Strategies · Library · Crypto
§18.3Hedges and structures
Sentiment analysis – naïve Bayes Bernoulli
Collect tweets with chosen keywords, mark each word as present or absent, and use a naive Bayes classifier to predict whether Bitcoin goes up or down. Trade on the predicted class.
- Asset class
- Crypto
- Held for
- Days
- Data
- Needs data ARCANE lacks
- Symphony
- Cannot be built as a symphony
The rule
- Signal
- Probability of an up or down class given the words in tweets
- Entry and exit
- Trade each period on the predicted direction
- Sizing
- Single position
How it fails
- Bot-generated noise
- Overfitting
- Sentiment lags price
Data it needs
- tweet text
- BTC prices
No social media text feed.
How ARCANE uses it
Shows how Alpha might weigh public sentiment in a Bitcoin Call. Paper only.
Needs text data and a classifier.
Source: Kakushadze and Serur (2018), 151 Trading Strategies, §18.3, p. 120 of the SSRN version. Described in our words. Hedges and structures is ARCANE’s filing, not the book’s.