Research dossier / release 3.0
Auction comparisons
Breakdowns within scheduled closing windows, with benchmarks and methodological cautions.
Bids through Oct 18, 2026America/Toronto reporting cutoffOct 17 excludedThe groupings below are scheduled closing windows, not verified unique auction IDs. Do not confuse scheduled times with confirmed closure.
Three-second comparisons within auction windows
| auction_window | hf_cross | hf_rapid3 | other_cross | other_rapid3 |
|---|---|---|---|---|
| 2026-10-18 13:00:00 | 1575 | 490 | 17802 | 1520 |
| 2026-10-18 18:30:00 | 501 | 162 | 5698 | 278 |
| 2026-10-18 21:30:00 | 9 | 1 | 7 | 1 |
| 2026-10-19 19:00:00 | 0 | 0 | 131 | 19 |
Better benchmark design
High-frequency accounts were selected using rolling-minute speed, so their faster gap distribution is partly a selection effect. Match controls on number of bids, active-day coverage, closing window, category and available inventory; hold out later days for validation. Compare complete days, using the same predeclared thresholds each time.
October 19 data refresh
These groups use reported scheduled closing windows as proxies. They are not necessarily equivalent to distinct auction identifiers and are not verified outcomes.
Full dataset export: completed-day bids · October 19 catalogue snapshot · coverage and limitations.