Research at the intersection of AI, sentiment analysis, quantitative trading and cryptocurrencies.

Research
August 27, 2019

Backtesting a Sentiment Analysis Strategy for Bitcoin – Part 2

Abstract Optimizing parameters from our previous strategy improved the simulated return and drawdown Adding trading fees made the strategy more realistic while finding optimal sentiment combinations and window sizes increased...
Research
August 22, 2019

Traces of manipulation: How the mind of the market predicts Tether and vice versa

Abstract Tether's sentiment shows forward correlations with Tether market cap and price Sentiment could be manipulated and/or exploited by insiders for creating arbitrage opportunities with Tether (USDT) If Tether market...
Research
July 30, 2019

Backtesting a Sentiment Analysis Strategy for Bitcoin – Part 1

Abstract A daily long/short strategy based on Augmento sentiments and backtested on Bitmex XBTUSD generated a positive return between 2017 and 2019 The signal which generated positive returns in a...
Research
June 19, 2019

Indicators based on sentiment analysis reveal all-time highs of Bitcoin FOMO

Abstract FOMO sentiment towards Bitcoin (BTC) recently reached all-time highs on Twitter and Bitcointalk but not on Reddit FOMO sentiment on Bitcointalk showed extraordinary price leading characteristics in recent months...
Research
May 10, 2019

Is the crypto bear market over? The sentiment perspective

Abstract Sentiment of crypto social media lead and accompanied the hyperbolic price run in 2017 Crypto sentiments are likely to have not been calm enough to repeat a 2017 scenario...
Research
May 07, 2019

The Augmento Sentiment Polarity Indicator for Cryptocurrencies

Abstract We show that building synthetic indicators based on opposing sentiments can have predictive capabilities in cryptocurrency markets  We create indicators based on three pairs of opposing sentiments: Positive/Negative, Bullish/Bearish...
Literature review
May 07, 2019

Literature review: Risks and Returns of Cryptocurrency by Liu and Tsyvinski (2018)

Link to paper Summarized findings The researchers analyzed the relationship between cryptocurrencies and stocks, commodities, currencies, macro factors and sentiment. Next to momentum sentiment is the most promising factor with...
Research
May 02, 2019

Perils of crypto sentiment seasonality

Abstract Based on 29 months worth of data we analyzed on which weekdays Twitter users express more positive or negative sentiments regarding Bitcoin  Simplistic seasonality analysis of sentiment in Bitcoin...