Non-Parametric Methods for Analyzing Temporal Dependencies in Financial Time Series.

EasyChair Preprint 15029

Abstract

Financial time series data often exhibit complex temporal dependencies, including autocorrelation, heteroscedasticity, and non-stationarity. Traditional parametric models, such as ARIMA and GARCH, rely on strong assumptions regarding the underlying data distribution and linear relationships, which may not capture the true dynamics of financial markets. Non parametric methods, on the other hand, offer a flexible alternative for analyzing such dependencies without the need for strict assumptions about the data. This abstract provides an overview of non-parametric approaches to analyzing temporal dependencies in financial time series, focusing on methods such as kernel-based estimation, local polynomial regression, and resampling techniques like the bootstrap. These methods are highly adaptable and capable of capturing nonlinear relationships, long memory processes, and structural breaks. Kernel-based approaches, for instance, allow for smoothing time-varying effects, while resampling methods provide robust inference without relying on specific distributional assumptions.

Keyphrases: - **Bandwidth selection**, - **Bootstrap methods**, - **Complex dynamics**, - **Empirical Mode Decomposition (EMD)**, - **Financial time series**, - **Heavy tails**, - **High-frequency trading**, - **K-Nearest Neighbors (KNN)**, - **Kernel Density Estimation (KDE)**, - **Kernel regression**, - **Machine learning in finance**, - **Market dynamics**, - **Non-linear dependencies**, - **Non-linearity**, - **Non-parametric methods**, - **Non-stationarity**, - **Parameter optimization**, - **Resampling techniques**, - **Risk management**, - **Time-Series Forecasting**, - **Volatility clustering**, - **Volatility regimes**

@booklet{EasyChair:15029,
  author    = {Wayzman Kolawole},
  title     = {Non-Parametric Methods for Analyzing Temporal Dependencies in Financial Time Series.},
  howpublished = {EasyChair Preprint 15029},
  year      = {EasyChair, 2024}}