Many claim they can predict stock and cryptocurrency prices using machine learning; but no one can prove a profit on live data. What’s the catch?
Note from Towards Data Science’s editors:_ While we allow independent authors to publish articles in accordance with our [rules and guidelines_](https://towardsdatascience.com/questions-96667b06af5), we do not endorse each author’s contribution. You should not rely on an author’s works without seeking professional advice. See our [Reader Terms_](https://towardsdatascience.com/readers-terms-b5d780a700a4) for details._
If you’re reading this article, you’ve likely seen blog posts/articles online using stock/cryptocurrency data and machine learning algorithms to “predict” future prices. Here, I’ve demonstrated a similar project in which I use some metrics like Tweet Volume, Google Trends Volume, Market Cap, and Trading Volume to “predict” tomorrow’s opening price. In this example, we will focus on Bitcoin prices since it is a hot topic, but this can be extrapolated to any asset as long as the data is available. Naively, one may think on a day where people are tweeting about Bitcoin, Googling Bitcoin, and trading Bitcoin more than usual, we will likely see the price of Bitcoin open higher tomorrow than it did today. Further, maybe we can actually predict tomorrows opening price with high enough accuracy based on these data to ultimately yield a profitable trading algorithm. To test this hypothesis, I’ve gathered some data from Nomics, Google trends, and other free online sources to yield the following dataframe.
From a machine learning scope, we can think of these four mentioned columns as our feature set and our ‘Price’ column as our target. More specifically, tomorrow’s price is the target for today’s features. With a bit of feature engineering, we can also include changes in these metrics for the past day, past two days, past three days, etc, where a value greater than 1 indicates an increase and a value less than 1 indicates a decrease.
Neural networks, as their name implies, are computer algorithms modeled after networks of neurons in the human brain. Learn more about neural networks from Algorithmia.
Recurrent neural networks, also known as RNNs, are a class of neural networks that allow previous outputs to be used as inputs while having hidden states.
The purpose of this project is to build and evaluate Recurrent Neural Networks(RNNs) for sentence-level classification tasks. Let's understand about recurrent neural networks for multilabel text classification tasks.
Artificial neural networks (ANNs), usually simply called neural networks (NNs), are computing systems vaguely inspired by the biological neural networks that constitute animal brains.
Convolutional Neural Network: How is it different from the other networks? What’s so unique about CNNs and what does convolution really do? This is a math-free introduction to the wonders of CNNs.