Currentline Data · Methodology

How every number earns its place.

The Currentline Data Platform turns the OEB's public record into structured, queryable rows. For the figures that matter — a revenue requirement, an allowed ROE, a rate base — “structured” is not enough. Each one is defined, bounded, anchored to the governing OEB provision, reconciled by machine, and signed by a person before it appears anywhere. This page documents that discipline in full, in public, because trust is the product.

See the data itself on the Data Platform (an Enterprise add-on), or read the editorial standards behind the wire.

The review discipline

Every figure is corpus-checked AND human-reviewed before publication.

Corpus-check: every extracted number is reconciled against the governing OEB provision before it can be served. A figure that contradicts the corpus or is physically impossible is failed and never reaches a reviewer or a reader.
Human review: nothing numeric publishes on the strength of an algorithm alone. Each rate-case fact is read and confirmed by a person against its source filing before it appears.
Published surfaces serve ONLY figures that are both approved and not failed by the corpus check — pending and rejected figures never appear on any reader-facing surface or in the API.
Flag, not fail, where the rule allows variation: a utility is entitled to ask for — and the OEB can award — values that diverge from the generic. Those are flagged for human judgement, not silently dropped, so a genuine departure is surfaced rather than hidden.

Every rate-case figure is confirmed by a person against its primary source document before it publishes.

Case Ledger

Every OEB proceeding, projected from the source record.

Case Ledger (proceedings)

proceedings

Every OEB proceeding we observe on the public record, as one typed row — applicant, proceeding type, status, filed and decision dates, and a count of filings. Projected directly from the source listing; no value is inferred.

Plausibility band: Structural, not numeric — there is no band to apply. A proceeding either appears on the OEB record or it does not.
How it's reconciled: One row per proceeding. The filing count is a live COUNT of linked documents, not a stored aggregate.
Corpus anchor: OEB public proceeding record (EB-number registry) — source-of-record projection, no corpus reconciliation required.

Caveats

›Coverage tracks what the OEB publishes; a brand-new filing may take a refresh cycle to appear.
›Status reflects the latest observed state and can lag the OEB's own internal status by a day.

Filing Ledger

Every intra-case document, typed, dated, source-linked.

Filing Ledger (filings)

filings

Every intra-case document — applications, evidence, interrogatories, decisions and orders — typed and dated, each linked back to its proceeding and to its PDF on the public record.

Plausibility band: Structural, not numeric. Document dates are read from the filing itself where available, otherwise from the source listing.
How it's reconciled: One row per document. Nothing is summed or averaged; the ledger is a faithful index, not a model.
Corpus anchor: OEB public proceeding record (per-case document listings) — source-of-record projection.

Caveats

›A document's filed date is taken from its own letterhead where we can read it, and from the listing otherwise — the two can differ by a day.
›Document typing (application vs. evidence vs. decision) is heuristic and is corrected on review when miscategorised.

Entities

The utilities and parties on the record.

Entities (utilities & parties)

entities

The regulated utilities and parties that appear as applicants on the record — legal name, entity type, and service territory where known — with a count of their proceedings.

Plausibility band: Structural, not numeric.
How it's reconciled: One row per entity. Proceeding counts are live COUNTs against the proceedings table.
Corpus anchor: OEB public proceeding record (applicant identities) — source-of-record projection.

Caveats

›Name normalisation merges obvious variants of the same utility; an unmerged duplicate is corrected on review.
›Service territory is populated where the record states it and is otherwise left blank rather than guessed.

Ask vs. Award (rate-case facts)

Requested vs. approved revenue requirement, rate base, ROE — extracted, corpus-checked, human-reviewed.

Revenue requirement

revenue_requirement

The total annual revenue a utility is permitted to recover from ratepayers in the test year — the headline ask of a rate application and the headline award of a decision. Captured at both postures (requested and approved) so the two sit side by side.

Plausibility band: Annual-revenue scale. A positive figure is required; a value above $10B is flagged for a human to confirm the figure and its units. This band is SEPARATE from rate base — revenue requirement is a yearly flow, not a capital stock.
How it's reconciled: Per-figure, posture-aware. Each requested and approved value is captured and reconciled on its own; nothing is averaged across cases. The ask-vs-award pivot only pairs like-for-like figures.
Corpus anchor: Distribution: Filing Reqs Electricity Distribution ch.2 §2.6 / §2.6.1 RRWF (Exhibit 6: Revenue Requirement). Transmission: Filing Reqs Transmission ch.2 §2.0.1–2.0.2 (Revenue Requirement); summary §2.3.1(A). The applicable reference is selected by regime (distribution is the default and the majority; when the regime is unknown we assume distribution and say so).

Caveats

›Distribution and transmission revenue requirements are governed by different filing requirements and are never collapsed into one anchor.
›Cost-of-capital parameters do NOT anchor this metric — revenue requirement is a build-up of costs, not a return figure.
›When the test-year basis is uncertain (settlement-derived, or an IRM/price-cap index year rather than an adjudicated rebasing), a bare requested-vs-approved delta is not meaningful and is annotated rather than presented as a clean comparison.

Rate base

rate_base

The value of the utility's invested capital on which it is allowed to earn a regulated return — the asset base, not an annual flow. Captured at requested and approved postures.

Plausibility band: Capital-stock scale. A positive figure is required; a value above $50B is flagged for a human to confirm the figure and its units. This band is SEPARATE from, and larger than, the revenue-requirement band — the two are never reconciled against one another.
How it's reconciled: Per-figure, posture-aware. Each value stands on its own; no cross-case averaging.
Corpus anchor: Distribution: Filing Reqs Electricity Distribution ch.2 §2.2.1 (Exhibit 2: Rate Base). Transmission: Filing Reqs Transmission ch.2 §2.5.1 (Exhibit 3: Rate Base). Selected by regime, defaulting to distribution when unknown (and saying so).

Caveats

›Rate base (a capital stock) and revenue requirement (an annual flow) carry distinct bands and distinct anchors — confusing the two is a category error we explicitly guard against.
›Per-project leave-to-construct capital costs are a different animal and anchor to a different provision; they are not rate base.

Allowed ROE

allowed_roe

The return on equity a utility is permitted to earn on the equity portion of its rate base. The OEB publishes a generic ROE each year; an individual utility may apply for — and, in rare cases, be awarded — a different figure with justification.

Plausibility band: Plausible band 3–15%; outside this a value is failed as almost-certainly a mis-extraction. Inside the band, a value within 0.05 percentage points of the published OEB-generic ROE for its year passes as "identical modulo rounding"; any larger divergence is flagged for human judgement.
How it's reconciled: Reconciled against the OEB-generic ROE for the figure's decision year, posture-aware. A REQUESTED divergence from the generic is routine and merely flagged for confirmation; an APPROVED divergence means the OEB actually granted a departure from its generic methodology, and the flag says so explicitly so a reviewer verifies the departure in the decision.
Corpus anchor: Cost of Capital 2026 parameters (letter EB-2025-0303, via EB-2024-0063 generic methodology); Filing Reqs Electricity Distribution ch.2 §2.5

Caveats

›We hold the OEB-generic ROE for 2025: 9%, 2026: 9.11%. With no published generic on file for a figure's year, we flag rather than silently pass.
›A divergence from the generic is a flag, never a fail — utilities are entitled to apply for a different ROE, and the OEB can award one. The flag asks a human to confirm; it does not assert the number is wrong.
›The 0.05pp match tolerance means "identical to the generic modulo rounding," not "close enough" — it exists only to absorb display rounding.

Deemed equity %

deemed_equity_pct

The common-equity share of the OEB's deemed capital structure used to set the allowed return — the regulator's standard is 40% common equity (with 56% long-term and 4% short-term debt).

Plausibility band: Must be strictly between 0% and 100%; outside that it is impossible and is failed. A value within 0.5 percentage points of the deemed 40% passes; a larger divergence is flagged.
How it's reconciled: Per-figure, reconciled against the deemed structure. No cross-case averaging.
Corpus anchor: Cost of Capital 2026 parameters (letter EB-2025-0303, via EB-2024-0063 generic methodology); Filing Reqs Electricity Distribution ch.2 §2.5

Caveats

›The deemed structure is 40% common equity; a utility may seek a different ratio with evidence, so a divergence is a flag, not a fail.

Rate / bill increase

rate_increase_pct

The percentage change in rates or a typical customer's bill that a filing requests or a decision approves.

Plausibility band: Hard band -50% to +100%; outside it the value is failed. The MEANINGFUL flag threshold is greater than 10% — this is the §2.8.12 per-class rate-mitigation trigger, not an arbitrary "large but possible" cut-off. A class increase above 10% is the point at which the OEB expects a mitigation plan, so it is exactly the figure a reviewer should see.
How it's reconciled: Per-figure, posture-aware. Each requested and approved percentage is reconciled on its own.
Corpus anchor: Filing Reqs Electricity Distribution ch.2 §2.8.11 + §2.1.2(H); §2.8.12 (≥10%/class rate-mitigation trigger)

Caveats

›This metric is anchored to the bill-impact / rate-mitigation provisions, NOT to cost-of-capital — a rate increase is an outcome of the whole application, not a return parameter.
›A bill impact varies sharply by customer class; a system-average figure can mask a much larger increase for a single class. The >10% flag is the per-class mitigation trigger for precisely this reason.

Utility scorecards

OEB-published performance indicators, transcribed as reported.

Utility scorecards

scorecard

OEB-published distributor performance indicators — reliability (SAIDI, SAIFI), customer-service percentages, and financial ratios (achieved/deemed ROE, debt-to-equity) — by utility and reporting year, as the OEB itself reports them.

Plausibility band: Physical-sanity only. Because the OEB publishes these values directly, the gate exists to catch a parse error, not to second-guess the regulator: SAIDI/SAIFI must be non-negative and within a wide reliability range; percentages within 0–100%; achieved ROE may legitimately be negative in an under-earning year; an unusual debt-to-equity (e.g. from near-zero or negative equity) is flagged for a human rather than dropped.
How it's reconciled: One row per utility, metric and reporting year, transcribed as published. No averaging — the values are the OEB's, not ours to recompute.
Corpus anchor: OEB RRR / Electricity Distributor Scorecard framework. Honest note: the document that DEFINES the scorecard framework sits in the OEB's Reporting & Record-keeping Requirements and is outside our rate-filing corpus slice, so we do not represent the bands as corpus-grounded beyond physical sanity. The in-slice provision that REQUIRES this performance data is Filing Reqs Electricity Distribution ch.2 §2.1.7 (Performance Measurement).

Caveats

›Plausibility bands here are physical-sanity checks only; they are not derived from the scorecard's defining document, which is not in our rate-filing corpus slice.
›Achieved ROE is an outcome and can be negative; it is not the deemed return target and is never reconciled against the cost-of-capital generic.
›We transcribe the OEB's published figures rather than recomputing them — a discrepancy means a parse error to fix, not a regulator to correct.

Coverage & scope

Cited to the indexed OEB record

v2026.05 · as of May 31, 2026

Currentline indexes the publicly posted OEB electricity proceeding record — proceedings, intra-case filings, the parties on the record, the rate-case figures we human-review, and the OEB-published utility scorecards. Coverage tracks what the OEB publishes; a brand-new filing may take a refresh cycle to appear.

What the indexed record excludes

—Confidential filings, and the confidential portions of otherwise-public documents.
—Personal-information redactions (FIPPA-based) made to public versions of filed documents.
—Non-relevance redactions made to public versions of filed documents.
—Settlement-conference materials, position papers and negotiations — these are not part of the OEB record by rule.
—Enforcement proceedings, which run on a separate procedural track — they have their own public record but are outside Currentline's current coverage scope.

Figures sourced from public filings may reflect redacted or summarized evidence where the OEB granted confidential treatment.
Settlement-derived figures reflect the settlement proposal as filed — not the underlying negotiations or any confidential supporting data.
Deferral- and variance-account figures are read from decision text; the Accounting Procedures Handbook (Uniform System of Accounts) that governs account definitions is outside the indexed corpus and should be consulted for accounting-grade use.

How we cite a source

Every datum links to its primary source — the case decision or, for values set by a generic OEB order, that order.

A note on leave-to-construct costs

Leave-to-construct project costs

Per-project transmission leave-to-construct capital-cost facts are reconciled against their own filing provision — never against revenue requirement or rate base. We surface them only as project-specific figures.

filing-req-transmission-ch4-leave-to-construct §4.3.2.6 (Project Costs)

Provenance & corrections

Every figure traces back to its filing on the public OEB record, with a source PDF and an approximate page. Structural rows (proceedings, filings, entities) are faithful projections of that record — we do not infer them.

Where the OEB's rule allows a utility to diverge from a generic value — a different ROE, a different equity ratio — we flag the divergence for human judgement rather than dropping the figure. A flag asks a reviewer to confirm; it never asserts the regulator was wrong.

If you believe a published figure is mistaken, tell us and we will check it against the source and correct the record. Accuracy is not a feature here — it is the whole proposition.