quant research // summarizing the paperLOADING
Kenichiro Shiraya, Tomohisa Yamakami, Akira Yamazaki · 2026-07-09
The paper builds a way to estimate the 'stochastic discount factor' (a curve describing how investors price different market outcomes) using only S&P 500 option prices and their implied volatility. It finds this curve is non-monotonic, showing a hump on the shallow-put side that becomes a W-shape at longer maturities, and links that shape to stochastic volatility dynamics. It then uses this to forecast the equity premium out-of-sample, reporting better performance than benchmarks like Martin's bounds.
Why it matters: The approach uses forward-looking option market data rather than historical returns to gauge expected market returns, which could be relevant for those building return or risk forecasts. The claimed out-of-sample edge over an established benchmark (Martin bounds) suggests option-implied information may add predictive value, though this is an academic result on one index.
⚠ Results are model-based and limited to S&P 500 options; out-of-sample statistical outperformance does not guarantee tradable or robust real-world gains.
This paper proposes a stochastic discount factor (SDF) scaled by time-varying volatility. By utilizing prices and market data implied solely from S\&P 500 options, the proposed framework recovers a stable, non-monotonic SDF that captures the pure forward-looking expectations of market participants while mitigating observation noise. Our empirical analysis reveals that the SDF exhibits a distinctive hump on the shallow put side, which transitions into a more clearly defined W-shape as the time to maturity increases, identifying maturity as a key factor influencing the intensity of the central hump. We show that this structural feature can be theoretically rationalized by stochastic volatility dynamics under a constant market price of risk. The equity premium derived from the time-varying volatility scaled SDF demonstrates superior out-of-sample predictive performance relative to existing benchmarks, such as the Martin bounds.
Go deeper: a full research-committee breakdown of this paper, its assumptions and failure modes, and how its method would apply to a specific ticker or your watchlist. See StockTools AI →
AI summary generated from the paper’s public abstract via arXiv; it may miss nuance — read the source before relying on it. Thank you to arXiv for its open-access interoperability; StockTools is not affiliated with arXiv, and all rights remain with the authors. Educational only, not financial advice.