Dr. Swetadri Samadder's research is centered on the quantitative analysis of financial time series, with a strong focus on uncovering complex structures and behaviors within global and Indian stock markets. His work integrates statistical, econometric, and nonlinear methods to investigate market dynamics, memory properties, and inter-market relationships.
Time Series Analysis
Investigating persistence, anti-persistence, and randomness in stock market indices using finite variance scaling methods.
Fractal and Multifractal Analysis
Application of Higuchi, Katz, and Multifractal Detrended Fluctuation Analysis (MFDFA) to characterize self-similar and multifractal behavior in stock prices.
Market Memory and Autocorrelation Structures
Identification of autoregressive (AR), moving average (MA), and mixed (ARMA) influences across global stock indices.
Causality and Connectedness
Applying linear and nonlinear Granger Causality tests to examine influence patterns among G7, E7, BRICS, and Asian markets. Analysis of temporal connectedness using CPR and Pearson correlation over rolling windows.
Cointegration and Portfolio Diversification
Studying short- and long-term relationships between Indian and developed market indices using Johansen cointegration and VECM-based causality. Emphasis on portfolio diversification potential.
Spectral and Periodic Analysis
Use of Ferraz-Mello DCDFT, Empirical Mode Decomposition (EMD), and Hilbert-Huang Transform to uncover cycles, pseudo-periods, and spectral properties in financial time series.
Nonlinearity and Determinism in Markets
Measured using Delay Vector Variance (DVV), 0–1 chaos test, Largest Lyapunov Exponent, and Recurrence Plot analysis to establish non-chaotic yet nonlinear deterministic behavior in stock market data.
Statement of Research:
The focus of my research has been the comprehensive quantitative analysis of global and Indian stock markets using advanced statistical, econometric, and nonlinear dynamical techniques. The work encompasses a broad spectrum of financial time series modeling, scaling analysis, memory characterization, fractality, market integration, and causality.
A scaling analysis was carried out on the daily closing price data of India’s key stock indices, SENSEX and NIFTY, along with major global indices, including those from G7 countries (USA, Canada, France, Germany, Italy, Japan, UK), developed markets (Australia, Spain, Netherlands), emerging African economies (Kenya, South Africa), advanced Asian economies (Hong Kong, Singapore, South Korea, Taiwan), and E7 nations excluding India (China, Brazil, Mexico, Russia, Indonesia, Turkey). Using the finite variance scaling method, markets were classified based on memory behavior—persistent (long-term memory), anti-persistent (short-term memory), and random. Notably, markets like India, USA, Germany, Japan, and China exhibited anti-persistence, while a majority of the others showed persistent behavior. Russia’s stock market displayed near-random features.
To identify underlying generating processes, autocorrelation coefficients and lag structures were examined to distinguish whether markets followed autoregressive (AR), moving average (MA), or ARMA processes. It was found that markets like Germany, UK, Australia, and several Asian and Latin American countries aligned with MA processes, while others like India, USA, Canada, and Japan followed mixed models.
To understand the connectedness and dynamic co-movements among these markets, window-based CPR (150, 250, and 350-point windows) analyses were conducted and results were benchmarked against Pearson’s correlation. The findings indicate that markets exhibit alternating phases of high and low connectivity, undermining the assumption of monotonic globalization.
Granger causality tests, both parametric and non-parametric, were used extensively. The causal influences of global indices on the Indian market were explored, particularly SENSEX and NIFTY with DOW JONES and selected major Asian markets. Results showed a dominant influence of DOW JONES, especially during periods of strong co-movement. A broader causality analysis among G7, E7, and BRICS countries revealed that G7 members have stronger internal causal structures, with countries like the USA, UK, and France acting as highly endogenous hubs. In contrast, BRICS and E7 members like China, Russia, and Indonesia were more exogenous. Nonlinear Granger causality further highlighted India’s centrality in BRICS and E7 as an endogenous influencer, while countries like Brazil and Russia remained exogenous.
Stock market integration studies were performed between India and major developed economies (Australia, Canada, France, Germany, UK, USA). Using Johansen cointegration and VECM-based Granger causality, the results demonstrated long-term cointegration, particularly with the USA, and short-term causality with Germany and France. The Indian market exhibited low correlation with France, suggesting potential for international portfolio diversification, especially in the short run.
In the context of Asian market linkages, cointegration and causality tests were applied to markets in Sri Lanka, Bangladesh, Hong Kong, South Korea, Japan, and China. While short-term associations were low—favoring diversification—long-term integration and both unidirectional and bidirectional causality were detected, with Japan being the dominant influencer.
To delve into nonlinear dynamics and fractal behavior, the study investigated the self-similarity and multifractal structures of Indian (BSE Sensex and NSE) and American (DOW JONES) markets. The Higuchi and Katz methods confirmed monofractal characteristics, while multifractal detrended fluctuation analysis (MFDFA) revealed heterogeneous scaling behaviors. NSE was found to possess richer multifractal properties than BSE. Singularity spectra and generalized Hurst exponents further substantiated these differences.
Given the fractal nature of markets, periodicities in the indices were analyzed using Date-Compensated Discrete Fourier Transform (DCDFT), Empirical Mode Decomposition (EMD), and Hilbert-Huang Transform. Several dominant and pseudo-cycles were discovered, including long-range periodicities of over 100 months for indices like DOW JONES and S&P500. The similarity in spectral behavior between DOW JONES and Indian indices was particularly noteworthy.
The nonlinear and deterministic nature of these time series was validated using the Delay Vector Variance (DVV) method. Further, the possibility of chaotic dynamics was tested using 0–1 chaos tests, Lyapunov exponents, and recurrence plots. The results established that while these markets are nonlinear and deterministic, they do not exhibit chaotic behavior.