A feature derived from news text
Question
How would you evaluate a news-embedding feature for next-day returns while controlling for duplication and timing?
Short answer
Use original publication and ingestion timestamps, cluster syndicated or near-duplicate stories, and prevent future revisions entering historical features. Fit any supervised reduction inside training folds and record the embedding model version and its provenance. Compare against simple text and non-text baselines under identical execution assumptions. Audit whether model pretraining or retrieval introduced future event knowledge relevant to the historical test.
Original question and answer written by SelfMadeQuant.com.
See the hint and the full worked solution in the question bank (free account).
Open the worked solutionLessons behind this question
- Finance-aware validation
- Cross-Validation and Information Criteria
- Sample Spaces, Events, and the Probability Axioms
Start with a free lesson: Probability Fundamentals is part of a free topic. Create a free account, no card needed, to read it in full with worked solutions.
Read it free