Forward Ledger

The design of this page is a time-stamped archive of every snapshot the pipeline publishes, followed by the realised outcome once each one-month window closes. What it actually holds is one snapshot and no outcome, and it says so.

Retired programmeHistorical snapshot, not a current signal. Model outputs only; not investment advice.
Status
Retired after the last run
Last run
2026-04-29 · V12
Evaluation window
2026-04-29 → 2026-05-29 (closed)
Realised outcome
Not computed; no evaluation published

Sample: 30 U.S. mega-caps; training lookback about three years of daily data; walk-forward out-of-sample panel 2024-09-30 → 2026-03-31 (19 months); prediction horizon one month. The page date is not a model date. Definitions and provenance of every metric are on the Research Validation page, and the unedited output file is published at /data/signals.json.

Snapshot register

PublishedPipelineWindowRealised outcomeStatus
2026-04-29V122026-04-29 → 2026-05-29not computedRetired after this run
2026-04-27V12superseded 2026-04-29not computedSuperseded, not archived on the site

The tracker that was meant to append a realised-versus-estimate row after each window never ran on the 2026-04-29 snapshot. That is a gap in the process, recorded here rather than filled with a number computed after the fact. No synthetic history is backfilled.

The 2026-04-29 snapshot, top five by rank score

TickerSectorTier1-Mo estimateStrategyRank scorePublished
INTCTechnologyHIGH+4.2%OTM Call 20952026-04-29
AMDTechnologyHIGH+4.2%ATM Call952026-04-29
ORCLTechnologyHIGH+2.8%OTM Call 20932026-04-29
AVGOTechnologyHIGH+2.4%ATM Call902026-04-29
PLTRTechnologyHIGH+2.4%Bull Spread872026-04-29

Rank score is the model rank score (30–95, monotone in the estimate; not a probability): the percentile rank of the 1-month estimate within the 30-name cross-section, floored at 30 and capped at 95, computed on the unrounded estimate. It is a rescaling of the estimate's rank, carries no calibration, and is not a confidence. The walk-forward IC on the Validation page is the measure of what the ranking is worth. The field is stored as ec in the published file.

Full 30-name snapshot on the Lab page.

How the ledger is designed to work

01

Publish

A run publishes a full snapshot of 30 estimates with tiers and options structures, together with its validation record.

02

Freeze

The snapshot is frozen with a date-stamped filename. Published outputs are not modified retroactively; the file the site renders is the run output, unedited.

03

Track

After the one-month window, a tracker compares the snapshot with realised returns and appends a row. This step did not run for the only snapshot.

04

Record

Realised versus estimate is published here with its caveats attached. Nothing is published until the window has closed and the row exists.

Hypothetical implementation rules for any tracked outcome

These rules define how a realised outcome would be computed. They have not been applied to any published outcome, because none exists. The stock leg carries no execution costs; the options leg would use the half-spread-per-leg estimate from the chain at publication.

Rebalancing cadencePer snapshot (intended bi-weekly)
Position sizingEqual-weight within tier
Stock-leg costsNone (pre-cost)
Options-leg costsHalf-spread per leg, chain at publication
BenchmarkSPY total return
Options assumptionHeld to expiry, modelled

Limitations

This ledger describes hypothetical implementation of model outputs, not actual trading. Hypothetical results cannot account for liquidity, market impact or operational constraints. No outcome has been computed for any snapshot. All outputs are for educational and research purposes only; not investment advice. Full disclaimer.