The Register records the original rule, the result, and any mistakes in the process. It exists so FP1 cannot quietly rewrite the story later. This entry compares two registered rules with their recorded outcomes; it is not an overall accuracy score.
Neither registered call changed. No thresholds were moved. The page was five days late, and that remains visible.
Because a forecast is only useful if you can tell whether the forecaster changed the rules after seeing the result.
The spread finished at -68.745, still inside its band.
The index crossed 55 once. The rule required two consecutive weeks.
No line was moved during the window.
The result was published five days after the grade date.
Replay a recorded decision
Follow the original position, the rule, the evidence and the result. These are published historical records, not hypothetical cases.
Belief · 20 July baseline · window opened 27 July 2026
The index describes the selected prediction markets, not physical capacity in use. The registered question was whether subsequent readings would meet one of two conditions.
Explore the boundaries in the diagram, then follow what happened during the window.
Belief · Rule fixed for this window
Alternatively, any one weekly reading below 40 would trigger a change. The rule must be applied as written; a large rise within the window is not a substitute.
Belief · 27 July–31 August · primary replay at 21:00 UTC
The previous week was 45.315. Only the final week was above 55, and none of the six readings was below 40. This is the reconstructed record, not a live feed.
Belief · 31 August grade · published 5 September 2026
The final jump mattered, but it did not meet the original rule. There was only one week above 55, and no weekly reading below 40. The recorded assessment held.
Published 5 September · five days late. The result was reconstructed from the recovered method and archived data. The delay remains attached to the record; the thresholds were not reset.
Inspect the full audit trail ↓Capital · 20 July baseline · window opened 27 July 2026
FNC-1 is FP1’s public-equity proxy for the AI buildout; SOX is the semiconductor index. The spread compares the two. It does not measure productive use of capacity.
Explore the boundaries in the diagram, then follow what happened during the window.
Capital · Rule fixed for this window
The registered wording was “closes to −40 points, or widens beyond −120 points.” A move that remained between those boundaries would leave the assessment unchanged.
Capital · 31 August · weekly-explicit reconstruction
FNC-1 was 134.943 and SOX was 203.688. The spread had moved from −80.9 toward −40, but remained between −120 and −40. These are the published endpoints; this replay does not invent an intervening weekly path.
Capital · 31 August grade · published 5 September 2026
Movement is not the same as meeting the rule. The grade-date spread had reached neither boundary, so the recorded assessment held. No threshold was moved.
Published 5 September · five days late. The result was reconstructed from the recovered method and archived data. The delay remains attached to the record; the thresholds were not reset.
Inspect the full audit trail ↓The recorded outcome is unchanged. The full audit trail below retains exact values and limitations, including the reconstruction. Read the published data ↗
FNC-1 is FP1’s public-equity proxy for the AI buildout; SOX is the semiconductor index used for comparison. The Belief Index reflects the selected prediction markets. These measure market prices and expectations, not physical capacity in use. Read the implementation note →
For anyone who wants the exact rule, opening value, grade-date value, and publication status.
| Field | FNC-1 / SOX spread | Belief Index |
|---|---|---|
| Registered threshold | Closes to -40, or widens beyond -120 | Two consecutive weekly readings above 55, or any one below 40 |
| Opening reading | -80.9 · 20 July weekly close | 49.7 · 20 July weekly close |
| Grade-date reading | -68.745 · FNC-1 134.943, SOX 203.688 | 57.490 · 21:00 UTC primary replay |
| Final result | No change | No change · one week above 55 |
| Threshold reset? | No | No |
| Publication | 5 September 2026 · 5 days late | 5 September 2026 · 5 days late |