UNIVARIATE AND MULTIVARIATE MODELLING AND FORECASTING OF INTERNATIONAL EQUITY INDEX RETURNS

This thesis investigates the validity and adaptability of the Efficient Market Hypothesis (EMH) across major international equity markets, focusing on the S&P 500 (United States), FTSE MIB (Italy), and Nikkei 225 (Japan) over the period 2020–2025. The research evaluates both the internal efficiency of individual markets and the cross-market transmission of information in a highly integrated global environment by combining univariate and multivariate econometric models. Univariate ARFIMA–GARCH and GJR-GARCH models are first estimated to capture the conditional mean and volatility dynamics of each market, accounting for persistence, asymmetry, and heavy-tailed distributions. A Vector Autoregressive (VAR) model is then employed to explore interdependence among indices and potential cross-market spillovers. Outof- sample forecasts, generated through a rolling-window procedure (2024–2025), are assessed both statistically – via RMSE and MAE – and economically through simulated trading strategies based on model-implied signals. The empirical results indicate that while forecast accuracy remains limited during tranquil periods, both GARCH-type and VAR models exhibit markedly improved directional performance in high-volatility regimes, where informational shocks are more persistent. Trading strategies built on these forecasts often outperform the benchmark in selected markets, particularly for the FTSE MIB and Nikkei 225, confirming the presence of regime-dependent inefficiencies. Moreover, the introduction of a threshold-based strategy, which filters out weak signals, enhances profitability and stabilises performance by reducing exposure to market noise. Overall, the findings suggest that market efficiency is not absolute but adaptive and contextdependent, consistent with Lo’s (2004) Adaptive Market Hypothesis. Econometric models can capture transient inefficiencies during turbulent phases, offering valuable insights for both academic and applied financial forecasting.