This paper presents a probability-driven trading framework based on condi-tional exceedance statistics rather than price or return forecasting. Using daily SPY data,conditional probabilities of large future price moves given large past moves are estimatedover fixed horizons using data up to the end of 2024. These probabilities are organized intotransparent lookup tables and translated directly into systematic, event-driven buy–hold–selltrading decisions with fixed holding periods. All parameters are frozen prior to evaluation,and performance is assessed strictly out-of-sample on 2025 data. Across multiple param-eter configurations, the resulting strategies exhibit strong risk-adjusted performance, withSharpe ratios exceeding 4 for selected regimes. The results suggest that long-horizon condi-tional structure in price dynamics can be exploited using simple, interpretable probabilitytables without reliance on complex predictive models.
ARUN RAMANATHAN S (Tue,) studied this question.