Volatility Targeting Is Risk Management, Not Alpha: Evidence from 31 ETFs, a Semiconductor Stress Test and a Decade of Gold

Abstract

Does scaling a single asset’s exposure by the inverse of its recent volatility improve its risk-adjusted return? We test causal daily volatility targeting on 31 liquid exchange-traded funds (ETFs) covering U.S. equity indices, sectors, Treasury and credit bonds, commodities, international equity and real estate, using total-return data from January 2008 to August 2026 (4,695 trading days). Net of costs and leverage financing, the mean change in Sharpe ratio relative to buy-and-hold is +0.01 (median 0.00); 16 of 31 assets improve, none significantly at the 5% level under the Ledoit–Wolf test, and the smallest false-discovery-rate q-value is 0.77. Moreira–Muir spanning alphas are significant for two of 31 assets and for none once the Fama–French five factors and momentum are added. A Cederburg-style real-time combination of the managed and unmanaged asset lowers the Sharpe ratio by 0.12 on average. What volatility targeting reliably does is reshape risk: volatility falls by 16% at the median, excess kurtosis falls from 10.3 to 1.9, and at matched volatility the maximum drawdown is smaller for 23 of 31 assets—but larger for gold, silver, gold miners and broad commodities, whose slow bear markets occur at low volatility. The gains are confined to 2008–2013 (mean +0.14); they are negative in 2014–2019 and 2020–2026. We also audit the exploratory study that motivated this paper. Its headline results—a semiconductor ETF (SMH) Sharpe ratio of 0.94 against 0.57 for holding it, and a gold walk-forward of 0.89 against 0.87—are reproduced exactly from code, and both dissolve. The SMH gap was manufactured by an unadjusted 2-for-1 split that the strategy happened to sit out; the “volatility-target” rule was in fact a moving-average trend filter whose returns correlate 0.94 across assets with the filter alone and 0.12 with volatility scaling alone; and the gold result depends on how two-decimal ties in the in-sample Sharpe ratio are broken. Over 2000–2026, pure volatility targeting of SMH has a lower Sharpe ratio than holding it (0.38 against 0.44) and lost more in the dot-com bust.Declaration of generative AI use: All research questions, hypotheses, study designs and conclusions are the author’s own, including the decision to re-audit the author’s earlier exploratory results, which identified data errors in them. Under the author’s direction, generative AI tools (Claude Code, Anthropic) wrote most of the analysis code and ran it, and assisted with drafting, editing and formatting the text, and with typesetting. The author reviewed all code, results and text and takes full responsibility for the content of this paper.