Polymarket Mid-Price History Data

The archive records bid, ask, and mid per 250ms row across every archived market — including the Venue's own published mid where it exists — so the mid-price history is complete and joinable to the ladder that produced it.

The mid is a derived quantity but a faithful one here: stored alongside bid and ask, recomputable, and never filtered when one side is absent.

Measuring the venue shows mid-price history is clean exactly when both sides rest — and meaningless when one side empties: 90% of final-second books in the 5-minute corpus are single-sided, and a mid pasted over an empty side is arithmetic, not a price.

The honest definition of the archive's mid-price explainer is the executable floor: at the bound, mid equals the one real price. Studied data confirms the venue executes at the bound, not at an interior mid.

Mid-price backtesting is the classic fill error: it assumes you traded a price no one offered. The archive's mid series is the honest correction set — run your strategy against the mid, then against the executable side, and the difference is your true cost.

The measured magnitude of that difference is the venue's own split: 16.9% one-sided snapshots plus 3.24% crossed mean mid-fill bias is structural, not incidental.

The mid is a display, not a promise: the archive serves it with bid and ask so every study can see when it was tradable, and the site's house position is that executable price — not mid — is the price that would have filled you.

• Mid price explained
• Fill simulation
• Current prices