1-Hour Polymarket Data

Price-threshold markets ("ETH above $4,000 by Friday"), event-linked contracts, and longer-horizon Up/Down-adjacent instruments live at horizons that map naturally to hourly bars. The archive resamples any of them to 1-hour OHLCV from the raw 250ms frames.

Because the raw data is second-level, an hourly study never has to choose between granularity and window length — the same market produces 1-hour bars and second-level rows from the same stored frames.

Hourly OHLC on binary markets builds from the executable price series: each 250ms row supplies the state, and aggregation produces the open, high, low, close, and volume per hour. The archive's metrics (bid, ask, mid, depth, volume) are stored per row precisely so this resampling is lossless to the raw.

The honest aggregation choice: build bars from the executable side you would trade, not from the mid — on a venue whose prices bunch at 98.6% on the 0.99/0.995 grid, bar construction decides the signal.

Aggregating to the hour hides the venue's one-sided structure — 16.9% of 5-minute snapshots and 76.2% of final-minute empty sides vanish into an hourly open-close pair. The measured discipline is to report the fraction of sub-hour frames that were one-sided or crossed alongside the bars.

The archive's crossed and one-sided flags travel with the data, so an hourly study can always report the texture underneath the aggregation.

There is no native hourly crypto Up/Down on the venue, and a page that claims one would be noise. The honest product is the resampling: keep the same 250ms source of truth, and let the hourly bar be a view instead of a separate dataset.

• 5-minute data
• OHLC data
• Intraday data