Decentralized Efficiency? Arbitrage in Bitcoin Markets

Abstract

Using tick-level bitcoin data from February 2013 through April 2018, we show substantial arbitrage spreads between global bitcoin markets. Spreads follow multiple consistent patterns. Minimum and maximum prices show significant clustering. Spreads increase during the early hours of a day (according to coordinated universal time), when new exchanges enter markets, and following bitcoin heists and hacks. The full year 2017 and the first quarter of 2018 each had exploitable net arbitrage profit opportunities of at least USD380 million that smart money failed to capture. Based on long-term analyses, we also found that bitcoin market inefficiency has increased over time.

Disclosure: The authors report no conflicts of interest.

Editor’s Note

This article was externally reviewed in our double-blind peer-review process. When the article was accepted for publication, the authors thanked the reviewers in their acknowledgments. Daniele Bianchi and Nicola Borri were the reviewers for this article.

Submitted 12 September 2019

Accepted 18 February 2020 by Stephen J. Brown

Acknowledgments

We would like to thank Executive Editor Stephen J. Brown and Co-Editor Daniel Giamouridis. We would also like to thank Johannes Bernius and the participants at the 2019 Annual Meeting of the Southern Finance Association and at the 17th INFINITI Conference on International Finance 2019.

Notes

¹ Ripple is a technology that acts as both a cryptocurrency and a digital payment network for financial transactions. It was first released in 2012 and was co-founded by Chris Larsen and Jed McCaleb (source: Investopedia). Launched in 2015, Ethereum is an open-source, blockchain-based, decentralized software platform used for its own cryptocurrency,  ether (source: Investopedia).

² For examples, see “Famous Bitcoin Heist[s] and Hacks” at https://www.gamblingsites.com/bitcoin/heists-hacks/.

³ Confirmation times as of the writing of this article are hovering around 10 minutes; see https://www.blockchain.com/charts/median-confirmation-time.

⁴ For details, see the CME fee schedule at https://www.cmegroup.com/company/files/cme-fee-schedule-2018-01-01.pdf.

⁵ For details, see Sedgwick (2018).

⁶ Based on the equal probability of a bid–ask price occurring and the expectation that both trades of the arbitrage strategy would be executed, we calculated bid–ask spread slippage as the maximum average bid–ask spread across the exchanges covered in any given year. The reason we used maximum bid–ask spreads is to ensure the robustness of our calculations in light of heterogeneous bid–ask spreads across exchanges (see also Borri and Skakhnov 2018a). The result of our calculations was slippage of 0.38% for 2013, 0.36% for 2014, 0.16% for 2015, 0.14% for 2016, 0.10% for 2017, and 0.08% for 2018. Data were sourced from www.bitcoinity.org and www.tokenspread.com.

⁷ The Cboe has announced that it will not be adding any further bitcoin futures contracts beyond March 2019. The CME, with approximately twice the trading volume of the Cboe, will continue issuance without any policy changes (see Baydakova 2019).