The Sealed Record: A Prediction Ledger Sealed Before the Fact and Graded in the Open

This paper publishes the graded record of a prediction desk: a forecasting ledger operated by Aldred Industries in which every call is fixed in writing before the event it concerns — the claim, its threshold, its check date, and the named source that will settle it — and graded afterward, in the open, hit or miss, against that source. As filed in the register's grading run of November 5, 2026 (09:19 EST), the desk stands at 672 predictions sealed, 231 graded: 196 hits and 35 misses, a hit rate of 84.8%, with 441 seals still open. Of the 231 grades, 124 are itemized in the paper seal by seal — every one reproducible from the register's own files, with its pre-registered line and its outcome printed, hits and misses in the same tables at the same size. The remaining 107 grades are a carried ledger line from the desk's first week (93 hits, 14 misses) whose individual item records do not survive as a single register; no item-level claim is made over them anywhere in the paper. On the 124 itemized grades, the seals carry their own baselines: election seals print a confidence figure at sealing, one central-bank seal prints the market odds it took, and the remaining binary claims are computed against a coin-flip baseline. Against those baselines the itemized record produced 103 hits where 86.9 were expected; the exact probability of a record at least this strong, if each call's true chance were its printed baseline and the calls were independent, is approximately 1.9 in 10,000. Collapsing each market settlement day to a single trial, it is approximately 1.1 in 1,000. Under the harshest dependence treatment considered — the whole of election night counted as one observation as well — it is 0.091, and the itemized record does not by itself exclude chance at the conventional level; that bound is printed in the paper at the same size as the headline figure. The method that generates the calls is proprietary and is not disclosed, by design: the record is built so that it can be checked without it, because every seal precedes its outcome and every grade names its source. What the record means is left open. What the record is, is stated exactly. Keywords: forecasting; pre-registration; prediction ledger; sealed predictions; calibration; binomial test; Aldred Industries; prediction desk. ---

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-05
DOI
https://doi.org/10.5281/zenodo.23982777
Primary Topic
Forecasting Techniques and Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The Sealed Record: A Prediction Ledger Sealed Before the Fact and Graded in the Open

William Alexander Aldred
1 links
Zenodo (CERN European Organization for Nuclear Research)
Forecasting Techniques and Applications
preprint

The Sealed Record: A Prediction Ledger Sealed Before the Fact and Graded in the Open

William Alexander Aldred
preprint en

Abstract

This paper publishes the graded record of a prediction desk: a forecasting ledger operated by Aldred Industries in which every call is fixed in writing before the event it concerns — the claim, its threshold, its check date, and the named source that will settle it — and graded afterward, in the open, hit or miss, against that source. As filed in the register's grading run of November 5, 2026 (09:19 EST), the desk stands at 672 predictions sealed, 231 graded: 196 hits and 35 misses, a hit rate of 84.8%, with 441 seals still open. Of the 231 grades, 124 are itemized in the paper seal by seal — every one reproducible from the register's own files, with its pre-registered line and its outcome printed, hits and misses in the same tables at the same size. The remaining 107 grades are a carried ledger line from the desk's first week (93 hits, 14 misses) whose individual item records do not survive as a single register; no item-level claim is made over them anywhere in the paper. On the 124 itemized grades, the seals carry their own baselines: election seals print a confidence figure at sealing, one central-bank seal prints the market odds it took, and the remaining binary claims are computed against a coin-flip baseline. Against those baselines the itemized record produced 103 hits where 86.9 were expected; the exact probability of a record at least this strong, if each call's true chance were its printed baseline and the calls were independent, is approximately 1.9 in 10,000. Collapsing each market settlement day to a single trial, it is approximately 1.1 in 1,000. Under the harshest dependence treatment considered — the whole of election night counted as one observation as well — it is 0.091, and the itemized record does not by itself exclude chance at the conventional level; that bound is printed in the paper at the same size as the headline figure. The method that generates the calls is proprietary and is not disclosed, by design: the record is built so that it can be checked without it, because every seal precedes its outcome and every grade names its source. What the record means is left open. What the record is, is stated exactly. Keywords: forecasting; pre-registration; prediction ledger; sealed predictions; calibration; binomial test; Aldred Industries; prediction desk. ---

Zenodo (CERN European Organization for Nuclear Research)
Forecasting Techniques and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon