Multiple Financial Time Series Research Articles

Improvement of LSTM-Based Forecasting with NARX Model through Use of an Evolutionary Algorithm

The reported work aims to improve the performance of LSTM-based (Long Short-Term Memory) forecasting algorithms in cases of NARX (Nonlinear Autoregressive with eXogenous input) models by using evolutionary search. The proposed approach, ES-LSTM, combines a two-membered ES local search procedure (2MES) with an ADAM optimizer to train more accurate LSTMs. The accuracy is measured from both error and trend prediction points of view. The method first computes the learnable parameters of an LSTM, using a subset of the training data, and applies a modified version of 2MES optimization to tune them. In the second stage, all available training data are used to update the LSTM’s weight parameters. The performance of the resulting algorithm is assessed versus the accuracy of a standard trained LSTM in the case of multiple financial time series. The tests are conducted on both training and test data, respectively. The experimental results show a significant improvement in the forecasting of the direction of change without damaging the error measurements. All quality measures are better than in the case of the standard algorithm, while error measures are insignificantly higher or, in some cases, even better. Together with theoretical consideration, this proves that the new method outperforms the standard one.

Time-Varying Mixture Copula Models with Copula Selection

Modeling the joint tails of multiple financial time series has many important implications for risk management. Classical models for dependence often encounter a lack of fit in the joint tails, calling for additional flexibility. This paper introduces a new semiparametric time-varying mixture copula model, in which both weights and dependence parameters are deterministic and unspecified functions of time. We propose penalized time-varying mixture copula models with group smoothly clipped absolute deviation penalty functions to do the estimation and copula selection simultaneously. Monte Carlo simulation results suggest that the shrinkage estimation procedure performs well in selecting and estimating both constant and time-varying mixture copula models. Using the proposed model and method, we analyze the evolution of the dependence among four international stock markets, and find substantial changes in the levels and patterns of the dependence, in particular around crisis periods.

Entropic Dynamic Time Warping Kernels for Co-Evolving Financial Time Series Analysis.

Network representations are powerful tools to modeling the dynamic time-varying financial complex systems consisting of multiple co-evolving financial time series, e.g., stock prices. In this work, we develop a novel framework to compute the kernel-based similarity measure between dynamic time-varying financial networks. Specifically, we explore whether the proposed kernel can be employed to understand the structural evolution of the financial networks with time associated with standard kernel machines. For a set of time-varying financial networks with each vertex representing the individual time series of a different stock and each edge between a pair of time series representing the absolute value of their Pearson correlation, our start point is to compute the commute time (CT) matrix associated with the weighted adjacency matrix of the network structures, where each element of the matrix can be seen as the enhanced correlation value between pairwise stocks. For each network, we show how the CT matrix allows us to identify a reliable set of dominant correlated time series as well as an associated dominant probability distribution of the stock belonging to this set. Furthermore, we represent each original network as a discrete dominant Shannon entropy time series computed from the dominant probability distribution. With the dominant entropy time series for each pair of financial networks to hand, we develop an entropic dynamic time warping kernels through the classical dynamic time warping framework, for analyzing the financial time-varying networks. We show that the proposed kernel bridges the gap between graph kernels and the classical dynamic time warping framework for multiple financial time series analysis. Experiments on time-varying networks extracted through New York Stock Exchange (NYSE) database demonstrate that the effectiveness of the proposed method.

Time-varying copula models for financial time series

AbstractWe perform an analysis of the potential time inhomogeneity in the dependence between multiple financial time series. To this end, we use the framework of copula theory and tackle the question of whether dependencies in such a case can be assumed constant throughout time or rather have to be modeled in a time-inhomogeneous way. We focus on parametric copula models and suitable inference techniques in the context of a special copula-based multivariate time series model. A recent result due to Chanet al.(2009) is used to derive the joint limiting distribution of local maximum-likelihood estimators on overlapping samples. By restricting the overlap to be fixed, we establish the limiting law of the maximum of the estimator series. Based on the limiting distributions, we develop statistical homogeneity tests, and investigate their local power properties. A Monte Carlo simulation study demonstrates that bootstrapped variance estimates are needed in finite samples. Empirical analyses on real-world financial data finally confirm that time-varying parameters are an exception rather than the rule.

Nonparametric Time-Varying Mixture Copula Models

Time-varying copula models have been shown to offer additional flexibility among multiple financial time series. In this paper we propose a new nonparametric time-varying mixture copula model, in which both weights and dependence parameters are deterministic functions of time. Asymptotic properties of the local constant estimators are established, and a revised expectation maximization (EM) algorithm is employed to reduce the computational burden. We use Monte Carlo simulation to examine the finite sample performance of the proposed estimators. In a financial example, we show that our proposed model outperforms the benchmark copula models.

Characteristics of the co-fluctuation matrix transmission network based on financial multi-time series

AbstractThe co-fluctuation of two time series has often been studied by analysing the correlation coefficient over a selected period. However, in both domestic and global financial markets, there are more than two active time series that fluctuate constantly as a result of various factors, including geographic locations, information communications and so on. In addition to correlation relationships over longer periods, daily co-fluctuation relationships and their transmission features are also important, since they can present the co-movement patterns of multi-time series in detail. To capture and analyse the features of the daily co-movements of multiple financial time series and their transmission characteristics, we propose a new term—“the co-fluctuation relation matrix”—which can reveal the co-fluctuation relationships of multi-time series directly. Here, based on complex network theory, we construct a multi-time series co-fluctuation relation matrix transmission network for financial markets by taking each matrix as a node and the succeeding time sequence as an edge. To reveal the process more clearly, we utilize daily time series data for four well-known stock indices—the NASDAQ Composite (COMP), the S&P 500 Index, the Dow Jones Industrial Average and the Russell 1000 Index—from 22 January 2003 to 21 January 2015, to examine the concentration of the transmission networks and the roles of each matrix—in addition to the transmission relationships between the matrices—based on a variety of coefficients. We then compare our results with the statistical features of the stock indices and find that there are not many discernible patterns of co-fluctuation matrices over the 12-year period, and few of these play important roles in the transmission network. However, the conductibility of the few dominant nodes is different and reveals certain novel features that cannot be obtained by traditional statistical analysis, such as the “all positive co-fluctuation matrix”, which is the most important node, although one stock index has negative correlation with the other three. This research therefore provides a novel method for analysing the co-movement behaviour of multiple financial time series, which can help researchers obtain the roles and relations of co-fluctuation patterns both over short and long terms. The findings also provide an important basis for further investigations into financial market simulations and the fluctuation of multiple financial time series.

Multiple Time Series Ising Model for Financial Market Simulations

In this paper we propose an Ising model which simulates multiple financial time series. Our model introduces the interaction which couples to spins of other systems. Simulations from our model show that time series exhibit the volatility clustering that is often observed in the real financial markets. Furthermore we also find non-zero cross correlations between the volatilities from our model. Thus our model can simulate stock markets where volatilities of stocks are mutually correlated.

Analysis of non-stationary dynamics in the financial system

Novel data-driven analyses, appropriate for detecting economic instability in non-stationary time series, are developed using functional principal component analysis (fPCA) and Synchrosqueezing. fPCA is applied in a new way, aggregating multiple financial time series to identify periods of macroeconomic instability. Synchrosqueezing, a technique which generates a time-series’ time-dependent spectral decomposition, is modified to develop a new quantitative measure of local dynamical changes and structural breaks. The merit of this integrated technique is demonstrated by analyzing financial data from 1986 to 2012 that includes equity indices, securities and commodities, and foreign exchange. Both procedures successfully detect key historic periods of instability. Moreover, the results reveal distinctions between periods of long-term gradual change in addition to structural breaks. These tools offer new insights into the analysis of financial instability.

Q-Gaussian distributions and multiplicative stochastic processes for analysis of multiple financial time series

This study considers q-Gaussian distributions and stochastic differential equations with both multiplicative and additive noises. In the M-dimensional case a q-Gaussian distribution can be theoretically derived as a stationary probability distribution of the multiplicative stochastic differential equation with both mutually independent multiplicative and additive noises. By using the proposed stochastic differential equation a method to evaluate a default probability under a given risk buffer is proposed.

Simple Tests for Models of Dependence Between Multiple Financial Time Series, with Applications to U.S. Equity Returns and Exchange Rates

Evidence that asset returns are more highly correlated during volatile markets and during market downturns (see Longin and Solnik, 2001, and Ang and Chen, 2002) has lead some researchers to propose alternative models of dependence. In this paper we develop two simple goodness-of-fit tests for such models. We use these tests to determine whether the multivariate Normal or the Student’s t copula models are compatible with U.S. equity return and exchange rate data. Both tests are robust to specifications of marginal distributions, and are based on the multivariate probability integral transform and kernel density estimation. The first test is consistent but requires the estimation of a multivariate density function and is recommended for testing the dependence structure between a small number of assets. The second test may not be consistent against all alternatives but it requires kernel estimation of only a univariate density function, and hence is useful for testing the dependence structure between a large number of assets. We justify our tests for both observable multivariate strictly stationary time series and for standardized innovations of GARCH models. A simulation study demonstrates the efficacy of both tests. When applied to equity return data and exchange rate return data, we find strong evidence against the normal copula, but little evidence against the more flexible Student’s t copula.