Settlement Source Documentation
AI Disruption Prediction Market Contracts
Settlement Methodology and Data Conventions
1.1
Settlement architecture
PYMNTS Intelligence is the settlement source for the AI Disruption contract series listed on a CFTC-registered designated contract market (the “Exchange”). PYMNTS Intelligence delivers verified settlement values; it does not operate the trading venue and does not define or trigger settlement. Every contract settles on an aggregate reading across a survey base or disclosure universe.
| Source | What it measures | Settlement use examples | Scale & cadence |
|---|---|---|---|
| Consumer Monitor | U.S. adult behavior across AI use, payments, commerce, finance, health, information consumption | Share of consumers reporting specific behaviors on a previous-day (24-hour) recall window | n=4,000 monthly, weighted to the U.S. adult population (American Community Survey) |
| Enterprise Monitor | AI adoption, deployment depth, ROI, agentic permissions, workforce expectations and sentiment among senior decision-makers | Share of U.S. enterprises reporting specific positions or outcomes | n=500 quarterly with industry sample quotas |
| Corporate Disclosure Monitor | Publicly disclosed quantitative facts: SEC filings, earnings releases/transcripts, official corporate communications | Counts and values aggregated across a company universe (e.g., share of universe companies making a specified disclosure) | Continuous monitoring of fixed, pre-registered company universes; tied to the disclosure calendar |
Principles common to all three streams
- Independence. PYMNTS takes no positions on contracts its data settles; compensation from the Exchange is not contingent on outcomes; all PYMNTS Intelligence employees are subject to a total trading embargo on these contracts and related instruments.
- Reproducibility. Every settlement value is documented with raw inputs, methodology, and calculation; records retained at least seven years per CFTC standards.
- Dual-researcher verification. Every settlement determination is made independently by two researchers and must match before settlement advances.
- Pre-registered methodology. Exact question wording, response options, sample frame, denominator, threshold, and rounding rule publish when the contract lists and never change during the life of the contract.
- Transparent documentation. Every contract carries a public methodology page — settlement source, threshold, calculation, and audit history — published when the contract lists.
1.2
Consumer Monitor
| Parameter | Specification |
|---|---|
| Population | U.S. adults 18+, no regional exclusions; English-only fielding (a disclosed limitation) |
| Sample | ~4,000 qualified completes per monthly wave; online panels — primary provider plus secondary top-up; ISO 20252-compliant providers, 1M+ active U.S. panel minimum |
| Design | Independent cross-sectional sample each wave (not a longitudinal panel); behavioral settlement items use previous-day (24-hour) recall; wave aggregates represent behavior across the month |
| Fielding | Collects over the full calendar month; the collection window closes at month-end |
| Weighting | Raking (iterative proportional fitting) on age, gender, race/ethnicity, education, household income, Census region; reference = most recent American Community Survey (ACS) 1-year estimates, updated annually effective the first wave after each release; weights trimmed to [0.2, 5.0], truncations logged; bounds reviewed if >2% of weighted sample trims |
| Quality controls | Applied in order, any single failure excludes: speeding (completion time below a pre-specified fraction of the wave’s median completion time); straight-lining (same response across all items in 2+ batteries of 5+ items); pre-specified consistency-pair review; open-text screening (verbatim question echoes, minimum-length and nonsense-pattern checks). Exclusion counts logged per wave and reviewed before settlement |
| Settlement value | Weighted percentage of the relevant population reporting the pre-registered behavior, computed independently by two researchers |
| Known limitations | Online-panel coverage; self-report; cross-sectional month-over-month variance; English-only fielding |
1.3
Enterprise Monitor
| Parameter | Specification |
|---|---|
| Population | U.S. enterprises, answered by senior AI decision-makers: the primary AI decision-maker or most-knowledgeable person on AI strategy; involved-but-non-influencing respondents screened out |
| Sample | 500 qualified enterprise respondents per quarterly cycle, via B2B panel specialist (ISO 20252; title/company verification through panel registration data) |
| Distinctives | Decision-maker-level data; measurements with no public benchmark (agentic permissions, autonomous payment authority) |
| Denominators | Conditional questions settle on the conditional subsample by default; denominator specified explicitly in every pre-registered methodology; achieved base n published; conditional contracts carry a pre-registered no-settlement floor |
| Quality controls | Identity/role verification, panel-history and IP-geolocation fraud checks, B2B-adapted speeding and straight-lining, internal consistency checks |
| Known limitations | B2B panel selection effects; self-reported strategy and social-desirability bias; smaller conditional subsamples (disclosed) |
1.4
Corporate Disclosure Monitor
The Corporate Disclosure Monitor is a structured-extraction program that reads primary corporate disclosures and produces verified, fully sourced figures. Three governing principles: (1) the automated extraction produces evidence, the research team decides — it never proposes, scores, or resolves a contract; (2) every figure is sourced from a primary document, never estimated or drawn from model training data; (3) only public primary sources are cited.
| Element | Specification |
|---|---|
| Source tiering | Tier 1 (only citable): SEC filings, issuer IR disclosures, earnings transcripts/releases, official corporate communications. Tier 2/3: discovery only, never a cited figure. Blocked entirely: analyst reports, Wikipedia, Crunchbase and other non-tiered secondary sources |
| Source authority | Simultaneous official disclosure governs over later filings (press-release value over 10-Q re-print); filed GAAP line items govern balance-sheet/cash-flow metrics; company statements over aggregators; each contract names its governing source, incl. capex basis, finance-lease treatment, fiscal-quarter mapping, before listing |
| Detection | SEC EDGAR polled daily (XBRL company facts, submissions, full-text search, by CIK against the fixed universe); issuer IR and wire services monitored on pre-announced earnings dates; workforce-event trackers and WARN filings monitored weekly for discovery only |
| Extraction | Structured XBRL parsed with per-contract US-GAAP concept tags; press releases via per-issuer format profiles; citation counts via pre-registered pattern definitions applied uniformly |
| Redundant extraction check | Every figure extracted twice by independent automated runs. Tolerance: 0.5% for dollar/percentage figures; exact match required for counts. Disagreements resolve to the research team against the source document. Quality control on figures only — never determines outcomes |
| Adjudication | Task 1: two analysts confirm the figure and governing source. Task 2: two analysts independently apply the threshold for the YES/NO outcome without seeing each other’s conclusion; match advances, mismatch reconciles. The automated system never produces the outcome |
| Threshold-proximity rule | Verified figure within 2.0% of the threshold (4× extraction tolerance) escalates to heightened review: source-authority scrutiny, documented rationale, third reviewer. The 2.0% band is the default and may be adjusted per contract at pre-registration |
| Provenance | Source URL and accession number, filing date, concept tag, short verbatim quote, UTC timestamp, extraction-check result; append-only source-call log; point-in-time capture with source archival |
| Fallback & failure modes | Manual fallback replicates the full procedure whenever automated detection fails; source-format changes are flagged, never silently resolved, with disposition per the Exchange’s rules; post-disclosure amendments logged but never reopen settlement; universe companies that do not disclose within the contract window are excluded from that contract’s settlement |
1.5
The six-stage settlement workflow
Applies to all three sources.
| # | Stage | What happens · Control |
|---|---|---|
| 01 | Acquisition | Wave closes, or automated extraction (or manual fallback) produces the verified figure with provenance · sample/data hits target; anomalies flagged |
| 02 | Determination | Two researchers independently reach the settlement determination · must match; reconcile until both agree |
| 03 | Verification | Cross-check against secondary indicators (context only — secondary sources are never a settlement input); threshold-proximity escalation for disclosure-settled contracts · verification log; anomalies resolved or escalated |
| 04 | Approval | Head of Research confirms independent matching determinations and signs off (does not compute the value — separation of duties) · sign-off timestamped |
| 05 | Delivery | Verified value transmitted to the Exchange under embargo · receipt confirmed; delivery logged |
| 06 | Release | Settlement value and the contract’s methodology page are made public following settlement · public timestamp reconciled against delivery log |
Pre-release data handling: between internal approval and public release, the settlement value is known only to a minimal, named trust boundary: the research staff performing determination and verification, the Head of Research, the Exchange’s designated delivery contact, and compliance personnel with audit responsibility. Commercial and executive leadership, marketing, and editorial staff have no pre-publication access. The value remains under embargo until its public release following settlement — and because every PYMNTS Intelligence employee is barred from trading these contracts at all times, no one inside the settlement source can act on a value at any stage, public or not.
Finality: the delivered value is the settlement value. Corrections and amended filings are logged in the audit trail but never reopen or re-settle a contract.
1.6
Input integrity: why the settlement inputs resist manipulation
The historic failure mode of submission-based benchmarks was economically interested submitters: parties who supplied the inputs also held positions on the outputs. The inputs to these contracts are structurally different. Survey respondents have no economic stake in any settlement outcome: no respondent is told which survey items settle contracts, and the Consumer Monitor draws an independent cross-sectional sample of roughly 4,000 adults each wave, so no respondent or coordinated group holds a persistent position in the instrument. Question wording is public once a contract lists, but a single response moves a settlement value by hundredths of a percentage point — influencing an outcome would require corrupting a meaningful share of thousands of independently recruited respondents, undetected by the quality-control screens in Sections 1.2 and 1.3. Enterprise Monitor respondents are role-verified senior decision-makers subject to the same logic at quarterly cadence. Disclosure-settled contracts rest on public documents filed under the securities laws, where misstatement carries its own legal consequences independent of this program. The residual conflict surface — the administrator itself — is governed by the controls in Sections 1.5 and 1.7.
1.7
Independence, conflicts, and change control
- Financial independence. Compensation is for methodologically sound data production, never contingent on which way contracts settle; no paid contract design favoring outcomes.
- Total trading embargo. All PYMNTS Intelligence employees and company executives — not only the research staff who touch settlement data — are prohibited from trading any contract that settles on PYMNTS data, plausibly correlated contracts, and securities of companies named in disclosure-settled contracts, with annual written certification. The embargo applies at all times, including while collection windows are open and contracts are trading.
- Editorial separation. PYMNTS Intelligence is operationally independent of PYMNTS.com’s editorial, advertising, and sponsorship businesses. Editorial staff have no pre-release access to settlement values, and research staff who produce settlement determinations do not write topical commentary on live contracts. Methodology decisions are made by the research team without advertiser or sponsor input.
- Design discipline. Instruments are designed for best measurement, never for anticipated contract outcomes; disclosure source-authority rules are set before listing.
- Methodology change rules. No change between listing and settlement. If unavoidable (e.g., panel vendor failure), the affected contract settles on the prior methodology one final time; the change takes effect next wave under Head of Research authority. Settled contracts are never reopened.
- Rounding. Half-up rounding at two decimals, uniform across all contract types.
These conventions describe how settlement values are produced. They apply uniformly to every contract, in every cycle. Per-contract settlement details — exact question wording, threshold, denominator, and governing source — publish on each contract’s methodology page when the contract lists. This page governs the production of settlement values only: trading hours, contract close, settlement execution, and the handling of open interest are governed by the Exchange’s rules and each contract’s listed terms, which are the authority on those matters. Cycle cadences shown reflect the current contract types; a new contract type publishes its own cycle terms at listing.
2.1
Cycle conventions
| Contract type | Collection window | Verified settlement value | Publication |
|---|---|---|---|
| Consumer Monitor | The full calendar month the contract is named for | Within 2–4 business days following window close | Public following settlement |
| Enterprise Monitor | The quarterly cycle stated in the contract’s listed terms | Within 2–4 business days following window close | Public following settlement |
| Corporate & Market Disruption settled by the Corporate Disclosure Monitor | The full calendar quarter | Within 2–4 business days following window close | Public following settlement |
Verified settlement values are delivered to the Exchange, and settlement of the associated contracts occurs per the Exchange’s settlement process. Delivery mechanics between PYMNTS Intelligence and the Exchange are governed by the parties’ agreement.
2.2
The verification interval
The interval between window close and delivery is reserved for data processing, quality-control exclusions, dual-researcher verification, and internal approval. The settlement value is fixed entirely by the data collected within the closed window. If a value cannot be verified on schedule, delivery is held and the Exchange is notified — the verification standard is never relaxed to meet the timeline.
2.3
Finality and corrections
The delivered value is the settlement value, and it is final once settlement occurs. Corrections, restatements, and amended filings disclosed after delivery are recorded in the audit trail but do not reopen or re-settle any contract. This first-print-final convention matches the fixing conventions of established data-settled markets.
2.4
Thresholds, relisting, and rotation
- Every contract’s question wording, response options, sample frame, denominator, threshold, and rounding rule lock at pre-registration, before the contract lists, and publish on its methodology page.
- Recurring series relist each cycle with thresholds set at that cycle’s pre-registration; strike ladders re-center each cycle. No threshold mechanically adjusts off its own prior readings.
- Slate changes occur only at cycle boundaries. A contract collecting in an open window always settles under the rules it listed with.
2.5
Seasonal and conditional series
- Series measuring school-year behaviors list only during the school year (September–May), with recall windows valid in every listed month; the underlying series fields year-round so the measurement never gaps.
- Series measuring annual-cycle behaviors (e.g., tax filing) use a 90-day recall window and a semiannual cadence rather than a daily-recall item that breaks out of season.
- Contracts settling on conditional subsamples publish the achieved base n each wave and carry a pre-registered no-settlement floor: if the settlement base falls below the floor, the contract voids and open interest is handled per the Exchange’s process.
2.6
The disclosure universe
Corporate & Market Disruption contracts
The company universe for each disclosure-settled contract is fixed at cycle open per a versioned universe registry. Only companies that disclose within the contract window enter that contract’s settlement calculation; universe companies that do not disclose within the window are excluded. Companies whose fiscal quarters do not align to the calendar quarter settle per the fiscal-quarter mapping locked in the contract’s pre-registered definition.
2.7
A worked example (hypothetical)
Illustrative only · Not a listed contract
The figures below are illustrative only; they belong to no listed or planned contract. Consider a hypothetical monthly contract: “Will 15.00% or more of U.S. adults report Behavior B in the prior 24 hours?” — with the question wording, response options, sample frame, denominator, threshold, and rounding rule locked and published at pre-registration before listing.
- Window closes. The wave collects over the full calendar month; the collection window closes on the final calendar day, with 4,012 qualified completes.
- Quality control. 118 responses are excluded by the pre-specified screens (speeding, straight-lining, consistency, open-text), leaving an analytic sample of 3,894. Exclusion counts are logged and reviewed for anomalies.
- Independent determination. Two researchers, working separately, each compute the weighted share of U.S. adults reporting Behavior B from the analytic sample. Both obtain 15.7346%. Had the two values differed, the discrepancy would be reconciled to a documented cause before anything advanced.
- Rounding and threshold. Half-up rounding at two decimals gives a settlement value of 15.73. Because 15.73 ≥ 15.00, the contract resolves YES. The margin is not close in this example; a disclosure-settled contract whose verified figure fell within 2.0% of its threshold would additionally trigger heightened review with a third reviewer.
- Approval, delivery, release. The Head of Research confirms the determinations were independent and matching, and signs off. The value 15.73 is delivered to the Exchange under embargo; contracts settle per the Exchange’s process; the value and its methodology page are then made public. The settlement is final: any later correction is logged but does not reopen it.
This page governs the production of settlement values only. Trading hours, contract close, settlement execution, and the handling of open interest are governed by the Exchange’s rules and each contract’s listed terms.