Every king node we publish is frozen at publish time and graded against the forward price path — no revisions, misses included. Updated daily after the close.
| Hour (PT) | Tag Rate ($10) | Setups | Median Tag |
|---|---|---|---|
| 06:00 | 74.1% | 27 | 5.0m |
| 07:00 | 69.6% | 46 | 10.0m |
| 08:00 | 70.9% | 55 | 10.0m |
| 09:00 | 69.1% | 55 | 5.3m |
| 10:00 | 76.7% | 60 | 5.0m |
| 11:00 | 94.1% | 55 | 5.0m |
| 12:00 | 100.0% | 46 | 5.0m |
| 13:00 | -- | 18 | -- |
How often the close landed near the final king node (48 sessions). A single-instant check — kept separate because the king migrates intraday.
How gamma exposure levels predicted market direction each trading day.
Held/missed grades the close against the final King Node — a single-instant check, like the End-of-Day Pin Rate above. The headline tag rates instead grade each level frozen at publish time against the next two hours of price action, so a "missed" close can still be a tagged setup.
Freeze at publish. Every 30 minutes during market hours, the current king node is frozen. Whatever the model says later, that snapshot is what gets graded.
Distance gate. If price is already sitting on the level, there's nothing to predict — those anchors are excluded, not counted as wins.
Windowed tag. A setup counts as tagged only if price actually touches the frozen level (within the $ band) at any point in the following 120 minutes.
Window guard. A non-tag only counts as a miss when the full window was observed — end-of-day truncation never inflates the score.
No revisions, no cherry-picking, misses included. The same pipeline that publishes the level grades it.
The live heatmap shows the king node, gamma walls, and regime in real time — the same levels this page scores.
Start 14-Day Trial