Replay markets
How to Replay Polymarket Markets
Replaying a Polymarket market means walking its recorded 250ms frames in order — every reprice, every empty side, every settlement meter — and it is the single most honest test of both the data and your strategy. This page is the replay recipe.
Figures measured as of 2026-09-24 on the published PolyOrderbooks archive.
The recipe
Replay in four steps
Validation
Replay as a data test
Reconstruction is the fidelity benchmark this site publishes: replaying recorded books alone reproduced 36.6% of on-chain settlement outcomes in the replay family — the honest statement of how much of the venue's record is book-derivable versus reference-dependent.
A replay that ends at the recorded outcome passes the data test; one that drifts has a bug in order, grid alignment, or flag handling — and the replay harness tells you which.
Worked example
Replay a 26-second collapse
Replay the Aug-23 multi-coin repricing night frame by frame: the family collapsed together within 26 seconds, ETH's bid side lost ~96% while the ask widened +491%, and the reference prints land mid-run. Stepped at 250ms, the sequence is the evidence; at minute granularity it is a mystery.
The same replay on a strategy tells you what you would have done at each second — before you ever risk real USDC.
Honest
The honest limits
A replay is only as faithful as its fill model: walking available depth at the executable side is honest; assuming fills at a mid is fiction. The archive's fill simulation defines the honest model this recipe uses.
FAQ
How do I replay a Polymarket market?
Pull the books and tape for a window, order frames by UTC second, step your strategy at the executable side, and converge on the recorded outcome.
What is the reconstruction benchmark?
Recorded books alone reproduced 36.6% of on-chain settlement outcomes in the replay family — the honest fidelity floor.
Can replay validate a strategy?
Yes — it replays your decisions second by second against the real ladder before any capital moves.