Data that preserves exactly what was known at each moment in history, with no later revisions leaking backward.
Point-in-time (PIT) data records the state of the world as it was actually known at each timestamp, rather than overwriting the past with later corrections. It matters because so much data gets revised: an economic figure is restated, a company refiles, a vendor corrects an error. A database that quietly applies those revisions backward lets a backtest "know" things that no one knew at the time.
That look-ahead bias is subtle and devastating - it inflates historical performance in ways that vanish the moment a strategy trades live. Maintaining PIT history means storing every value with the time it was first known and never silently rewriting it, so a query for "what did we know on this date" returns exactly that. It is one of the least visible and most important properties of a research-grade dataset.
A live evaluation measures EdgeOrigin against your coverage requirements: decision-ready real-time data, delivery latency into your systems, and record-level provenance.