Adding VXN gave no clear utility gain in a 2017–2026 Nasdaq-100 paper test with no leverage

We tested whether a better monthly risk forecast improved a fixed paper portfolio decision. The test has 116 monthly entries from January 2017 to August 2026. Each forecast uses its own ALFRED data vintage, cut two calendar days before the month began. We reuse the forecasts and market sample from an earlier SCI study; this is not an independent validation sample. Index allocations have a 100% cap. Entry is at the first session close of each month. Exact accounting includes weight drift, initial purchase, final sale, and assumed costs of 0, 10, or 25 basis points per dollar traded. At a 15% annual volatility target and 10 bp cost, adding Cboe VXN to a model of past returns raised the annual mean-variance utility score from 11.11% to 11.94%. The paired gain was 0.83 percentage points, with a 95% six-month-block interval of -0.99 to +2.85 points. All planned target, cost, and block-length intervals include zero. March 2020 supplies 71.8% of the full-sample gain. The VXN rule had a larger monthly-mark drawdown, and a raw 21-session risk control had a higher point utility score. Four stale-input months keep existing index units and cash. A change in outcome vintage leaves the result unchanged. Independent accounting checks passed for 9,744 rows, and all 15 core output files matched byte for byte in a clean repeat. We preserve code, source hashes, dated forecasts, model states, allocation schedules, uncertainty draws, figures, and negative results. Complete source-derived ledgers are retained under provider terms. The simulation uses price-index returns and zero cash yield. It does not establish an executable fund return, return-direction skill, or a reliable positive utility gain.

Nasdaq-100 · VXN · volatility timing · dated data · ALFRED · paper trading · transaction costs · mean-variance utility · negative result · reproducibility