Asset Pricing of Cryptocurrencies and Momentum Based Patterns

This thesis develops a three-factor asset pricing model for cryptocurrencies by using a market factor, a size factor and a factor related to the transaction volume relative to an asset's market capitalisation. This model explains on average about 35% of the variance of weekly returns. Additionally, significant momentum-based returns were revealed by using several formation, holding and weighting periods and different ways of constructing the portfolios. Those returns did not only have abnormal returns up to 74.11 basis points per day by using the aforementioned asset pricing model for risk adjusting, but also have a remarkable Sharpe ratio of more than 3 by using their raw returns. Not only discrete momentum premia were calculated, but also patterns in those premia investigated. Evidence for superior returns of buying past 'winners', a so-called Long-Only approach, in comparison to buying past 'winners' and short-selling past 'losers' was found. Moreover, the analysis suggests investors to utilise a formation period of three weeks and a holding period of two weeks, while ignoring a waiting period. The used data set is covering 15 cryptocurrencies chosen by their market capitalisation and their data availability for a period from April 2016 to July 2017.