The Profitability of Standard Trading Strategies in Cryptocurrency Markets
|Author:||Bc. Miroslav Duda|
|Year:||2019 - summer|
|Leaders:|| doc. PhDr. Ladislav Krištoufek Ph.D.
|Work type:|| Bachelors
|Awards and prizes:|
|Abstract:||The thesis attempts to determine how strategies used for forecasting and trading on foreign exchange and stock markets perform when applied to cryptocurrency markets. The approaches explored are ARIMA, VAR, MA Crossover, and
Granger Causality using gold prices and S&P 500. The currencies traded are
Bitcoin, Ethereum, Binance Coin, and Basic Attention Token. The models are
trained on logarithmically transformed and differenced time series composed
of the currencies’ daily and hourly closing prices. Applying these strategies
mostly leads to ambiguous results, with MA Crossover generally performing
better than VAR, which in turn performs better than ARIMA. However, every
strategy was moderately successful for at least one of the currencies examined.
Trading on the hourly dataset was negatively influenced by sudden price jumps.
ARIMA and VAR perform better in the inter-bubble periods. No significant
Granger causality was found.