Nikkei 225
The Nikkei 225 (Nikkei Stock Average) is the leading Japanese equity index — a price-weighted index of 225 large companies, so higher-priced stocks dominate its moves. It is deeply liquid: the Osaka Exchange (JPX) lists large, mini and micro futures (the large contract is Nikkei 225 × JPY 1,000, tick JPY 10) with day and night sessions spanning over 20 hours, and the index also trades as yen- and USD-denominated futures on SGX and CME, linked through the long-standing CME-SGX Mutual Offset System. Costs are low, so — as for the S&P 500 and DAX — the Nikkei sits well inside the cost regime where regime-switching timing strategies remain executable.
The Nikkei is the case in this vault where regime filtering looks most useful, and the reason is economic, not statistical luck. After Japan’s late-1980s asset bubble burst, the Nikkei suffered a prolonged, persistent post-1990 bear market — exactly the deep, multi-year drawdown a two-state Markov Regime-Switching Model is designed to step out of. The empirical payoff is visible across both anchor studies. Bulla et al. 2010 report the Nikkei’s largest annualised excess return of their five indices — about 201.6bp over buy-and-hold after 10bp costs — far above the trivial 18.5bp on the S&P 500. Shu Yu and Mulvey 2024 find the regime strategies materially lifted the Nikkei’s very low buy-and-hold Sharpe (0.12): the HMM-guided 0/1 strategy raised return to 2.5% from buy-and-hold’s 0.8% while cutting the maximum drawdown from a punishing -79.1% to -48.6% (the jump-model variant reached 4.7%).
This makes the Nikkei a precise illustration of where and why HMM regime filtering helps — and of its limits. Regime detection delivers its largest gains in markets with deep, persistent bear phases, because sidestepping a multi-year decline both cuts drawdown and lifts return; in a market that mostly trends upward, the same filter mostly costs turnover. But even here the gain is best read as drawdown avoidance plus the mechanical return that comes from not riding a long bear market down — not as evidence of a repeatable directional edge. The Nikkei is the strongest single data point for regime filtering precisely because Japan’s lost decades were an unusually favourable environment for it; that is a reason to be cautious about generalising the result, not a demonstration of standalone alpha.
Shu Yu and Mulvey 2024 [trades_market] Nikkei 225 Bulla et al. 2010 [reports_profitability] Regime-Based Asset Allocation Nikkei 225 [supports] Volatility Regime
Connections
- Hidden Markov Model Regime Detection — trades_market, source: https://arxiv.org/html/2402.05272v3
- Markov Regime-Switching Model — detects_regime, source: https://mpra.ub.uni-muenchen.de/21154/1/MPRA_paper_21154.pdf
- Bulla et al. 2010 — reports_profitability (largest excess return, ~201.6bp), source: https://mpra.ub.uni-muenchen.de/21154/1/MPRA_paper_21154.pdf
- Shu Yu and Mulvey 2024 — reports_profitability (drawdown cut -79.1% to -48.6%), source: https://doi.org/10.1057/s41260-024-00376-x
- Regime-Based Asset Allocation — tests_strategy, source: https://doi.org/10.1057/s41260-024-00376-x
- Volatility Regime — detects_regime, source: https://mpra.ub.uni-muenchen.de/21154/1/MPRA_paper_21154.pdf