Oil and gas trade price fluctuations and profit divergence

Customer Case · Oil and Gas Trade Risk and Profit Management

Under price fluctuations, how to make 'spot profit' the final profit

From static spreads and diversified hedging to full lifecycle economic profit and AI decision collaboration

Scope of Application: Crude oil, refined oil, fuel oil, naphtha, and related cross-border spot trade

Version 2026.08 | Joint reading by management, trading, risk control, operations, and finance

Executive Summary

What the customer sees at the time of the transaction is the static gross margin between the purchase price and the selling price, but the final economic profit is jointly determined by pricing windows, benchmarks and basis, actual quantities, shipping schedules/month differences, freight and demurrage, exchange rate financing, inventory valuation, and claims. Simply 'hedging if the direction is right' cannot guarantee profit locking and may even result in losses in both spot and derivatives due to timing and mismatch of the underlying.

Overview of Case Conclusions

First unify profits, then automate hedging; first calculate risks clearly, then discuss AI.

Use 'effective coverage rate' instead of nominal hedge ratio, and 'final profit deviation rate' as the North Star metric.

Adopt 'one profit standard, five layers of control, one closed-loop platform' to bring trading, operations, risk control, and finance onto the same event chain.

1. Customer Background and Pain Points

The case client is a medium-to-large cross-border oil trading platform (anonymous), mainly dealing in diesel, fuel oil, and naphtha. Procurement and sales usually use floating pricing, with voyages lasting 20–40 days, and they also use commodity derivatives, foreign exchange forwards, and trade financing. As transaction volumes expanded, the divergence between book gross profit and final settlement profit became the most challenging operational issue for management to explain.

Three Observable Signals of Customer Pain Points

The expected gross profit at the time of the deal was considerable, but the closer to settlement, the more frequent the downward revisions; the execution team could not explain the extent of the revisions in advance.

When the futures side is profitable, the spot inventory may not yet reflect the depreciation according to market value; when the spot side is profitable, hedging losses are accounted for separately.

The risk meeting discussed 'oil price fluctuations,' yet it did not address which day, which benchmark, what quantity, or which currency is still exposed.

2. Diagnostic Method: Tracing from Contract Price Difference to Final Cash

The project team replaces department interview-based attribution with a 'full lifecycle trade playback': representative settled trades are selected, and contracts, pricing, derivatives, logistics, inventory, foreign exchange, financing, invoices, claims, and demurrage fees are aligned by a unique trade number, with economic profit recalculated daily.

Unified profit equation

Principles of Diagnosis

Nominal tonnage can only describe the cargo quantity and cannot represent risk coverage; coverage must be calculated based on price sensitivity and the valuation date.

Derivative profits are not independent performance and must be evaluated together with the corresponding physical risk.

Any unhedged exposure must have a risk owner, authorized limit, maturity date, and disposal action.

3. Key Findings: Eight Types of Drivers Jointly Devour Gross Margin

4. Representative Example: How a Book Gross Profit of 1 Million USD Is Reduced to Only 100,000

The following is a case calculation example for 50,000 tons of diesel based on reference materials. The amounts are used to demonstrate the profit bridge logic and do not represent the audited data of any specific company. The static price difference at the time of transaction is $20/ton, corresponding to an expected gross profit of $1 million.

5. Root cause of the problem: The risk was not translated to the same calendar

Purchasing, sales, and hedging are often managed based on the total contract volume, but what truly determines profit and loss is the risk sensitivity on each pricing day. Even if the nominal hedging rate reaches 95%, if there is a mismatch in the date, underlying asset, or month, the net exposure can still be significant.

Root Cause Chain

Contract terms are unstructured → Unable to generate an accurate pricing calendar.

Shipping schedule, quantity, and quality events are not automatically updated → positions still remain in the original plan.

Hedging is executed on a single contract basis → Repetition or offsetting at the portfolio level is not identified.

Logistics, capital, and claims costs lag → expected profits continue to be overestimated.

Performance is based on transaction gross profit or departmental profit and loss → Risk bearers and benefit recipients are not aligned.

6. Solutions Overview: One Profit Metric, Five Layers of Control, One Closed-Loop Platform

Event stream of the closed-loop platform

7. Layers One to Three: Quotation, Exposure, and Layered Hedging

1. Calculate lockable profit before quoting

The pricing model uses the currently executable buy and sell prices and lockable costs, and sets aside risk reserves for unhedgeable basis, quantity and quality, shipping schedule, and tail-end scenarios. 'Risk profits' that still depend on future market conditions must not be regarded as confirmed gross profit.

2. Establish the pricing quantity calendar

Separate fixed price exposure from floating price exposure; record buy risk as positive and sell risk as negative.

Recalculate daily based on the bill of lading, ETA, actual loading, batch sales, and customer quantity changes.

Nominal coverage is only auxiliary; the core metric is the effective sensitivity coverage that matches the actual risk.

3. Layered Hedging

VIII. Fourth Layer: Profit and Loss Statement for the Entire Lifecycle of a Single Transaction

Each transaction uses a unique number, linking purchasing, sales, shipping and storage, inspection, derivatives, foreign exchange, financing, invoicing, payment collection, demurrage, and claims. The system saves snapshots at key points in time, allowing profit changes to be explained, attributed, and reviewed.

Profit Bridge Attribution Sequence

Changes in contract terms

Market and Basis Changes

Changes in Quantity and Quality

Changes in shipping schedule, freight, and demurrage

Financing and Exchange Rate Changes

Claims, Taxes, and Final Settlement

9. Fifth Level: Real-Time Warning, Authorization, and Action Mechanism

Thresholds must be tied to actions, and red alerts cannot be limited to emails. The table below provides recommended values for reference; companies should calibrate based on transaction size, liquidity, and risk tolerance.

Division of governance

10. Implementation Roadmap: Form a sustainable closed loop in 12 months

90-Day Priorities

Select 20–30 representative settled transactions, reconstruct the profit bridge, and create a risk dictionary.

Unify products, units, currencies, price sources, timestamps, and the association keys of transactions/batches/hedges.

Upload the quotation approval list, daily pricing calendar, and end-of-day tripartite reconciliation.

Select a high-frequency oil product and two routes for parallel trial operation, without immediately expanding the automatic execution authority.

Key data interface

ETRM/CTRM, ERP/General Ledger, Futures and Swap Platforms, Banking/Cash Management, AIS Shipping, Warehousing Tank Areas, Market Prices and Foreign Exchange, CRM/Credit, Contract and Document Repository. The data layer must retain original values, revised values, sources, timestamps, and responsible persons.

11. Outcome Indicators and Business Value

The following values are the 'trial run/target range' designed based on the scale of case problems, used for project acceptance and value measurement, and do not constitute a uniform industry benchmark, nor do they represent audited performance by clients.

12. AI Evolution: From Data Assistant to Controlled Decision Collaboration

Typical AI Alert

Five Red Lines That AI Must Not Cross

Each recommendation can be traced back to real transactions and risk exposure; exceeding the physical risk is considered potential speculation and triggers an upgraded approval process.

It is not allowed for general-purpose generative AI to place unlimited orders directly.

Automatically downgrade to manual decision-making when there is data interruption, major contract changes, extreme price jumps, or model conflicts.

The output must include the use of data, residual basis, extreme scenarios, confidence, and failure conditions.

The criteria for evaluating AI are profit deviation, limit breaches, margin forecasts, and human time, rather than simply the accuracy of price predictions.

13. Replicable Experiences and Professional Conclusions

Six Replicable Experiences

First rebuild the profit bridge of settled transactions to avoid building a large system based on incorrect metrics.

Contract terms must be structured around benchmarks, formulas, windows, locations, quality, currency, and tolerances.

Hedging decisions must take into account both price risk and cash margin to prevent 'economically efficient but liquidity compromised' situations.

Basis is not unmanageable: alignment of terms, spread tools, shortening cycles, substitution of cargo flows, and reduction of reserve combinations.

Performance should be measured using risk-adjusted return on capital to avoid rewarding accidental profits from unauthorized exposures.

The highest value of AI is to shorten the cycle of recognition—decision—execution—review, not to replace authorized responsibility.

Professional conclusion

Caliber and Instructions for Use

This report is compiled based on 1–5 case materials provided by the user. Clients, transactions, and metrics are expressed anonymously and in case-based form; USD/ton, tonnage, time, and target values are used to demonstrate management logic. Prior to formal implementation, calibration should be conducted based on actual contracts, master data, accounting policies, risk limits, derivative liquidity, local regulations, and the company's risk tolerance.

Next step recommendation: Conduct a 6-week profit bridge diagnosis using 20–30 settled transactions, create a list of data discrepancies, a risk dictionary, and an initial business blueprint, and then decide the scope of system transformation.