Accuracy
When we say 70%, does it happen 70% of the time?
| We said | It happened | Events |
|---|---|---|
| 4% | 3% | 75 |
| 14% | 11% | 18too few to read |
| 25% | 13% | 24 |
| 33% | 29% | 7too few to read |
| 45% | 41% | 17too few to read |
| 55% | 60% | 5too few to read |
| 64% | 43% | 7too few to read |
| 75% | 64% | 11too few to read |
| 85% | 89% | 18too few to read |
| 97% | 100% | 104 |
How this is measured
The last price we showed at least 30 seconds before each event closed, against what actually happened. Only events that have settled, 286 of them. A price taken in the final seconds is nearly the answer, which is why it is not used.
What this does not prove
Most of these are short Bitcoin windows rather than a spread of subjects, and most of the prices came from our own pricing rather than from a crowd of traders. Any row with fewer than 20 events is marked, and should be read as nothing at all.
How events settle is on how events are decided. What trading costs is on what it costs.