Library term·FinTech & data science
Missing Data in Financial Time Series — Imputation Without Lookahead
MCAR/MAR/MNAR taxonomy, forward-only fills, Kalman-style smoothers vs naive ffill dangers in backtests.
Authored by·Editorially reviewed
Onur Erkan YıldızFounder, Financial Engineer · CMB-licensed
Higher education in Financial Engineering and Money & Capital Markets. SPK (Turkey CMB) licence. 16 years across institutional markets, research, and quant-driven analytics.
Taxonomy
MCAR (missing completely at random), MAR (conditional on observed), MNAR (informative missing) — each imputation policy differs.Causal imputation
Never use future rows to fill past holes in research features — classic subtle leakage.Methods
Piecewise linear bridge, EMGARCH state-space, Kalman filtering when model identified; simplest last observation carried forward may destroy variance structure — document side effects.Finvestopia
Transparency on data cleaning is part of trustworthy quant storytelling.Related entries
Walk-Forward Matrix: Statistical Design for Honest Out-of-Sample Evaluation
Rolling train/test grids, multiplicity control, causal features and why a matrix of splits beats single-window storytelling.
PostgreSQL TimescaleDB for Tick & OHLC Warehouses
Hypertables, compression tiers, retention, continuous aggregates — scaling research & BI without losing SQL.
Educational content authored by our team — informational only, not investment advice.