The VIX term ratio gave no clear gain in weekly Nasdaq variance forecasts in a 2018–2025 archive test
We tested whether the log ratio of three-month VIX to one-month VIX improves weekly Nasdaq Composite variance forecasts. Each Monday forecast used an ALFRED snapshot dated on the preceding Friday. A fixed ridge model used recent variation and one-month VIX. A second model added the term ratio. Both used the same last 260 eligible completed weeks. The main test had 401 weeks in 2018–2025. The added input changed mean QLIKE loss by -0.00428, with a paired 95% eight-week block interval from -0.02422 to +0.01770. Four- and 13-week blocks, fresh inputs, final-vintage outcomes, and removal of 2020 gave no clear added gain. Variance squared error was lower by 23.75%, but its interval included zero. Yearly changes had both signs. A source outage removed 16 planned test origins. Early training features use a 2014 historical archive; all later features are fixed at their own cutoff. The package preserves code, forecasts, fit states, source hashes, aggregate scores, figures, negative findings, checks, and repeat instructions. Raw source data and direct outcomes are retained under source terms. This is a risk-forecast test. It does not establish a trading profit.
volatility forecasting · VIX term structure · Nasdaq Composite · point-in-time data · chronological evaluation · negative result
References
4- usesPast VIX values reduced weekly Nasdaq Composite variance forecast loss in 504 tests from 2016 to 2025 sci:sha256:4adcb5f1bc1bda2df920b7b53f7fb09b57658a671e2df019f32b1c4d19f5143c
Its dated weekly VIX study motivates the comparison model. Our code, Monday origins, rolling window and interval-based sources differ.
- usesdoi:10.1093/jjfinec/nbp001
Motivates past variation inputs at several horizons. We use daily closes, not intraday realized variation.
- supportsdoi:10.1080/00036846.2025.2601895
Its abstract motivates a test of VIX maturity information. It does not support our specific added-ratio result.
- usesdoi:10.1016/j.jeconom.2010.03.034
Context for comparison with a noisy variation measure and use of QLIKE.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesAdding a VIX term ratio raised weekly Nasdaq return forecast error by 2.4% in a 2020–2025 archive test
We reuse its retained ALFRED source bytes, MIT archive and calendar routines, downloader, and source timing rules. We test signed returns and negative-return probabilities with added signed-return controls and a later main sample.
- usesThe VIX term ratio gave no clear gain for weekly Nasdaq 3% loss probability forecasts in a 2020–2025 archive test
We reuse its dated ALFRED source bytes and MIT archive reader, calendar rules and downloader. We change the target to fixed loss events and remove historical backfill.
- usesThe VIX term ratio gave no clear monthly Nasdaq return forecast gain with dated inputs in 2018–2025
We reuse its retained dated VIX and VXV source files and its documented source outage. It tests weekly variance. We test monthly mean returns with original code.