Sentiment has long been an important indicator in the financial markets and with a market as potentially volatile as Bitcoin, watching sentiment can prove even more rewarding.
There are a number of ways to gauge sentiment , with some systems now going as far as to scan Twitter and other social media for that extra edge, but a good, easily available method is to look at Google Trends.
Google Trends provides historical data on any search term you care to think of and so with this in mind we decided to see if we could find a correlation between the number of searches for various Cryptocurrency related search terms, and the price of Bitcoin.
We began by collecting data for a large number of different Bitcoin and crypto related search terms, along with the USD price of Bitcoin, all going back to January 2017. Amongst them were terms like ‘Bitcoin’, ‘Ethereum’, ‘Crypto Trading’, ‘Coinbase’, the list goes on.
We then took the search terms that had the highest correlation to the price of Bitcoin and ran them through an SKLearn Linear Regression model to see if we could get a meaningful fit, and the results are rather interesting.
Here we see the top 5 search terms graphed with the USD price of Bitcoin, showing a good correlation, and possibly even an element of predictive power.
Having fed this data through our model we were then able to draw comparisons between the model’s predicted price, versus the actual price of Bitcoin and displayed them in the scatter plot above.
We also graphed the difference between the two (the residuals) in the chart above, this shows the actual price of Bitcoin versus the predicted price, and as you can see Bitcoin is currently just over $2,000 under where our model predicts it should be, based on the recent Google Trends data.
And finally, this chart shows a simple line graph of both, and one can now clearly see that with the model predicting a price of $8,023 (given the recent Google Trends data), BTC at $5,900 still potentially looks cheap. GL.
Join us at BigCoinData.io or follow us on Medium for more.