Ethereum Future Price Prediction
Can we predict cryptocurrency prices using machine learning? We’re going to build a Keras deep learning model that attemps to predict the future price of cryptocurrencies like Bitcoin and Ethereum in this video. The type of model i’m using is a bidirectional LSTM recurrent network. Ethereum future prices as well as other cryptocurrency prices are hard to predict, but with the power of machine learning we can find a suitable prediction.
Code for this video:
Want more education? Subscribe! And connect with me here:
More learning resources:
Join us in the Wizards Slack channel:
And please support me on Patreon:
Signup for my newsletter for exciting updates in the field of AI:
Hit the Join button above to sign up to become a member of my channel for access to exclusive content!
Most people don't understand the concept of "buying the dip" buying the dip is all about buying digital assets when their prices are down and selling off when the price rises. Holding is profitable, although trading is far more profitable. I was able to grasp the knowledge of trading crypto assets early enough, but i was still limited due to my lack of technical understanding of how to analyse the digital market, all that changed when i encountered Daron Wilber's services., although i have been into numerous services but he stands out with experience and expertise playing in his favour. I must confess it wasn't an easy task to learning the routes on trading but with the assistance of Wades it was more easier to understand, Here are his details on ω𝐇𝓐𝓽ѕ𝓐ρ𝐩 +1(431)813-5396 for his assistance on strictly crypto related concerns.
You know trading, depending on the startup this can be quite quick, or fall in to the Long Term bracket. You do your due diligence on the asset, buy your cryptocurrencies, buy their asset, and hope you chose correctly selling when the price moves higher and for better investment chances i believe broker24 platforme is the real deal
Guys $ALPA Trading Contest is ongoing don't miss on #MXC 15,000 $ALPA is waiting to you https://twitter.com/MXC_Exchange/status/1330699118247698434
You should look into Zilliqa its much better then ETH its like ETH 3.0 much faster fees are really low
God bless you sir, my masters thanks you 🙏🏾
Anyone has a data file that works? Please!
Now i see why google hired you XD
Can anyone give me a reason why he shuffles training data in a time series problem where the order would matter ??
Does anyone know why..
@siraj: The model is taking too much time to train any specific reason for the same?
How do you predict the future prices ? Do you take the last window size data and predict next value and use this predicted value as our prediction set to make the feature columns, because in the test data you are not given the prices so how do your create the price window for the prediction dataset ?
Hello Siraj, amazing video, I have a question, why would you do that normalization and why is it useful? why not using other kinds of normalization? Thank you in advance.
I will try to predict everything price!
I am a big fan of your work.
Why aren't we using train_test_split directly and then from sklearn.metrics to import confusion matrix and classification report? That would save us a lot of time right?
Well things have changed since this review…
Don't you know you can create your own AI using LEDU coins? They can guide you step by step. Purchase LEDU coins, then go to LiveEdu dot tv and purchase the premium account then you are on your way.
So why are you still here and making videos to "teach" us this instead of making billions of dollars out there implementing this? You must be such an altruist.
I'd like to see a model based on Bitcoin, Ethereum, etc. users and adoption
Man, this is the best youtube channel that I've seen by far, even better that TED's, I have a youtube playlist that it's call SiRaj.
this is a very good tutorial on cryptos prices prediction, using bidirectional LSTM is interesting. however, in reality, how can people predict crypto/stock prices using data from the future?
anyway i appreciate that u take into account some important cryptos features such as the volume, hash rate, block size, market cap…etc but not just open and close prices.
i think i will implement my own version which takes into account of those features but just using a normal lstm recurrent recurrent network
Why bidirectional? I dont get it…
Scam. Everyone knows when someone publishes something smart that "makes money" it wont work. If it worked, it would not be published. You don't need an Einstein degree to understand that…
YOU are the man dude!!
Siraj, why shuffle the data? If we are interested in sequences, the order should matter to the ML, as it could have some periodic features. When you shuffle the data you don`t loose this things???
Thanks for the video..
Siraj, why you don't drive a Lamborghini if it works?
Hey Siraj! I’m from the DigiByte community and would like to ask your oppinion about it. Thanks!:))
For the others who doesn’t know Digi do some research on it, DGB is the most powerful UTXO(BTC,LTC,VTC..etc) blockchain in the family. The community is awesome so feel free to join! #TheFutureIsDigiByte 👽🔮
Your tutorials are so great, thanks man.
Hi Siraj. thank you for this video. Is it possible to fit the RNN also during testing? Or how can I fit a RNN also during run time? I want to keep a RNN updated with real time data streamed from a API and still learning (fit) as long the app is running. Or is the method test also updating the RNN?
Anyone built or building a bot based on Siraj's model? Collaborating together to speed up the build? Would love to hear any success stories applied to markets if you're already ahead, but I'm just now stumbling on this video
I really love this video <3 <3. However, I cannot replicate without the .cvs dataset. :((
Your training and your test data overlap by sequence_length-1. This might have biased the results.
90k telegram no. 1 channel
Join this channel now for daily 50-100% profit signals
I have predicted ETH and DASH prrice in Jun 2017 ( find my comments) I said 1000USD for 1 ETH up to end of the 2017 .
Now I am saying ETH and DASH price will be easy 2200-2600 USD by the end of the Year.
Thank Siraj for the tutorial! Will you be able to share the data file you use?
Siraj, Isn't using confusion matrix and precision/recall/F1 for classification problems only? Isn't this a regression problem (MSE) and should use R2 score instead?
You are interested in cryptocurrencies and you look for where to invest? TokenGo – the platform with unique approach and the relevant ideas. There is a ready site for investors in case of the organization of own ICO. I think the company will find the niche on a blockchain market.
Great video my dude. Btw, it's not listed in your crypto playlist. I got a bit lost trying to find it again.
Anyway, I'd love it if you could do a follow up where you have real time data being fed through.
I'd love to see how you approach the issue because my attempts aren't going so well.
I was thinking of using Coinigy for realtime data or maybe even an exchange directly (I know binance has websockets api)
Or even have a separate application cache the data in redis and just fetch the information from there (I know that's what I'll probably do).
Corl.io is looking for machine learning developers who are crypto enthusiasts. If you're interested, send us an e-mail with your CV.
Why do we shuffle the data? dont we lose the connection through time?
Hey Siraj do you mind explaining why you used a recurrent network instead of a normal feed forward network. Is it because since the recurrent network has the LSTM layers its able to remember and take into account previous price points making it better for predicting the price of bitcoin?