We have two bitcoin and cryptocurrency enthusiasts on our team, and only of us made money during its peak earlier this month. Cryptocurrencies are just too volatile, and its value depends too much on how the public feels about it. How people think and talk about a cryptocurrency affects it price to a large extent, unlike stocks which also have the support of the market, shareholders and the company itself.
What it does
Our website scrapes for thousands of social media posts and news articles to get information about the required cryptocurrency. We then analyse it using NLP and ML and determine whether the price is likely to go up or down in the very near future. We also display the current price graphs, social media and news trends (if they are positive, neutral or negative) and the popularity ranking of the selected currency on social platforms.
How I built it
The website is mostly built using node.js and bootstrap. We use chart.js for a lot of our web illustrations, as well as python for web scraping, performing sentimental analysis and text processing. NLKT and Google Cloud Natural Language API were especially useful with this. We also stored our database on firebase. Google Cloud: We used firebase to efficiently store and manage our database, and Google Cloud Natural Language API to perform sentimental analysis on hundreds of social media posts efficiently.
Challenges I ran into
It was especially hard to create, store and process the large datasets we made consisting of social media posts and news articles. Even though we only needed data from the past few weeks, it was a lot since so many people post online. Getting relevant data, free of spam and repeated posts, and actually getting useful information out of it was hard.
Accomplishments that I'm proud of
We are really proud that we were able to connect multiple streams of data, analyse them and display all relevant information. It was amazing to see when our results matched the past peaks and crashes in bitcoin price.
What I learned
We learned how to scrape relevant data from the web, clean it and perform sentimental analysis on it to make predictions about future prices. Most of this was new to our team members and we definitely learned a lot.
What's next for
We hope to further increase the functionality of our website. We want users to have an option to give the website permission to automatically buy and sell cryptocurrencies when it determines it is the best time to do so.
We bought the domain name get-crypto-insights.online for the best domain name challenge since it is relevant to our project. If I found a website of this name on the internet, I would definitely visit it to improve my cryptocurrency trading experience.
We are Discord team #1, with @uditk, @soulkks, @kilobigeye and @rakshaa