I’m now writing on medium, sharing my knowledge with potential future practitioners interested in the field of data science.

Sentiment analysis leverages natural language processing algorithms to understand whether a string of text expresses a positive, neutral or negative opinion about a specific topic.

This information is defined in algorithmic trading as alternative data, non-financial records that can have an influence on prices, and might give you an edge over the market.

Over the course of a month and a half, the database recorded 13,000,000 tweets, averaging roughly 300,000 a day.

Most of the research I read about the topic made the mistake of downloading historical tweets, which comes at a high cost in terms of statistical accuracy. Twitter heavily limits access to historical tweets, allowing researchers to download only a tiny fraction of them.

Streaming, on the other hand, includes every tweet published on the platform from the moment you start the stream until you stop it. Once you state the search keyword you’re interested in, you’re ready to go.

As the sentiment increases, Bitcoin prices tend to increase and viceversa.

It’s important to remind that correlation does not mean causation. We cannot state that price is directly related to sentiment. If that were the case, users would be able to manipulate the market, which is not true.