Algorithmic Trading with Twitter Sentiment Analysis!

I packaged up my 5th and final project at Metis!

Te this project I combined my passion for cryptocurrency and Gegevens Science!

Specifically, my project wasgoed focused on predicting the price of Bitcoin using Twitter gegevens!

For this I used Facebook Prophet to create what I call a Nested Time Series within a Linear Regression prototype!

You can view the entire project on my GitHub.

How does Twitter gegevens relate to Bitcoin Price?!

Previous research has shown that Twitter sentiment can be used to predict market movement of financial instruments and other securities.

I desired to investigate if I could find the same relationship with Bitcoin price!

Why Bitcoin?

Originally, I desired this project to be focused on the cryptocurrency known spil Ethereum. Unluckily, I wasn’t able to do this because of gegevens!.

I had intended to use Tweepy and the Twitter Surplus API to query tweets relating to #ETH, #Ethereum #Crypto, #hodl. However, Twitter has stringent rate boundaries and I wouldn’t be able to collect enough gegevens te time to finish this project. I only had Four weeks to take a stab at this!

Spil such, I found a public GitHub repository by Miles Luders who is attempting to recreate a Stanford Paper that is working on an identical project!

Fortunately, he had already collected twitter gegevens strakheid Oct 2018 – Feb 2018. It is this twitter gegevens that is the ondergrond of my own project! I downloaded the available Twitter gegevens from his repo on Aug 31st 2018.

Spil such, the main limitation te this project is the gegevens! Ideally, I would have had more te addition to a broader gegevens span. During Oct ’16 – Feb ’17, Bitcoin wasgoed relatively stable and did have dramatic variation’s ter price. I would need to re-run this investigation on a larger scale to obtain more sturdy results.


Ter the Final Notebook on GitHub, I go into the details of what I mean by a nested Time Series Specimen.

Google Glides

You can view the presentation I talent @ Metis regarding this project here.

Be sure to read the presenter notes on slide 8! You can also see the overall results of this project ter the slips.


I truly hope you love this project – I learned a ton extra gegevens science devices!

This concludes my Metis journey – I head back to grad schoolgebouw on Monday.

Fear not, I will proceed blogging anything and everything that I find interesting te grad schoolgebouw.

I’ll be graduating te June 2018 and I’m actively looking for employment – please reach out if you are hiring for gegevens science roles!

Related movie: The Favours Of Your Creator (Powerful Recitation)