OilPriceAPI Research

Energy-data experiments, with the stop rules written first.

These pages document what is hard about each dataset, how the first probe works, what current evidence supports, and which measurements will determine the next investment. No research-stage dataset is presented as an available OilPriceAPI product.

PARM means Problem → Approach → Research → Measurement.

The framework keeps preliminary source research separate from product availability and forces every experiment to end in an observable decision.

Experiment 1collection probe

Posted prices

Can public purchaser bulletins become a reliable, versioned history of crude grades, locations, effective dates, and corrections? This experiment starts with source preservation—not a product claim.

Cadence
Daily source checks; immutable capture whenever content changes
First decision
Can OilPriceAPI collect enough direct purchaser postings, with sufficient continuity and rights clarity, to justify a normalized historical API?
Review the research
Experiment 2collection probe

Production vintages

Can repeated snapshots reveal how federal and state oil-production estimates change after first publication—and create an auditable ‘as known on’ history for models and backtests?

Cadence
Daily availability checks plus immutable snapshots at each official release
First decision
Do official oil-production series revise often and materially enough—and disappear thoroughly enough—to justify a retained point-in-time API?
Review the research
Experiment 3collection probe

Refinery outages

Can public regulatory and operator evidence identify refinery outages quickly and accurately enough to support a useful event history without pretending that a proxy is a confirmed outage?

Cadence
Hourly checks for fast public signals; daily evidence review; explicit closure tracking
First decision
Can a public-source event ledger reach commercially useful timeliness and accuracy, or is licensed/analyst-confirmed data required from the beginning?
Review the research
Experiment 4collection probe

Pipeline notices

Can repeated captures of public pipeline EBBs preserve critical notices, planned service outages, revisions, and cancellations well enough to create a normalized historical record?

Cadence
Five-to-fifteen-minute polling during the probe, with immutable versions and explicit failures
First decision
Are public EBB notices transient and heterogeneous enough—and buyer-relevant enough—to justify continuous capture and cross-pipeline normalization?
Review the research

Demand validation

Fragmented regulatory-data hypotheses

Keyword estimates, market evidence, current product behavior, and direct workflow conversations are kept separate so a compelling data-engineering problem does not get mistaken for demand.

Measured demand probe

Would joined flaring authorization and observation data replace manual compliance research?

A measured demand probe for a well- and operator-level dataset joining regulatory flare/vent exceptions, reported volumes, and satellite observations. This is research, not an available OilPriceAPI product.

Review evidence and stop rules
Measured demand probe

Is produced-water and disposal capacity data valuable enough to normalize across regulators?

A measured demand probe for produced-water volumes, Class II/UIC permits, injection monitoring, well locations, pressures, and operating status. This is not a live dataset or endpoint.

Review evidence and stop rules
Measured demand probe

Does a permit-to-production lifecycle join beat another isolated well dataset?

A measured demand probe for normalized permit, spud, completion, first production, status, and plugging milestones keyed by API number. No lifecycle endpoint is currently promised here.

Review evidence and stop rules

What the first collection produces

Immutable raw captures, source and rights inventories, change and failure logs, minimal schemas, and a Day-14 comparison of coverage, transience, normalization cost, and buyer evidence. Production endpoints come only after a probe passes its gates.