Abstract:Pairs trading and multiple trading strategies are types of market-neutral strategies to use a pair or a combination of stocks and other financial instruments with co-integration or co-movements to generate potential profits, which may not be affected by the direction of the overall market. Commonly used pairs and multiple trading strategies are constructed using the Kalman Filter (KF) to utilize mean reversion in nonstationary but co-integrated asset prices. In this paper, we propose novel resilient pairs trading and multiple trading strategies using the combination of the KF algorithm and the KF innovation volatility interval forecasts using neural networks. The proposed trading strategies are implemented and investigated using the hourly prices of Bitcoin, Ethereum and Bitcoin Cash in the bear market. Those crypto assets are selected because they move in the same direction in the long term and have high trading volumes. The experimental results reveal performance for the proposed trading strategies with the upper and lower trading intervals using KF innovation volatility interval forecasts superior to that of the trading strategies with upper and lower trading bands using KF innovation volatility point forecasts. The performance and robustness of the proposed trading strategies using a proper assumption of transaction costs have also been examined. The strategies using innovation volatility interval forecasts consistently generate higher profits and a more robust number of transactions with or without transaction costs than those using innovation volatility point forecasts.

Currency Forecasting using Multiple Kernel Learning with Financially Motivated Features

An efficient multiple kernel computation method for regression analysis of economic data

Machine learning based forecasting of significant daily returns in foreign exchange markets

A sparse multiple kernel learning method for listed companies financial distress prediction

Financial Forecasting Using Pattern Modeling and Recognition System Based on Kernel Regression

Multiple Kernel Support Vector Regression for Economic Forecasting

Forecasting multi‐frequency intraday exchange rates using deep learning models

Ideal Kernel-Based Multiple Kernel Learning for Spectral-Spatial Classification of Hyperspectral Image.

Boosting Based Multiple Kernel Learning and Transfer Regression for Electricity Load Forecasting

Financial Distress Prediction with Group-Sparsity Multiple Kernel Learning

Learning the kernel matrix via predictive low-rank approximations

Multiple Kernel Learning via Low-Rank Nonnegative Matrix Factorization for Classification of Hyperspectral Imagery

Two-stage multiple kernel learning with multiclass kernel polarization.

A Cryptocurrency Multiple Trading Strategy with Kalman Filter Innovation Volatility Interval Forecasts

Learning with multiple Gaussian distance kernels for time series classification

Representative Multiple Kernel Learning for Classification in Hyperspectral Imagery

Unsupervised Multiple Kernel Learning

Koopman Kernel Regression

Cryptocurrency Trading-Pair Forecasting, Using Machine Learning and Deep Learning Technique

An interpretable deep learning multi-dimensional integration framework for exchange rate forecasting based on deep and shallow feature selection and snapshot ensemble technology

GBP/USD Currency Exchange Rate Time Series Forecasting Using Regularized Least-Squares Regression Method.