Pairs Trading Strategy for A and H Shares Based on Kalman-HMM Approach

Pairs trading is a statistical arbitrage strategy that takes advantage of unbalanced financial markets. A common difficulty for quantitative trading participants is the detection of market institutional changes in financial markets. In order to solve this issue, the hidden Markov model (HMM) is applied for status detection. The research objective is to use Kalman filter to predict and the hidden Markov model (HMM) to identify state transitions on the basis of screening transaction pairs with obvious co-integration relationship. This research would prove the profitability of the strategy and the ability to resist risk through the combination of these two methods with real data. The empirical results showed that compared with the traditional cointegration strategy, the holding yield increased from 1.6% to 16.2% and the maximum pullback reduced to 0.02%. Further research is required to improve trading rules.

Unsupervised Learning for Pairs Trading in the National Stock Exchange of India

Pairs Trading is a widely known and used market-neutral trading strategy that utilizes the concept of statistical arbitrage. It is based on the idea of mean-reverting time series and relies on the spread between two assets to demonstrate that property to buy an asset at a relatively undervalued price and an asset at a relatively overvalued price. This allows investors to manage risk if the market moves strongly in only one direction by making money on one side of the bet. The main challenge of pairs trading is selecting pairs that have an actual underlying relationship and their spread has real statistical significance. In this paper, we present the use of machine learning, specifically unsupervised clustering to construct our search space for pair selection and compare it against a traditional way of selecting pairs. We see that not only are we able to pick out more profitable pairs, these pairs are also less volatile and have less exposure to the market.

Machine Learning in Futures Markets

In this paper, we demonstrate how a well-established machine learning-based statistical arbitrage strategy can be successfully transferred from equity to futures markets. First, we preprocess futures time series comprised of front months to render them suitable for our returns-based trading framework and compile a data set comprised of 60 futures covering nearly 10 trading years. Next, we train several machine learning models to predict whether the h-day-ahead return of each future out- or underperforms the corresponding cross-sectional median return. Finally, we enter long/short positions for the top/flop-k futures for a duration of h days and assess the financial performance of the resulting portfolio in an out-of-sample testing period. Thereby, we find the machine learning models to yield statistically significant out-of-sample break-even transaction costs of 6.3 bp—a clear challenge to the semi-strong form of market efficiency. Finally, we discuss sources of profitability and the robustness of our findings.

On cointegration and cryptocurrency dynamics

This paper aims to model the joint dynamics of cryptocurrencies in a nonstationary setting. In particular, we analyze the role of cointegration relationships within a large system of cryptocurrencies in a vector error correction model (VECM) framework. To enable analysis in a dynamic setting, we propose the COINtensity VECM, a nonlinear VECM specification accounting for a varying systemwide cointegration exposure. Our results show that cryptocurrencies are indeed cointegrated with a cointegration rank of four. We also find that all currencies are affected by these long term equilibrium relations. The nonlinearity in the error adjustment turned out to be stronger during the height of the cryptocurrency bubble. A simple statistical arbitrage trading strategy is proposed showing a great in-sample performance, whereas an out-of-sample analysis gives reason to treat the strategy with caution.