artificial intelligence but so far only machine learning has been achieved and you are seeming honest up to that level .)

[Wexnident]

Hmmm I don't really see any use for it. At the very least, whenever I make a mistake in my portfolio, I have myself to blame. But using ML? Nah, I'd just make up excuses and the like when the fault really lies to me, and that kind of ruins the attitude I should have when dealing with the market. Plus, ML still has a lot of problems imo, yes the idea seems pretty nice and good, but hell whenever an odd pattern is felt, ML has that possibility to actually ruin everything it has learned by trying to adjust that certain pattern. In the end, it's literally just filled with uncertainties that makes the choice of self management a lot better than ML.

Machine learning has become a hot topic. With the release of GPT-3 we are in quickly changing times. As we continue this trend, we will begin to see more investigation into how we can apply machine learning to crypto portfolio management. In this article, I briefly discuss how I use machine learning to select my portfolio.

It's legit you just pressing 3 buttons and you're done tbh.

[stompix]

7 days have passed and we should change the title to
"Why not use machine learning to predict the price"

The first column in the predicted price, the last that actual price in 7 days


That is for their top earners.

Honorable mention:
They predicted the price of bitcoin to be  $9,116.18...we all know how that went. 

[el kaka22]

First of all I would like to thank you for using the word machine learning instead of trying to hype it up and call it an A.I. , a lot of places call machine learning artificial intelligence, it is a moral and well.. vocab related discussion if machine learning can be considered A.I. and what is actually A.I., if a software is given parameters and learns within that and becomes better, is that intelligence?

I don't know any of that, but when people say A.I. everyone thinks of stuff like ultron so I am sure most places uses that knowing people will think that there is some be all end all robot that can dominate the world with how smart it is and can make jokes but instead it helps you trade bitcoin for some reason. Anyway I think machine learning is the right thing to say and I want to thank you for that.

[joniboini]

Wonder how their algorithm is. Looks like they use a short-time frame and count how many times price action failed/break through a resistance and support and then decide where the price will go. Which is why they predict bitcoin to go down. If that algorithm is really true then I guess they won't predict a new bull run.

[stompix]

Wonder how their algorithm is.

Tossacoin? 

Looks like they use a short-time frame and count how many times price action failed/break through a resistance and support and then decide where the price will go. Which is why they predict bitcoin to go down.
If that algorithm is really true then I guess they won't predict a new bull run.

It might be but, but I'm wondering why the sudden spikes in some cases and the lack of it in others, although the number of spikes seems quite the same or totally the opposite in some other pairs

https://nomics.com/assets/band-band-protocol
https://nomics.com/assets/kava-kava

In the 1 month graph you can see their prediction, I think they are adding something in that, wonder if it's volume or "social activity" and at one point they do recalculate the previous price evolution and adjust the prediction.
It can be clearly seen here:



Probably some TA experts might be able to guess what they are using, for me once they failed like this, a service that gives you a 50% gain backed by "machine learning" and ends with 30% loss, is just toilet paper material.