Seven Hedging Pain Points for Trading Companies

Trading companies hedge differently from manufacturers in one important way: their exposure is not inventory exposure, it is contract exposure. The goods may not have been bought, the route may not have moved, the customer may not have priced — but the price risk is already on the company.

That makes hedging structurally harder for a trader. A manufacturer's risk object is physical goods in a warehouse: visible, countable. A trader's risk object is a set of contract terms plus a future pricing date, and both have to be translated, checked and passed along by people.

The seven pain points below apply to almost every trading company of any size.

Pain point 1: Contracts and futures positions are two books that do not know each other

This ranks first of the seven, because it distorts the other six.

The typical scene: sales holds a procurement contract priced on a basis three months out; trading holds the matching long futures position. The problem is that these two things sit in two different systems, with nothing forcing a link between them.

So every morning you see the same scene: business reports how much exposure is left, trading reports how many positions it holds, and risk reconciles the two by hand in a spreadsheet. That takes 2 to 3 hours a day, and the more troublesome part is that the error rate usually sits between 2% and 5%.

What does 2%-5% mean? It means the company's actual hedge ratio may be well below, or well above, what the policy says. When prices rise it may be luck; when prices fall it is a real loss. And because the error is there every day, nobody can point to where it comes from — which is a worse problem than the error itself.

Pain point 2: The more complex the pricing rules, the more likely a manual pricing error

Traders win customers by staying flexible on price: fixed, deferred, average, average basis — sometimes a different method for different batches of the same contract. Each of those has its own algorithm, and each requires three inputs to be assembled: futures price, basis and delivery date.

The result of manual calculation: 5 to 10 minutes to price a single contract. During busy periods that turns directly into a backlog, and the backlog raises the probability of a misprice — because the error rate rises sharply when people are tired and under pressure.

More worrying is the directional character of a misprice. A basis mispriced by RMB 20 is RMB 200,000 on a 10,000-tonne contract and RMB 2 million on a 100,000-tonne contract. The more complex the pricing rules, the more fatal this multiplier.

Pain point 3: No single definition of exposure, so no basis for decisions

In a trading company there are at least three versions of exposure:

• Business view: by customer — how much has this customer not priced yet;
• Risk view: by commodity — is total exposure on this commodity over the limit;
• Finance view: by accounting period — how much did the cost of this batch move.

All three are correct, but they cannot be compared directly. So "how exposed are we right now" often has no standard answer inside a trading company — everyone's number is right, and everyone's number means something different.

The consequence is that discussion of the hedge ratio stops being a calculation and becomes a contest of positions. Business says exposure is large, risk says it is within limits, both talk past each other, and the outcome is usually set by whoever is louder rather than by the actual risk.

Pain point 4: The night session is the most expensive part of a trading day

Pricing requests are distributed very unevenly across the day. Collated industry practice puts night-session pricing at around 35% of the total.

For a trading company the night session is not overtime — it is where the business happens. A customer pricing at night is telling you they are serious and intend to deal. If you are absent then, the damage is doubled: the customer waits and goes elsewhere, or prices at a level you cannot accept, and a live opportunity is wasted.

The difficulty of staffing the night session by hand is that it is a low-frequency but high-demand role. There may be only a handful of pricing requests in a week, but every one of them needs someone to respond within 2 minutes. Staffing it around the clock is prohibitively expensive; staffing it on demand cannot guarantee the response time. That is why this problem is ultimately solved by a system rather than by adding headcount.

Pain point 5: Attribution errors are going to happen

Already calculated above: manual attribution takes 8 to 12 minutes per trade, so with dozens to a hundred trades a day, an error rate of 1%-3% is normal rather than exceptional.

For a trading company the impact is larger than for a manufacturer, because the chain is longer: a futures fill has to be attributed to a contract, the contract to a customer, the customer to a salesperson, and finally reconciled with the finance ledger. Any break in that chain gets amplified along it.

Trading companies also typically layer on basis business — a basis trader earns the spread, the spread is thin, and one attribution error can flip the direction of the P&L.

Pain point 6: Many traders in parallel, no clear chain of responsibility

Futures trading in a trading company is naturally collaborative: different commodities, different customers, different accounts, often different people. That division of labour is reasonable, but it creates two problems.

The first is inconsistent definitions. Two traders tag their fills according to their own understanding, so the same kind of business is attributed under different rules in different hands, and the totals do not reconcile.

The second is conflicts are not detected in time. On the same account, one trader believes a position is still open and another believes it is closed — in a manual setup that conflict usually surfaces only at reconciliation.

The more important point is the audit view: when something goes wrong, management needs to know who made the call and on what basis. If the operation log holds only the final result and not the process, every review has to rely on memory — and memory three months later is unreliable.

Pain point 7: Audit, inspection and disclosure all need a complete evidence chain

Trading is one of the most heavily regulated business models. Group audit, external audit, state-asset compliance checks, disclosure once listed — each of them asks the same question: can your hedging process be reproduced in full?

The reality is usually that the data exists, but is scattered across systems, files and versions. Retrieving history and reconciling beyond 30 days already takes 15 minutes a time as an optimistic estimate, and it is easy to miss something that matters.

There is also a dimension of time: what compliance expects keeps tightening. Regulators now want not only outcome data but the full chain — who approved it, under which rule, how exceptions were handled. No manual process, however skilled, reliably meets that.

Seven pain points, one comparison table

If you can only do one thing

These seven pain points have a clear priority order: fix the contract-position-exposure mapping first.

The reason is simple: exposure is the input to every decision. How the ratio is set, how limits are approved, how much margin is held and whether the hedge worked all depend on its accuracy. If exposure is still assembled by hand, optimising the other six processes is building on unstable ground.

A practical way to start: pick one commodity and one group of customers, put the existing contract ledger and the futures positions on the same table, reconcile the difference trade by trade and explain every difference. Only when 100% of the differences are explained has the mapping actually been established.

After that, the second move is to parameterise the pricing rules and the third is to hand routine pricing and attribution to a system. The order cannot be reversed — automating an inaccurate process only produces untrustworthy results faster.

Data sources and notes

The practice data in this article — the 2 to 3 hours a day of manual reconciliation and the 2%-5% error rate produced by split contract and position ledgers; manual pricing time of 5-10 minutes per contract; night-session pricing at around 35% of the total; manual attribution time of 8-12 minutes per trade and an error rate of 1%-3%; and history search and reconciliation beyond 30 days taking about 15 minutes a time — come from RMC's diagnostic records for trading clients and from before-and-after comparisons taken across systemisation projects. This article adds analysis of the three definitions of exposure, the amplification effect along the chain and the priority order. Nothing here constitutes an endorsement or evaluation of any specific software product.