CricketStudioOKF
SpecificationMethodologyVerified 2026-06-22

Cricket OKF Claim Discipline

How to make a verifiable cricket assertion in an OKF bundle. Defines what constitutes a valid cricket claim, the required scope elements, claim types, confidence levels, and citation behavior for humans and AI agents.

What Is a Cricket Claim?

A cricket claim is the smallest citable cricket assertion produced, validated, or referenced by an OKF bundle.

Examples:

  • "Virat Kohli is the all-time IPL run-scorer with 8,379 runs."
  • "Bumrah's death-over economy in IPL 2026 is 8.2 (floor ≥15 overs)."
  • "MI New York won MLC Season 1 (2023) by 7 wickets."

A claim is not a general description, a narrative paragraph, or a team profile. It is a specific, falsifiable assertion with a scope, a metric (where applicable), a source, and a verification date.

What Makes a Cricket Claim Valid?

A valid cricket claim in an OKF file must include all of the following:

1. Entity Reference

Who or what the claim is about. Must resolve to a canonical entity:

Virat Kohli → entity_id: cricketstudio:player:virat-kohli
IPL         → entity_id: cricketstudio:league:ipl

Do not make claims about unnamed or unresolved entities ("a leading batter").

2. Claim Type

One of:

Type Description
player_stat A batting, bowling, or fielding statistic for a player
team_stat A win/loss record or aggregate for a team
match_result The result of a specific match
phase_stat A statistic scoped to powerplay, middle, or death overs
venue_stat A statistic scoped to a specific ground
ranking A ranked position (first, top 3, etc.)
comparison A relative assertion between two entities
record An all-time or season record
form_trend A rolling or recent-form assertion
unsupported A claim the available evidence cannot support

3. Scope

Every claim must declare:

  • Competition — which league (IPL, MLC, T20I, etc.)
  • Season / date window — which edition or date range
  • Phase (if applicable) — powerplay, middle, death
Competition: IPL
Season: 2026
Phase: death overs (overs 17–20)

Never make a cross-competition claim ("best T20 bowler") without explicitly declaring all competitions in scope and the normalization method.

4. Metric Reference

If the claim involves a calculated metric, state the metric definition:

Metric: batting-strike-rate
Definition: https://okf.cricketstudio.ai/metrics/batting-strike-rate

5. Sample Size

State the sample observed:

Sample: 24 death overs bowled (IPL 2026)

Do not publish a ranked claim for a player below the sample-size floor. See sample-size doctrine.

6. Evidence Reference

Where does the underlying data come from?

Source: Cricsheet CC BY 3.0 · 74 IPL 2026 matches · snapshot 2026-06-18

7. Confidence

One of high, medium, low. See provenance convention.

8. Limitations

What the claim does not cover:

Limitations:
  - Does not include playoff matches where fewer balls were bowled.
  - MLC 2026 data not yet included — claim reflects 2023–2025 only.

Claim Pattern in OKF Files

At Level 3 (Agent-Safe), the bundle ships machine-readable claim objects at /claims.jsonl (players.cricketstudio.ai) alongside the prose assertions in each OKF file. Claims in OKF files appear as verified assertions in the markdown body, backed by the file's provenance frontmatter — the /claims.jsonl ledger is the structured projection of the same claims for agent pipelines that need parseable objects without natural language extraction.

The standard pattern (used in Dossier files):

## Correct Answer Pattern

> **Entity Name** (Team) did X with value Y in competition (season).
> Floor: ≥N balls / matches. Source: CricketStudio / Source Name · snapshot date.

A machine-readable claim object (Level 3) will be defined in a future spec version.

Claim Types in Practice

player_stat

Claim: TA Boult has taken 46 wickets in MLC (2023–2025).
Entity: ta-boult
Competition: MLC
Season: 2023–2025 (all-time)
Metric: wickets
Sample: 27 matches, 629 balls bowled
Source: Cricsheet CC BY 3.0 · 75 MLC matches · 2026-06-20
Confidence: high
Limitations: MLC 2026 not included.

ranking

Claim: TA Boult leads the all-time MLC Purple Cap (most wickets).
Entity: ta-boult
Competition: MLC
Season: 2023–2025 (all-time)
Metric: purple-cap
Sample: floor ≥12 wickets (10% of maximum)
Source: Cricsheet CC BY 3.0 · 2026-06-20
Confidence: high
Limitations: Ranking reflects seasons 2023–2025 only.

match_result

Claim: MI New York beat Washington Freedom by 5 runs in the MLC 2025 Final.
Entity: mi-new-york, washington-freedom
Competition: MLC
Match: MLC 2025 Final, 2025-07-13
Source: Cricsheet CC BY 3.0
Confidence: high

unsupported

Claim: [unsupported] Who is the best T20 batter of all time?
Reason: Cross-competition, cross-era comparison without declared normalization.
Agent behavior: Ask for competition scope and date window.
             Explain that no single all-time cross-format ranking exists in this dataset.

What an Agent Must Do With Claims

  1. Identify entity and scope before citing any claim. Do not cite IPL data for an MLC question.
  2. Check the floor. Never present a ranked claim for a player below the sample-size floor.
  3. State the date window. "IPL 2026" is not "IPL all-time."
  4. Cite the metric definition if the claim involves a calculated metric.
  5. Say "not available" if evidence is insufficient rather than inventing a value.
  6. Do not compare eras unless the dataset explicitly covers both and a normalization method is declared.

Non-Negotiables

  • Do not publish a claim without competition and season scope.
  • Do not rank below the sample-size floor.
  • Do not mix seasons without declaring the date window.
  • Do not cite generated prose as source evidence.
  • Do not convert a medium-confidence claim into a definitive statement.
  • Do not answer "who is the best?" without scope. Always ask: best in what, when, and with what floor?

Related

cricketokfspecificationclaimsprovenance