gincarbone / machine-learning-bitcoin

A machine learning LSTM model to predict Bitcoin prices

Machine learning Bitcoin

Is a machine learning LSTM model to predict Bitcoin prices of netx month. For pricing it needs historical bitcoin price data from 2015 to date. The designed neural network is an adaptation of the one used in the course "Deep Learning A-Z™: Hands-On Artificial Neural Networks" but with the increase in the number of 64 neurons in the Hidden Layers. Thanks for this! Some changes have been made to the Python script in order to configure the main things in the first lines of code.

How It Works

Machine learning Bitcoin DOES NOT use algorithms and financial metrics. It is entirely developed in Python3 with Keras Lib and Tensorflow backend. It uses an RNN (Recursive Neural Networks) neural network made up of several layers of 64 neurons each. The historical depth of analysis that is able to manage for the prediction of cryptocurrency prices is 120 days. This means that for the calculation of future prices the experience of the previous historical series is used with a 120-day window of action.

Dataset

Dataset from Investing.com historical data https://www.investing.com/crypto/bitcoin/btc-eur-historical-data

Predicted prices vs real price

Offer me an Italian Coffee

these are the digital addresses to make a small offer and support the project. Thank you.

BTC = 17fYA4hB6BDikGSqcyYjC3dCnU56bjeN2J

ETH = 0xdc66Bbf32580d358e3eC021A20B51a09AFeB993e

LTC = LgzpfTNJjjmqP2mcpkJNYeR4HYupU4BFTP

Thanks

Deep Learning A-Z™: Hands-On Artificial Neural Networks