Crypto Nostradamus

• 

  Recent tweet analysis

• 

  First Dashboard

• 

  Sentiment trend vs Market trend based on historical tweets

• 

  Market price Prediction

Inspiration

Recently, retail investors have been letting social media activity effect their decisions to short-sell stocks. Causing volatility in crypto prices based on public consensus.

What it does

Our project predicts the market price change of crypto currencies based on sentiment scores of tweets

How we built it

• Collaborated using Google Colab
• Used Twitter API to extract past 24 hour cryptocurrency tweets
• Performed sentiment analysis on the currencies
• Used Coin Gecko API to extract USD 24 hour % change of cryptocurrencies
• Created a Dashboard of real time twitter sentiment score and 24 hr price change
• Analyze historical datasets of Bitcoin and built a prediction model with scikitlearn

Challenges we ran into

We had to deal with unicode characters and emojis in the data. We restricted the number of tweets to 100 per cryptocurrency since we had a rate limit. We had a hard time sourcing the right historical data.

Accomplishments that we're proud of

We were able to succesfully generate a summary view of how tweets impact cryptocurrencies with evidence in the form of graphs and charts using sentiment analysis and we built a prediction model which predicts price change from sentiment score

What we learned

We learnt how to implement sentiment analysis on textual data

What's next for Crypto Nostradamus

• filter for verified user tweets
• create a list of most influential twitter users like Elon Musk whose tweets are known to sway the crypto market
• train a better model using better data
• do a weighted averge of sentiment scores based on number of retweets

Built With

• colab
• natural-language-processing
• python
• twitter