Analyse de marché
The Monte Carlo analysis of JPY carry trade unwind uses a Merton Jump-Diffusion model to simulate USD/JPY dynamics over a 90-day horizon, capturing both continuous exchange rate movements and sudden discontinuities from forced position closures. The model incorporates calibrated parameters including a spot rate of 158.5, stressed volatility ranging from 9% to 18%, and jump intensities based on the August 2024 episode (12% yen strengthening) and July 2026 events.
Three scenarios were simulated with 10,000 daily trajectories. The baseline scenario assumes limited intervention with a -1% drift, 9% volatility, and no jumps, resulting in a stable yen near current levels (mean 158.1, median 158.0) and only a 12% probability of USD/JPY falling below 150. The mild unwind scenario (25-30% subjective probability) features -6% drift, 12% volatility, and two annual jumps averaging -6%, producing significant yen strengthening (mean 151.9, median 152.3) with a 41.8% chance of breaking 150 and maximum drawdown around 38%. The severe unwind scenario (12-18% tail probability) replicates and amplifies the 2024 episode with -11% drift, 18% volatility, and four annual jumps of -12%, driving USD/JPY below 140 with 52% probability and causing cascading liquidations, with maximum drawdown reaching approximately 63%. Risk metrics for the severe scenario show extreme outcomes: 5% expected shortfall of 115-120, and 95% VaR of 124.
The analysis indicates that even moderate unwind now has significant probability of pushing USD/JPY below 150 within 90 days, with nonlinear second-round effects on global liquidity. Operational implications include sizing carry positions based on mild/severe scenarios rather than yield alone, maintaining put options or risk-reversals given current negative skew (1M risk-reversal at -2.2), and monitoring for jump intensity increases when COT shorts exceed 100k contracts or the rate differential compresses below 2%. The model concludes that unwind risk is no longer purely theoretical tail risk, warranting proactive risk management through appropriate position sizing, hedging, and surveillance of key market indicators. Calibration sources include implied volatilities from Saxo/MacroMicro (August 2026), 2024-2026 historical data, CFTC positioning, and standardized Steelldy jump-diffusion stress parameters.
