OHLCV Data Quality Pipeline

1 min read

Placeholder

Placeholder one-sentence summary of the OHLCV data quality pipeline.

Symbols covered

500+

Anomalies flagged

0.3%

Uptime

99.9%

Placeholder

Challenge

So I noticed downstream models were silently training on stale and split-adjusted bars, and identified that vendor feeds needed per-source normalization rules.

Solution

Designed a staged validation pipeline with schema checks, corporate action reconciliation, and anomaly quarantine before data hits the feature store.

Placeholder

PythonAirflowPostgreSQLGreat Expectations

Result visual