Middle East Geopolitical Risk and Monthly Systematic Tail Risk Identification for Time Series Momentum in Chinese Commodity Futures

Abstract

Fluctuations in Middle East Geopolitical Risk (ME-GPR), a regional application of Geopolitical Risk (GPR), rarely remain confined to political news. They permeate commodity futures markets through oil supply expectations, shipping security, global risk-off sentiment, and dollar-based financial conditions. Leveraging 42 continuous contracts of Chinese commodity futures, this paper constructs two Time Series Momentum (TSMOM) strategies with $J=60$ and $J=250$, and derives ME-GPR, Middle East and North Africa Geopolitical Risk (MENA-GPR), change terms, and a positive ‘Shock’ variable from the Caldara-Iacoviello country-level GPR data. Here, ‘Shock’ denotes an upward month-to-month risk-change indicator rather than a separately identified event shock. The analysis does not target extreme single-day tail loss events. Instead, it identifies whether the strategy enters a systematic tail risk state over the following month; the model draws on risk information observable in the previous month to assess the next-month high-risk state. The empirical evaluation is conducted strictly at a monthly frequency. Empirical diagnostics indicate that the monthly Logistic scoring architecture retains non-trivial predictive utility for isolating systematic tail states. The more stable predictive content, however, comes from market-risk benchmark variables, including the CBOE Volatility Index (VIX), the CBOE Crude Oil Volatility Index (OVX), Brent, the U.S. Dollar Index (DXY), commodity market volatility, and strategy-state variables. In the 2022-2025 main sample, the Market-risk benchmark Logistic model records an Area Under the Curve (AUC) of 0.701; in the 2021-2025 robustness sample, the AUC is 0.730. The inclusion of GPR variables does not yield a consistent statistical gain in out-of-sample predictive performance. Monthly dynamic risk control compresses risk exposure and maximum drawdown: in the main sample, the gross maximum drawdown narrows from -15.01% to -9.47%, and the 5 basis points (bp)-cost drawdown narrows from -17.17% to -10.16%. After transaction costs based on actual turnover are included, however, absolute-return gains remain weakly robust. Consequently, the architecture is better interpreted as a risk-budgeting tool than as an alpha-generating strategy.