 SCORECARD ARENAS DISCOVERY FOUNDATIONS FOR AGENTS 

The One Lab is an autonomous research institution. Its models make predictions about housing, markets, and industry — in public, before the answers are known — and every prediction is scored against what actually happens.

Forecasts locked before outcomes are known. Graded against reality. Every miss public.

The honest number is always small. That is the point.

FORECASTS LOCKED 30 written before their outcomes existed

GRADED 0 first verdicts due late Oct 2026

MISSES PUBLISHED insufficient data nothing has been graded

MISS RATE insufficient data denominator is zero — no rate exists over an empty set

Nothing has been graded yet. The forecasts above are locked and waiting on the official figures they will be judged against, so there is no accuracy number on this site — and there will not be one until the first of them matures.

Most research asks you to trust the author. Ours asks you to check the ledger.

02 / LIVE SCORECARD

The scorecard is the product.

Backtest and forward results are never averaged together. The seam below is the boundary: nothing under it was fitted.

{
  "endpoint": "/v1/scorecard",
  "separation": {
    "backtest": "excluded_from_track_record",
    "forward": "graded"
  },
  "forward": [
    {
      "arena": "Housing / United States",
      "locked": 30,
      "graded": 0,
      "misses": null,
      "summary": "awaiting_first_grade"
    },
    {
      "arena": "Housing / Canada",
      "locked": 0,
      "graded": 0,
      "misses": null,
      "summary": "no_forecasts_yet"
    }
  ],
  "backtest": {
    "graded": null,
    "rates": "not_published"
  }
}

BACKTEST · MIGRATED FROM v1 · FITTED AFTER THE FACT · NOT A TRACK RECORD

These were fitted against data that already existed, so they are not evidence that a forecast made in advance will be right. Their completeness is unmeasured, which is why counts appear below and no error rate does — not because the rate is unflattering, but because its denominator is unestablished and the figure would not mean anything.

INSUFFICIENT DATA

No migrated backtest history could be read.

 SEAM · NOTHING BELOW THIS LINE WAS FITTED FORWARD RECORD BEGINS 2026-08-04 

FORWARD · LOCKED AT ISSUANCE · GRADED ON RESOLUTION

ARENA	LOCKED	GRADED	MISSES	SUMMARY
Housing / United States	30	0	—	AWAITING FIRST GRADE
Housing / Canada	0	0	—	NO FORECASTS YET

MISSES ARE COUNTED IN THE SAME COLUMN SET AS PASSES. NO ARENA IS EXCLUDED FOR UNDERPERFORMANCE.

EVERY VINTAGE, ONE PER ROW →

03 / THE TRUTH ENGINE

Five steps, each leaving an artifact you can check.

A forecast that cannot be checked is an opinion. Every stage below writes a record you can read back; the grade is arithmetic on records that already exist.

ST-01

Issue

A model writes a forecast with an explicit resolution date, interval and data cutoff. Nothing is published yet.

forecast.json

ST-02

Lock

The forecast is written to a table that cannot be edited or deleted, and the database stamps the time. Neither we nor anyone else can change the content or backdate the stamp afterwards.

write-once row + db timestamp

ST-03

Publish

The locked vintage appears on the arena page and in the API on the same second, with its limitations attached.

vintage page + /v1

ST-04

Realize

The primary source releases the outcome. We record the release, the revision window, and the exact series used.

observation.json

ST-05

Grade

Error is arithmetic on two files that already existed. A miss publishes with the same prominence as a pass.

grade.json

04 / ARENAS

One arena issuing. One standing by. The rest say so.

An arena appears here because the Lab intends to cover it. It carries figures only once it has issued forecasts — the four marked in development have no pipeline running and no numbers to show, and saying so is the point of the list.

AR-01 LOCKED · AWAITING FIRST GRADE

Housing / United States

House-price forecasts for the United States and the nine Census divisions, built on the FHFA Purchase-Only index through ALFRED's real-time vintage archive. Issued on each publisher vintage at three, six and twelve months, locked before the outcome exists.

 30 locked · 0 graded 

AR-02 NO FORECASTS YET

Housing / Canada

House-price forecasts for Canada and eight metropolitan markets, built on Statistics Canada's New Housing Price Index. The archive is being re-collected under stricter rules about recording when each figure was published; no forecast is issued until that finishes.

 No forecast locked yet — first lock not yet scheduled 

AR-03 IN DEVELOPMENT

Quant

Planned. Cross-sectional equity signals graded on realized forward returns rather than backtested equity curves. No signal ships until every trial is logged including discards, and no public figure exists yet.

 No forecasts issued — this arena is not running 

AR-04 IN DEVELOPMENT

Financial Stability

Planned. Systemic stress indicators for banking and credit channels, published as model-family ranges rather than point estimates. Nothing has been issued.

 No forecasts issued — this arena is not running 

AR-05 IN DEVELOPMENT

Geospatial

Planned. Land-use and infrastructure change graded against subsequent imagery and permits, using pretrained open foundation models as backbones. Nothing has been issued.

 No forecasts issued — this arena is not running 

AR-06 IN DEVELOPMENT

Mine Design

Planned. Pit and stope economics as stochastic distributions with assumptions exposed, graded against realized costs. Nothing has been issued.

 No forecasts issued — this arena is not running 

05 / FOR AGENTS

Machine-readable by construction.

Every graded number on this site has a stable anchor, a schema.org record, and a JSON manifest at the same path. Agents read the same ledger humans do — no summary layer in between.

The v1 API has not been cut over yet. It still serves the design fixture set and stamps sample:true on every payload, so it does not yet contain the locked forecasts shown above. Read those from the vintage pages until it does.

schema.org/Dataset · Claim · Observation
stable anchors: /vintages/<vintage-uuid>
calibration credential: signed, re-verifiable

GET /v1/vintages JSON

Every locked forecast, paged, filterable by arena and grade.

GET /v1/vintages/{id} JSON

One forecast's full life: issue, locked payload, realization, grade.

GET /v1/calibration JSON

Reliability curves and interval coverage per arena and horizon.

MCP server · tools: list_vintages, get_grade, get_manifest

The One Lab · Toronto · Every number on this site resolves to a primary source or a graded model output. Where it can't, we say so.

 FOUNDATIONS ↗ DESIGN SYSTEM ↗ OBSERVATORY · DIRECTOR ↗ READ FROM THE RECORD AT BUILD TIME 