A six-month risk decay rule gave no clear forecast gain in 20 US market tests in 2005–2025 under assumed archive release dates

We tested whether an old market risk state should move toward its ten-year level as the forecast lead increases. Twenty annual French Data Library archives supplied July inputs for 240 later monthly US market excess returns, from October 2005 to September 2025. Historical release by September 30 and archive preservation are assumptions. The primary six-month decay rule had a mean QLIKE gain of 0.1054 over a fixed July state. Its paired 95% annual-block interval was -0.0428 to +0.2812. The gain over a ten-year baseline was also uncertain. At 5 basis points per dollar traded, a long-only paper account gave an annual utility gain of 0.146 percentage points, with interval -0.664 to +1.167. No signed-return model was tested. The package preserves all fixed rules, 3600 forecast rows across three data cases, 450000 uncertainty draws, source hashes, original code, figures and repeat instructions. Independent checks passed, and a fresh run reproduced 34 numeric result and ledger files byte for byte. The same market outcomes occur in earlier SCI archive tests. This test gives no clear gain under the slow data schedule; it does not establish a reliable trading rule.

risk forecasts · mean reversion · annual archives · point-in-time data · QLIKE · forecast horizon · paper trading · negative result