How to Become Rich Using Quantitative Trading?

Abstract

Bitcoin has become a very common investment product, also known as 'digital gold' because of its global discussion and rising prices. This paper develops a model that can only rely on these data to predict future price fluctuations and plan trading decisions. The trading decision planning model consists of three sub-models that are mutually undertaken. For model 1, in order to judge the price trend reliably, this paper uses the Long Short-Term Memory (LSTM) neural network, which is sensitive to the time series, to learn all the time series price data up to the trading day, and to predict the future price data of them respectively, which is provided to another sub-models for subsequent work. For Model 2, after obtaining the follow-up price data, all the technical indicators needed are calculated in this model, and then the moving average convergence divergence (MACD) rule is used to determine whether the transaction is carried out for the first time, and then the predicted price trend is used for the second judgment. For model 3, the model begins to make decision- making planning. This model mainly solves the bi-objective programming problem with the goal of maximizing returns and minimizing risks. In the range of controllable risks, the model will give the optimal solution of maximizing returns. Finally, using the model, the sensitivity of the model to transaction cost is studied by changing the transaction commission, and it is found that the final investment income is more sensitive to the transaction commission of bitcoin.