Forecasting prices of alternative assets

Recently, Cryptocurrencies have been the new alternative asset choice for retail investors around the globe. Due to Cryptocurrencies’ characteristic high volatility, Central Banks do not acknowledge them as a store of value and therefore, not as money. While some scholars argue that cryptocurrencies provide a sort of a sink for excess funds, thus keeping inflationary pressures at bay, their high volatility has caused wealth destruction effects at the same time. Cryptos are regarded as unforecastable and the best known strategy regarding crypto so far has been the buy-and-hold strategy which has provided investors returns in excess of 100%. This paper aims to establish a causal relationship between pairs of cryptocurrencies and utilise the information to forecast returns with the help of a novel OLS-LSTM model proposed within. The practicality of this approach is shown by implementing a simple trading strategy based on the signals generated by the model. We further aim to execute a pairs trading algorithm between different pairs of cryptocurrencies