A sentiment-based model for the BitCoin: theory, estimation and option pricing

TL;DR

This paper develops a bivariate continuous-time model for Bitcoin prices incorporating sentiment indicators, proving its arbitrage-free nature, deriving a closed-form option pricing formula, and empirically fitting it to market data using Google search and volume proxies.

Contribution

It introduces a novel sentiment-based bivariate model for Bitcoin, providing a closed-form option pricing formula and empirical validation with market data.

Findings

Model is arbitrage-free under mild conditions.

Closed-form formula for European option prices on Bitcoin.

Model fits market data using sentiment proxies like Google searches.

Abstract

In recent literature it is claimed that BitCoin price behaves more likely to a volatile stock asset than a currency and that changes in its price are influenced by sentiment about the BitCoin system itself; in Kristoufek [10] the author analyses transaction based as well as popularity based potential drivers of the BitCoin price finding positive evidence. Here, we endorse this finding and consider a bivariate model in continuous time to describe the price dynamics of one BitCoin as well as a second factor, affecting the price itself, which represents a sentiment indicator. We prove that the suggested model is arbitrage-free under a mild condition and, based on risk-neutral evaluation, we obtain a closed formula to approximate the price of European style derivatives on the BitCoin. By applying the same approximation technique to the joint likelihood of a discrete sample of the bivariate process, we are also able to fit the model to market data. This is done by using both the Volume and the number of Google searches as possible proxies for the sentiment factor. Further, the performance of the pricing formula is assessed on a sample of market option prices obtained by the website deribit.com.