Polymarket Spread History Data

The archive stores bid, ask, mid, and the derived quoted spread for every 250ms row across the eight archive coins, sports, and event markets — so spread behavior is not a snapshot you chase but a historical series you query.

Because executable prices bunch on the 0.99/0.995 grid 98.6% of the time, the quoted spread concentrates on small fixed steps, and the series becomes a measurement of when it is one step, when it is wide, and when one side simply vanishes.

In the 5-minute corpus the spread is narrow exactly where the market has made its decision — near settlement the rest sits at the bound — and wide where two sides still contest the price. One-sided books (16.9% of snapshots) formally have infinite spread, and 3.24% crossed snapshots have negative quoted spread; the honest series reports both states instead of filtering them.

Cost-to-cross — the distance from mid to the executable side — is the fill-based version of the same number and is available per row.

Spread history feeds execution modeling: a strategy that crosses costs a round trip of the quoted spread plus slippage, and the archive makes both measurable per market and per minute.

For venue research, spread persistence — how long the spread stays wide before re-tightening — is the honest summary of thinness, and it is queryable from the series directly.

Quoted spread is a minimum, not a price: the honest execution analysis also models slippage against the resting ladder, and both quantities come from the same recorded rows.

• Spread explained
• Slippage data
• Bid-ask data