Pregame line movement concentrates in the final hour (NBA most of all)
pregame per-series consecutive-snapshot |devigged_prob| movement, bucketed by minutes-to-commence, from own scraped line_history
| sport | n series used | n move pairs | final hour movement share |
|---|---|---|---|
| kbo | 0 | 0 | - |
| mlb | 5794 | 1307128 | 0.1813 |
| nba | 160 | 182054 | 0.3247 |
| npb | 0 | 0 | - |
| soccer | 66 | 565548 | 0 |
| soccer_intl | 494 | 164072 | 0.261 |
How to read this figure
Bars are minutes-to-start buckets; height is the mean absolute devigged-probability move between consecutive snapshots, and final_hour_movement_share is the fraction of total absolute movement that happens in the last 60 minutes.
Published operands
Pregame movement coverage by sport
Mean movement per pair is total absolute movement divided by move pairs. Final-hour movement share is a separate fraction of total movement, not a per-pair mean.
| Sport | Availability | Observation window | Snapshot cadence | Pregame snapshots | Move pairs | Series used | Mean movement per pair | Final-hour movement share | 6h+ mean | 3-6h mean | 1-3h mean | 0-1h mean |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| KBO | Unavailable | 2026-07-04 to 2026-07-17 (13 days; 13 files) | Not published | 0 | 0 | 0 | Not published | Not published | Not published | Not published | Not published | Not published |
| MLB | Published | 2026-06-18 to 2026-07-17 (30 days; 30 files) | median 1.083, p90 2.033, mean 2.205 min | 1,403,814 | 1,307,128 | 5,794 | 0.01% | 18.13% | 0.01% | 0.01% | 0.02% | 0.02% |
| NBA | Published | 2026-06-18 to 2026-07-17 (30 days; 30 files) | median 1.117, p90 1.183, mean 1.866 min | 183,478 | 182,054 | 160 | 0.01% | 32.47% | 0.01% | 0.02% | 0.02% | 0.12% |
| NPB | Unavailable | 2026-07-04 to 2026-07-17 (14 days; 14 files) | Not published | 0 | 0 | 0 | Not published | Not published | Not published | Not published | Not published | Not published |
| SOCCER | Published | 2026-06-18 to 2026-07-17 (30 days; 30 files) | median 1.100, p90 3.400, mean 3.131 min | 606,832 | 565,548 | 66 | 0.00% | 0.00% | 0.00% | Not published | Not published | Not published |
| International soccer | Published | 2026-06-19 to 2026-07-17 (29 days; 29 files) | median 1.067, p90 1.533, mean 3.198 min | 169,976 | 164,072 | 494 | 0.03% | 26.10% | 0.02% | 0.02% | 0.05% | 0.07% |
| TENNIS | Published | 2026-07-03 to 2026-07-17 (15 days; 15 files) | median 1.117, p90 1.217, mean 2.374 min | 840,232 | 809,986 | 2,510 | 0.01% | 10.70% | 0.01% | 0.01% | 0.01% | 0.02% |
| WNBA | Published | 2026-07-04 to 2026-07-17 (14 days; 14 files) | median 1.117, p90 1.167, mean 2.086 min | 334,592 | 333,540 | 920 | 0.02% | 11.76% | 0.03% | 0.02% | 0.02% | 0.04% |
Independent-game counts are not published. Series used and dense snapshots are reported separately and do not supply an independent-game count.
Continue with the data
Open the interactive analysis: Devigged probability movement by time to closeWhat it means
The artifact records when movement occurred; it does not identify its cause. In the NBA the per-snapshot price motion inside the last hour (0-1h mean_abs_move 0.00117) is about seventeen times the size of the early-window motion (6h+ mean_abs_move 7e-05).
Caveats and confounds
descriptive_only -- this is consecutive-snapshot movement from our own scraped line_history (min_moves_floor 20), a microstructure measurement, not a trading signal.