Graduation Year

2026

Date of Submission

4-2026

Document Type

Open Access Senior Thesis

Degree Name

Bachelor of Arts

Department

Economics

Second Department

Mathematics

Reader 1

Ricardo Fernholz

Reader 2

Evan T. R. Rosenman

Terms of Use & License Information

Terms of Use for work posted in Scholarship@Claremont.

Rights Information

Β© 2026 Prateek D Vyas

Abstract

This thesis empirically tests the rank-rebalancing mechanism of Stochastic Portfolio Theory (SPT) across commodity futures, foreign exchange futures, and equity ETFs. The Reverse Price-Weighted strategy (RPW) assigns, to each asset, the market weight of the asset at the opposite price rank, and generates an annualized excess return of 2.90% over the price-weighted (MKT) commodity benchmark, during the period of November 1977 to October 2025 (HAC t = 2.058, pΒ = 0.040). The differential Sharpe ratio (dSharpe) of 0.245 is confirmed by a stationary block bootstrap, with a 𝑝-value of 0.015, and factor regressions controlling for carry, momentum, and value yield an alpha of 2.90% (p = 0.041), and the SPT master equation decomposition is satisfied by construction; the excess-growth component estimated from rolling covariance matrices produces a cumulative correlation of βˆ’0.735 with realized outperformance (R2 = 0.540), a limitation of the short-window covariance estimator discussed in Chapter 5. The premium is positive for all four subperiods spanning 48 years, including the post-2018 out-of-sample period that included the COVID-19 commodity shock and the 2022 energy crisis. When applied to foreign exchange futures and equity ETFs, RPW fails exactly as predicted: in markets where cross-sectional price ranks are too stable for local time to accumulate meaningfully. The 12-month rank autocorrelation AC(12) acts as a continuous cross-asset moderator of RPW viability, with a slope of βˆ’0.352 across seven observations from three asset classes, transforming a mathematically sufficient condition into a usable screening tool.

Download

Share

COinS