The Standard
The Promise Tracker Standard
The rubric behind every verdict, published in full. Two readers applying this standard to the same filings must reach the same verdict; that is the test it is written to pass.
Standard v1.0 · 2026-07-04 · canonical
§1
Principles
- Mechanical. A verdict is an arithmetic comparison between a target the company stated and an outcome the company reported. No discretion at any step.
- Company-sourced on both sides. The target comes from a company disclosure; so does the actual. SpotValue never substitutes its own estimate for either.
- Cited. Every promise and every verdict carries a source reference to the filing it came from.
- Immutable history. Once a fiscal year’s promises and verdicts are extracted and validated, they are never edited by hand; corrections happen by re-extraction under a new standard version.
- Reproducible statistics. Hit rates are computed by fixed formulas, disclosed wherever the number is displayed.
§2
What qualifies as a promise
A promise is a forward-looking commitment that is specific, measurable, time-bound, and company-made in an official channel.
Excluded, always: aspirational statements without a measure (“world-class”, “well positioned”); third-party forecasts; boilerplate risk language; restatements of past facts; and market-wide commentary unless tied to a company-specific measurable target.
Separable commitments are recorded as separate promises: “FY25 production of 20–22 Mt at unit cost of $103–113/t” is two promises, not one.
§3
Eight categories
| Category | Covers |
|---|---|
| Production guidance | Output volumes, sales volumes, utilisation |
| Financial guidance | Revenue, EBITDA, NPAT, margins, unit costs, cost-out programmes |
| Capex | Capital expenditure envelopes, project budgets |
| Capital returns | Dividends, payout ratios, buybacks, capital-management frameworks |
| Strategic milestones | Project delivery, first production, rollouts, approvals sought |
| ESG and regulatory | Emissions targets, safety targets, rehabilitation, undertakings |
| M&A and divestment | Announced transactions and their stated terms and timelines |
| Recovery and turnaround | Explicit recovery commitments after downgrades, incidents, or losses |
§4
Five verdicts
| Mark | Verdict | Meaning |
|---|---|---|
| ○ | Open | Not yet due, or due but the resolving disclosure has not yet been published. |
| ✓ | Hit | Outcome met or beat the target under the rubric. |
| ✗ | Missed | Outcome fell short of the target on the adverse side. |
| ~ | Partial | Inseparable mixed outcome, materially-late delivery within grace, or materially reduced scope. |
| ⟳ | Reframed | Target restated, replaced, or withdrawn before resolution; never graded hit or missed, and linked to its successor. |
Terminal verdicts are final; a promise resolves exactly once.
§5
The mechanical rubric
Every promise records a target type and a direction at extraction; the verdict is a pure function of target and actual.
- Ranges. Within the band, or beyond its favourable bound, is a hit; beyond the adverse bound is a miss.
- Points. Compared within a stated tolerance: ±5% when the company said “approximately”, ±2% when it did not.
- Floors and ceilings. “At least X” and “no more than X” compare directly.
- Milestones. Delivered by the due period is a hit; up to six months late is partial; later, or quietly dropped, is a miss.
- Same basis. The comparison uses the basis the promise specified: statutory or underlying, constant currency, managed or equity share. If the company stops reporting that basis, the promise is reframed, visibly.
- Silence is an outcome. If the due period passes and no disclosure settles the promise within the following full reporting cycle, the verdict is a miss, recorded as not addressed.
§6
Fixed formulas
Open promises are excluded from every rate. Reframes are counted separately and never merged into the hit rate.
assessed = hit + missed + partial hit_rate_% = 100 × (hit + 0.5 × partial) / assessed strict_hit_% = 100 × hit / assessed reframe_rate_% = 100 × reframed / (assessed + reframed)
A rate is displayed only when at least five promises have been assessed; below that, counts are shown instead, because a percentage of two promises is noise. The formula accompanies the number wherever it appears.
§7
Disclosure and versioning
Historical promises in the dataset have been coded against this Standard retrospectively; the Standard was authored prior to coding. Both the rubric and per-promise sourcing are public, so any verdict can be verified by any reader.
The Standard is versioned. Changes that could affect a verdict (rubric, tolerances, formulas) require a version bump and a changelog entry, and every compiled ledger records the version it was computed under. A typeset, versioned PDF edition accompanies public launch.