Reading the evidence
Data quality is fitness for a specific use across completeness, accuracy, consistency, timing, provenance and meaning. Clean formatting does not establish that observations represent the intended facts.
Make it click
Jars can be neatly arranged yet contain the wrong ingredient, an old batch or a missing label. Organisation helps only after identity and freshness are checked.
Now put that into markets
Market records need stable identifiers, publication times, units, corporate-action treatment and lineage. Missing fields should remain missing rather than guessed.
Validation checks schema, ranges, uniqueness and reconciliation; provenance links records to sources and parser versions. Corrections need append-only audit.
Start from the decision or measure the data must support, then test each required field. Reconcile totals and inspect samples at boundaries.
Automated checks catch known failure modes, not semantic mistakes no rule anticipates. More rows amplify systematic error rather than dilute it.
Takeaway
