From Geopolitical Event to Testable Market Thesis

A geopolitical market note is strongest when it predicts an observable sequence rather than offering a directional opinion.

“Oil will rise because the conflict is serious” is not a mechanism.

A useful thesis identifies:

• the actors;
• the decisions they can make;
• the constraints on those decisions;
• the physical or commercial system those decisions enter;
• the market expected to react first;
• the observations that should follow;
• the strongest rival explanations;
• and the evidence that would kill the view.

This article is the reusable method I extracted while rebuilding my Onyx submission.

For the actual US–Iran and Hormuz case study, see The Strait Is the Price.

The method does not make geopolitical episodes clean. It prevents the writing from becoming cleaner than the evidence.

1. Define the decision before the topic

A topic is not yet a research question.

Topic:
The US–Iran conflict and oil.

Weak question:
Will oil go up?

Decision question:
Which political and commercial sequence would create persistent delivery stress,
which market should confirm it first,
and what evidence would reverse the view?

The decision question establishes what the note is for.

Possible outputs include:

• a directional position;
• a relative-value position;
• an options structure;
• a monitoring framework;
• a procurement decision;
• a policy assessment;
• a risk limit;
• or a scenario trigger.

Do not begin by collecting every available article. Begin by naming the decision.

Write the mechanism in one sentence

Use this structure:

Because [actor/action] changes [physical or commercial constraint],
I expect [first observable state] to move before [later outcome].
The thesis weakens if [specific contradiction].

For the Onyx reconstruction:

Because conflict risk changes insurance and owner decisions,
delivery reliability should deteriorate before the full effect appears
in production, inventories, products and flat crude.
The thesis weakens if ordinary cover, traffic and product supply recover.

The sentence is provisional. Its purpose is to make later revision visible.

Define the unit of analysis

Specify whether the note concerns:

• one announcement;
• one military action;
• one shipping incident;
• an embargo;
• a policy intervention;
• a sequence of events;
• or a persistent regime.

A one-hour event window and a multi-month disruption require different evidence.

2. Freeze the evidence boundary

The first operational task is not collecting links.

It is freezing the cutoff.

Without a cutoff, later confirmation leaks backward into the original thesis. The note becomes a story about what happened rather than a record of what could have been known.

Evidence states

State	Definition	Permitted use
pre_draft	Known before the first thesis was written.	Background, prior evidence and initial assumptions.
at_cutoff	Publicly available by the formal submission or publication cutoff.	May support the original claim.
post_cutoff_pre_event	Available after submission but before a later focal event.	Tests the thesis but cannot be backfilled into the submission.
post_event	Observed after the event sequence.	Retrospective support, revision or contradiction.
current_revision	Added in a later public revision.	Must be visibly dated and separated from the original record.

pre_draft

Definition

Known before the first thesis was written.

Permitted use

Background, prior evidence and initial assumptions.

at_cutoff

Definition

Publicly available by the formal submission or publication cutoff.

Permitted use

May support the original claim.

post_cutoff_pre_event

Definition

Available after submission but before a later focal event.

Permitted use

Tests the thesis but cannot be backfilled into the submission.

post_event

Definition

Observed after the event sequence.

Permitted use

Retrospective support, revision or contradiction.

current_revision

Definition

Added in a later public revision.

Permitted use

Must be visibly dated and separated from the original record.

The labels can differ. The separation cannot.

Record event date and publication date separately

Every source record should retain:

• publicationDate
• eventDate
• accessDate
• evidenceState
• knownAtSubmission
• currentRevisionOnly

Publication date and event date are not interchangeable.

A report published on 15 July may describe tanker activity from 10 July. It can support a retrospective account of the 10 July event, but it was not available to a submission made on 13 July.

Preserve source vintages

Official forecasts must retain their vintage.

A July forecast is not “wrong” merely because later events overtook it. It records the information set and assumptions available at that date.

Store:

• publisher
• release
• vintage
• data period
• forecast horizon
• assumptions
• subsequent revision

Do not silently replace an earlier vintage with the latest series.

Apply the evidence-state test

Before publishing a factual paragraph, ask:

• When did the event occur?
• When was the source published?
• Could it have been known at the relevant cutoff?
• Is it being used as contemporary evidence, hindsight or current context?
• Is that distinction visible to the reader?

If a paragraph mixes states without saying so, rewrite it.

3. Build actor and event ledgers before writing the story

“Geopolitics caused oil to rise” removes both the actors and the transmission system.

Replace it with one row per actor and one row per event.

Actor ledger

Minimum actor-ledger fields

Field	Purpose
actor	Names the decision-maker rather than attributing agency to an abstract event.
statedObjective	Preserves the actor’s public justification.
inferredObjective	Labels interpretation as inference rather than fact.
constraints	Explains what the actor cannot do.
availableActions	Defines the realistic action set.
observedAction	Records behaviour rather than commentary.
counterpartyResponse	Captures strategic interaction.
physicalExposure	Connects the actor to infrastructure, production or flows.
commercialExposure	Connects the actor to insurance, finance, contracts or logistics.
marketChannel	Names the market expected to react first.
sourceIds	Provides provenance.
confidence	Limits overclaim.

actor

Purpose

Names the decision-maker rather than attributing agency to an abstract event.

statedObjective

Purpose

Preserves the actor’s public justification.

inferredObjective

Purpose

Labels interpretation as inference rather than fact.

constraints

Purpose

Explains what the actor cannot do.

availableActions

Purpose

Defines the realistic action set.

observedAction

Purpose

Records behaviour rather than commentary.

counterpartyResponse

Purpose

Captures strategic interaction.

physicalExposure

Purpose

Connects the actor to infrastructure, production or flows.

commercialExposure

Purpose

Connects the actor to insurance, finance, contracts or logistics.

marketChannel

Purpose

Names the market expected to react first.

sourceIds

Purpose

Provides provenance.

confidence

Purpose

Limits overclaim.

Do not turn an inferred objective into a fact through repetition.

Introduce the inference once and retain the label.

Controlled event classes

Use controlled classes such as:

• threat
• policy_announcement
• diplomatic_action
• sanction_or_licence
• military_act
• shipping_incident
• physical_disruption
• commercial_response
• policy_buffer
• negotiated_change
• operational_normalisation

A threat, an attack, an owner withdrawal and a production shut-in are different event types.

If the chronology collapses them, the note will overfit headlines.

Event ledger

Minimum fields include:

• eventId
• eventDateTime
• publicationDateTime
• actor
• action
• eventClass
• statedReason
• inferredObjective
• counterpartyResponse
• physicalEffect
• commercialEffect
• marketChannel
• immediateResponse
• laterResponse
• alternativeExplanation
• sourceIds
• sourceQuality
• evidenceState
• confidence
• publicSafe

Build the chronology before writing the narrative.

Narrative prose creates pressure to preserve a coherent arc. A ledger makes contradictions visible.

In the Onyx case, the useful chronology was not a one-step story from war to higher oil prices.

It was closer to:

4. Convert chronology into testable causal arrows

A chronology orders events. It does not establish a mechanism.

The next step is to propose the causal chain.

1. 1

   Political action → disruption probability

   Timing: Minutes to hours · Confidence: high

2. 2

   Disruption probability → insurer/shipowner behaviour

   Timing: Hours to 1–2 days · Confidence: high

3. 3

   Insurer/shipowner behaviour → war-risk premium/tanker availability

   Timing: Same day to week · Confidence: high

4. 4

   Tanker availability → vessel flow/delivery reliability

   Timing: 1–7 days · Confidence: high with data caveat

5. 5

   Flow/reliability → shut-ins/rerouting

   Timing: Days to weeks · Confidence: high

6. 6

   Shut-ins/rerouting → refinery/inventory effects

   Timing: Weeks · Confidence: high

7. 7

   Refinery/inventory effects → crude spreads/product cracks

   Timing: Days to weeks · Confidence: high

8. 8

   Spreads/cracks → flat price/volatility

   Timing: Minutes to days · Confidence: medium-high

9. 9

   Sustained prices/shortages → macro demand response

   Timing: Weeks to months · Confidence: medium-high

View text version

Transmission links (text fallback)

Link	Timing	Confidence
Political action → disruption probability	Minutes to hours	high
Disruption probability → insurer/shipowner behaviour	Hours to 1–2 days	high
Insurer/shipowner behaviour → war-risk premium/tanker availability	Same day to week	high
Tanker availability → vessel flow/delivery reliability	1–7 days	high with data caveat
Flow/reliability → shut-ins/rerouting	Days to weeks	high
Shut-ins/rerouting → refinery/inventory effects	Weeks	high
Refinery/inventory effects → crude spreads/product cracks	Days to weeks	high
Spreads/cracks → flat price/volatility	Minutes to days	medium-high
Sustained prices/shortages → macro demand response	Weeks to months	medium-high

Sources: EIA / IEA / Reuters public reporting cited in the article body · Cutoff 2026-07-13 · Current revision 2026-08-01

Test every arrow separately.

Arrow-ledger fields

Field	Question
fromState	What changes first?
toState	What should change next?
mechanism	Why should the first state affect the second?
expectedSign	Is the effect positive, negative or non-monotonic?
expectedLag	Should it appear in minutes, days, weeks or months?
confirmingMeasure	What observation supports the arrow?
contradictingMeasure	What observation weakens it?
alternativeChannel	What else could produce the same outcome?
sourceIds	Which evidence supports the claim?
confidence	How certain is the link?

fromState

Question

What changes first?

toState

Question

What should change next?

mechanism

Question

Why should the first state affect the second?

expectedSign

Question

Is the effect positive, negative or non-monotonic?

expectedLag

Question

Should it appear in minutes, days, weeks or months?

confirmingMeasure

Question

What observation supports the arrow?

contradictingMeasure

Question

What observation weakens it?

alternativeChannel

Question

What else could produce the same outcome?

sourceIds

Question

Which evidence supports the claim?

confidence

Question

How certain is the link?

A short example:

Example arrow tests

From	To	Mechanism	Confirmation	Contradiction
Attack probability	War-risk terms	Underwriters reprice correlated loss risk.	Higher premium or reduced capacity.	Ordinary terms remain stable.
War-risk terms	Vessel behaviour	Owners require compensation or avoid the route.	U-turns, delays or higher freight.	Traffic remains normal.
Lower traffic	Export constraints	Storage fills and cargoes cannot clear.	Shut-ins, floating storage or lower loadings.	Production and exports remain normal.
Export constraints	Product stress	Refineries lose feedstock or export capacity.	Lower runs, inventory draws or higher cracks.	Products normalise first.

Attack probability

To

War-risk terms

Mechanism

Underwriters reprice correlated loss risk.

Confirmation

Higher premium or reduced capacity.

Contradiction

Ordinary terms remain stable.

War-risk terms

To

Vessel behaviour

Mechanism

Owners require compensation or avoid the route.

Confirmation

U-turns, delays or higher freight.

Contradiction

Traffic remains normal.

Lower traffic

To

Export constraints

Mechanism

Storage fills and cargoes cannot clear.

Confirmation

Shut-ins, floating storage or lower loadings.

Contradiction

Production and exports remain normal.

Export constraints

To

Product stress

Mechanism

Refineries lose feedstock or export capacity.

Confirmation

Lower runs, inventory draws or higher cracks.

Contradiction

Products normalise first.

The chain can fail at any arrow.

That is useful. A mechanism that cannot fail is not a testable mechanism.

Distinguish necessary from sufficient conditions

A political agreement may be necessary for de-escalation but insufficient for commercial normality.

Safe passage may also require:

• underwriting capacity;
• willing owners;
• available crews;
• acceptable freight;
• functioning terminals;
• compatible cargoes;
• and repeated successful voyages.

Do not treat one condition as proof of the full state.

Model adjustment speeds

Different states update on different clocks.

Illustrative adjustment speeds

State	Typical speed
Headline expectations	Seconds to minutes
Futures and options	Seconds to hours
Insurance terms	Hours to days
Owner and charter decisions	Hours to days
Vessel positioning	Days to weeks
Production shut-ins and restarts	Days to months
Refinery recovery	Days to months
Inventory rebuilding	Weeks to months
Infrastructure investment	Years

Headline expectations

Typical speed

Seconds to minutes

Futures and options

Typical speed

Seconds to hours

Insurance terms

Typical speed

Hours to days

Owner and charter decisions

Typical speed

Hours to days

Vessel positioning

Typical speed

Days to weeks

Production shut-ins and restarts

Typical speed

Days to months

Refinery recovery

Typical speed

Days to months

Inventory rebuilding

Typical speed

Weeks to months

Infrastructure investment

Typical speed

Years

A benchmark reversal can therefore precede physical recovery.

The lag structure belongs inside the thesis.

5. Decompose the market move and define rival explanations

Do not ask whether “the event moved oil”.

Ask which combination of shocks changed.

At minimum, separate:

• realised supply loss;
• expected future supply loss;
• precautionary demand;
• inventory demand;
• freight;
• war-risk insurance;
• refinery disruption;
• product scarcity;
• policy supply;
• non-affected producer response;
• global demand;
• currency;
• risk appetite;
• positioning and forced flows.

Expectations are not fake effects.

A market can move before current output changes because supply news changes expected scarcity, inventory demand and future macroeconomic conditions.

Treat event windows as descriptive by default

An annotated price chart is useful for chronology.

It is not causal identification unless the design has a defensible counterfactual.

For each event window, record:

• windowStart
• windowEnd
• eventTimestamp
• otherNews
• marketOpenState
• priceChange
• volume
• volatility
• reversalWindow
• interpretation
• alternativeExplanation

Avoid selecting only events whose price response matches the thesis.

Predict a market hierarchy

Ask which market should react first.

For a shipping shock, a plausible hierarchy is:

• insurance and freight;
• vessel behaviour;
• regional differentials;
• nearby time spreads;
• product cracks;
• flat crude;
• macroeconomic variables.

The exact ordering can differ. It must be declared before the observations are used as confirmation.

Use negative controls

A useful analogue or event set includes episodes that should not confirm the mechanism.

Examples:

• a large geopolitical headline with little physical supply effect;
• a demand shock without route disruption;
• a refinery outage without upstream loss;
• a political agreement without commercial normalisation.

A negative control prevents the evidence set from becoming a collection of supportive stories.

6. Select analogues by mechanism, not chart shape

Match historical episodes on:

• source of disruption;
• route or infrastructure affected;
• realised supply loss;
• expected duration;
• available inventories;
• spare capacity;
• product constraint;
• demand response;
• policy intervention;
• recovery mechanism.

Illustrative analogue matrix

Analogue	Mechanism tested	Useful similarity	Critical difference
1973–74 embargo	Producer-state coercion and importer response	Supply was used as foreign-policy leverage and later produced new emergency institutions.	Selective production restrictions were not the same as unsafe maritime transit.
1990 Gulf shock	Replacement-supply credibility	A large premium can reverse when replacement supply becomes credible.	It was not a clean commercial closure of Hormuz.
2019 Abqaiq	Repair credibility	A large outage can be short when repair and inventories are credible.	The shock centred on production infrastructure rather than route governance.
2022 Russia invasion	Commercial adaptation	Sanctions, insurance, rerouting and product effects persisted after the first shock.	There was no single physical chokepoint.
9/11	Negative control	A major geopolitical shock affected expectations and demand.	There was limited direct crude-supply loss.

1973–74 embargo

Mechanism tested

Producer-state coercion and importer response

Useful similarity

Supply was used as foreign-policy leverage and later produced new emergency institutions.

Critical difference

Selective production restrictions were not the same as unsafe maritime transit.

1990 Gulf shock

Mechanism tested

Replacement-supply credibility

Useful similarity

A large premium can reverse when replacement supply becomes credible.

Critical difference

It was not a clean commercial closure of Hormuz.

2019 Abqaiq

Mechanism tested

Repair credibility

Useful similarity

A large outage can be short when repair and inventories are credible.

Critical difference

The shock centred on production infrastructure rather than route governance.

2022 Russia invasion

Mechanism tested

Commercial adaptation

Useful similarity

Sanctions, insurance, rerouting and product effects persisted after the first shock.

Critical difference

There was no single physical chokepoint.

9/11

Mechanism tested

Negative control

Useful similarity

A major geopolitical shock affected expectations and demand.

Critical difference

There was limited direct crude-supply loss.

An analogue is evidence about one mechanism or arrow. It is not a template for the whole event.

7. Choose the market that expresses the mechanism

Only after specifying the mechanism should the note choose an instrument.

Mechanism-to-market mapping

Mechanism	Candidate market	Why
Global expected shortage	Flat crude	Captures the broad balance and expectations.
Immediate local scarcity	Nearby spread	Responds to prompt availability and inventories.
Route or location constraint	Regional differential	Isolates exposed grades and destinations.
Vessel scarcity	Tanker freight	Measures the direct delivery-cost channel.
Refinery or product shortage	Diesel or jet crack	Links the thesis to usable-product scarcity.
Discontinuous escalation	Options	Expresses convex event risk.

Global expected shortage

Candidate market

Flat crude

Why

Captures the broad balance and expectations.

Immediate local scarcity

Candidate market

Nearby spread

Why

Responds to prompt availability and inventories.

Route or location constraint

Candidate market

Regional differential

Why

Isolates exposed grades and destinations.

Vessel scarcity

Candidate market

Tanker freight

Why

Measures the direct delivery-cost channel.

Refinery or product shortage

Candidate market

Diesel or jet crack

Why

Links the thesis to usable-product scarcity.

Discontinuous escalation

Candidate market

Options

Why

Expresses convex event risk.

A route thesis expressed only through flat Brent may be overwhelmed by global demand, currency or unrelated supply.

Minimum trade specification

Before publishing a trade, record:

• instrument
• contract and expiry
• direction
• entry logic
• carry
• liquidity
• spread and execution cost
• margin
• volatility and skew
• position size
• maximum loss
• stop
• time stop
• fundamental invalidation
• data timestamp

If those inputs are unavailable, publish a monitoring framework rather than a synthetic trade.

Separate trade risk from thesis invalidation

A trade can lose while the thesis remains plausible because timing, carry, volatility or execution was poor.

The thesis can also fail while the trade profits for an unrelated reason.

Record both:

• trade stop
• fundamental invalidation

They are not the same field.

8. Replace composite scores with update rules

Composite scores are attractive because they create one number.

They are dangerous when the weights, calibration and decision mapping are unsupported.

A score averaging traffic, insurance, products and flat price can hide the disagreement that matters.

Prefer an update matrix.

Scenario update matrix

State	Trigger	First confirmation	Later confirmation	Contradiction
De-escalation	Sustained restraint and a credible transit agreement.	Ordinary cover and repeated voyages.	Production, refinery and inventory recovery.	Better headlines but continued commercial exclusion.
Constrained transit	Some voyages under exceptional terms.	Intermittent traffic and elevated freight.	Partial crude recovery with persistent product stress.	Ordinary traffic and products normalise.
Sustained disruption	Continued attacks or cover withdrawal.	Owner avoidance and freight stress.	Shut-ins, refinery loss and inventory draws.	Traffic and cover recover despite escalation.

De-escalation

Trigger

Sustained restraint and a credible transit agreement.

First confirmation

Ordinary cover and repeated voyages.

Later confirmation

Production, refinery and inventory recovery.

Contradiction

Better headlines but continued commercial exclusion.

Constrained transit

Trigger

Some voyages under exceptional terms.

First confirmation

Intermittent traffic and elevated freight.

Later confirmation

Partial crude recovery with persistent product stress.

Contradiction

Ordinary traffic and products normalise.

Sustained disruption

Trigger

Continued attacks or cover withdrawal.

First confirmation

Owner avoidance and freight stress.

Later confirmation

Shut-ins, refinery loss and inventory draws.

Contradiction

Traffic and cover recover despite escalation.

Probabilities may be added as a judgement layer.

They are not a substitute for observable triggers.

Update direction before magnitude

For each new observation, record:

• supports
• weakens
• ambiguous
• unrelated

Then record magnitude and confidence.

This prevents a dramatic but weakly connected headline from dominating the update.

Use contradiction deliberately

Every scenario should contain an observation that would surprise it.

Examples:

• headlines improve but insurers do not;
• traffic improves but products remain tight;
• products normalise without a large crude recovery;
• freight falls while visible traffic remains low;
• flat price rises while mechanism-specific markets weaken.

Contradiction tells you which arrow needs rebuilding.

9. Define invalidation before publication

An invalidation condition is an observation, not a feeling.

Weak:

The geopolitical situation improves.

Testable:

Ordinary underwriting capacity returns,
war-risk terms fall materially,
multiple laden voyages complete without exceptional arrangements,
production restarts,
and the predicted product or spread signal fails to respond.

Use three levels.

Arrow invalidation

One link in the mechanism fails.

Example:

Insurance terms rise, but vessel traffic remains normal.

Thesis invalidation

The central mechanism no longer explains the sequence.

Example:

Traffic, production and products normalise while the benchmark remains high because an unrelated demand or supply shock dominates.

Expression invalidation

The chosen market is a poor implementation.

Example:

Freight confirms delivery stress, but the flat-price position is overwhelmed by macroeconomic demand weakness.

This separation improves both research and trading discipline.

10. Maintain a claim ledger and reproducible evidence package

Every public factual sentence should map to a structured claim.

Minimum claim-ledger fields

Field	Purpose
claimId	Stable reference.
claim	Exact proposition being published.
sourceIds	Provenance.
sourceType	Primary, official, academic, reporting or internal.
eventDate	When the underlying event occurred.
publicationDate	When the source became public.
accessDate	Retrieval record.
evidenceState	Cutoff discipline.
verified	Whether the source supports the wording.
exactOrApproximate	Prevents false precision.
confidence	Limits certainty.
alternativeExplanation	Preserves the strongest rival account.
publicSafe	Controls publication boundaries.
destination	Identifies the article, chart or caption using the claim.

claimId

Purpose

Stable reference.

claim

Purpose

Exact proposition being published.

sourceIds

Purpose

Provenance.

sourceType

Purpose

Primary, official, academic, reporting or internal.

eventDate

Purpose

When the underlying event occurred.

publicationDate

Purpose

When the source became public.

accessDate

Purpose

Retrieval record.

evidenceState

Purpose

Cutoff discipline.

verified

Purpose

Whether the source supports the wording.

exactOrApproximate

Purpose

Prevents false precision.

confidence

Purpose

Limits certainty.

alternativeExplanation

Purpose

Preserves the strongest rival account.

publicSafe

Purpose

Controls publication boundaries.

destination

Purpose

Identifies the article, chart or caption using the claim.

Match the wording to the evidence quality.

Use:

• EIA estimated…
• IEA reported…
• Reuters reported…
• The submitted paper argued…
• The available evidence suggests…
• I infer…

Avoid:

• proved
• demonstrated conclusively
• caused
• guaranteed

unless the design genuinely supports that language.

Separate the reader-facing article from the research package

The article should explain the mechanism.

The repository should preserve enough structure to reproduce the evidence.

A useful conceptual layout is:

research/
  manifest/
    sources.json
    claims.json
    events.json
  raw/
    official/
    market/
    news-metadata/
  cleaned/
  derived/
  charts/
  diagrams/
  scripts/
  notes/

Use the repository’s actual conventions rather than creating this exact tree blindly.

For each source, retain:

• sourceId
• title
• publisher
• url
• publicationDate
• eventDate
• accessDate
• licence
• publicOrPrivate
• immutablePath
• checksum
• notes

Raw files should remain immutable. Cleaning and transformation should produce separate outputs.

Every chart or diagram should identify:

• input data;
• transformation script;
• output path;
• source note;
• cutoff;
• approximation;
• caption;
• accessible text.

If a visual cannot be reproduced from public data, label it as approximate, keep it private or omit it.

Do not manufacture a complete series from isolated observations.

Fail loudly

A build or research pipeline should fail when:

• a required source is missing;
• a data vintage is wrong;
• a checksum changes unexpectedly;
• an embed ID is missing;
• a public article references a draft page;
• a private artefact is about to be committed.

Silent substitution destroys cutoff discipline.

11. Publication test

Before publishing, answer the following.

Research

• Is the decision explicit?
• Is the cutoff frozen?
• Are evidence states visible?
• Are actors named?
• Are actions separated from commentary?
• Is the causal chain explicit?
• Does every arrow have confirming and contradicting evidence?
• Are rival explanations stated?
• Is the first confirming market named?
• Are analogues selected by mechanism?
• Are invalidation conditions observable?

Market expression

• Does the instrument match the mechanism?
• Are carry, liquidity and execution recorded?
• Is sizing defined?
• Are the trade stop and thesis invalidation separate?
• Is the data timestamp visible?
• Would a monitoring framework be more honest than a trade?

Publication safety

• Are inferred motives labelled?
• Are private documents excluded?
• Are publisher licences respected?
• Are current-revision facts separated from submission-time facts?
• Does every public link resolve?
• Does the prose avoid presenting indicative data as a transaction panel?

Reproducibility

• Does every key claim map to a source?
• Are raw files immutable?
• Are data vintages recorded?
• Are visual outputs deterministic?
• Will missing data fail loudly?
• Can another researcher identify what changed between revisions?

Compact operating checklist

1. Define the decision.
2. Write the mechanism in one sentence.
3. Freeze the cutoff.
4. Create evidence states.
5. Build the actor ledger.
6. Classify events.
7. Build the chronology.
8. Draw the causal graph.
9. Test every arrow.
10. Decompose the market move.
11. State rival explanations.
12. Select analogues by mechanism.
13. Name the first confirming market.
14. Choose an instrument or monitoring framework.
15. Define update rules.
16. Define arrow, thesis and expression invalidation.
17. Maintain the claim ledger.
18. Preserve source vintages.
19. Build deterministic visuals.
20. Fail loudly when required evidence is missing.

What this process can and cannot establish

The process can:

• force an explicit mechanism;
• preserve the information available at a cutoff;
• separate political actions from commercial and physical outcomes;
• predict an ordering of evidence;
• identify the market that should confirm first;
• expose disagreement between signals;
• state what would kill the view;
• preserve a reproducible public record.

It cannot:

• convert a complex geopolitical episode into clean causal identification by formatting it well;
• recover private intent from public statements;
• invent missing tanker, insurance, freight or options data;
• make indicative quotes equivalent to transactions;
• turn post-event confirmation into submission-time knowledge;
• guarantee that the selected trade will profit.

The correct response to those limits is not vagueness or false precision.

It is a dated mechanism, explicit rival explanations and observable update rules.

References