Dynamic Determination of the Margin Ratio Based on a Market State Indicator

Individual investors engaged in margin trading must themselves decide at what level to maintain the margin ratio of their account. This paper formulates an investment rule that sets this target level each quarter as a linear function of a continuous market state indicator (constructed from the change in the U.S. 10-year Treasury yield and from the VIX), does not make additional purchases of the NASDAQ100-linked asset through margin trading, and adjusts the difference with a gold-linked asset; the rule is frozen and published before the start of operation. In a simulation using yen-denominated monthly data from July 1991 to July 2026, compared with a fixed margin ratio of 132.08% that matches the average position ratio over the full sample period, the annualized return from January 1997 onward was 16.15% versus 14.06%, the Sharpe ratio 0.770 versus 0.700, the Sortino ratio 1.187 versus 1.057, and the regression intercept α 1.919% per year (Newey–West t-statistic 2.55). No additional margin (margin call) occurred in either account, but the minimum drawdown to margin call was smaller under this rule (24.72% versus 28.99%). However, because the coefficients include values transferred from prior studies and data from the same period were used to design the rule, these results are not treated as evidence of effectiveness. This paper presents a plan, with the comparison benchmark, evaluation metrics, and decision criteria fixed in advance, to evaluate the rule in a virtual account over the 120 months following publication.