AI Science Tracker
EN

Rolling ridge forecasts failed to beat zero in an ETF return backtest

KTH Royal Institute of Technology; Stanford University · 27 Oct 2026 · other

Discovery

Rolling ridge forecasts failed to beat zero in an ETF return backtest

Source check: Source checked. Not independent scientific confirmation.

Finding: In a 2005–2025 ETF study, a rolling ridge model using past return features produced negative mean out-of-sample R² at every tested regularization level against a zero-return forecast. Scores approached zero as stronger regularization shrank the predictions.

Technology: Ridge regression with rolling retraining

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The fitted regression generated return forecasts from past observations, enabling a retrospective comparison of predictive signal against the zero baseline.
Test and result
The model was repeatedly fitted on trailing observations and scored on the following period. The paper compared multiple regularization levels with a zero forecast.
Limitation / still unknown
The ETFs were selected from ones still trading at study time, producing survivorship bias. The result concerns the chosen predictors, horizons and historical universe.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.36491v1 · https://arxiv.org/html/2610.36491v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: §§2, 5.3 and 6.3, Figure 25 and the disclosed universe selection.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse other discoveries.