Risk management on an energy trading desk is how the firm decides how much it may lose, and how it checks every day that it cannot lose more. It starts from two numbers the desk already has: the position, which says how much it holds, and the mark-to-market P&L, which says what that is worth tonight. Everything in this article is built on those two.

It is written for analysts, engineers and anyone new to a desk. If you have not met positions or mark-to-market yet, read position aggregation and mark-to-market P&L first.

Jordan Dimov (A115). Published 14 November 2026. The numbers in the examples are illustrative, round and marked "say". Sources are at the end. This is a summary for learning, not advice.

The risks a desk watches

  • Market risk. Prices move against the position. A desk long power loses when power falls.
  • Credit risk. A counterparty fails to pay, or fails to deliver, and the desk has to replace the trade at a worse price.
  • Liquidity risk. The desk is right about the market but runs out of cash before it is proved right. This is the one beginners overlook, and the one 2022 made famous.
  • Operational risk. A trade booked twice, a nomination missed, a price feed that stopped at noon. It is real, but it is about process and systems more than markets.

The rest of this article is mostly about the first three.

Position limits

The simplest control is a cap on how much the desk may hold. Limits are set on the net position, by product and delivery period, because that is what moves the P&L. Position aggregation shows how that net figure is built from every trade in the book.

Say a desk may hold no more than 10,000 MWh net of power for any one month. It is long 10 MW of next month's baseload, every hour of a 30-day month: 10 x 720 = 7,200 MWh. It has used 72% of its limit.

When a trade would take the position past the limit, one of three things happens. The trade may be refused before it is booked. It may be booked, and the trader must bring the position back by a set time. Or a named person approves the breach in writing. Which one depends on the desk. What matters is that someone outside the trading team sees every breach.

A limit in MWh treats every megawatt-hour as equally dangerous. It is not: a megawatt-hour of next week's power moves far more than one of next year's. That is why desks also use a measure that accounts for how much prices move.

Value at Risk in plain terms

Value at Risk (VaR) answers one question: on a normal bad day, how much could this position lose? It is stated with a confidence level and a time horizon. "One-day 95% VaR of £35,600" means that on 95 days out of 100, the desk expects to lose less than £35,600 overnight. That holds only if prices behave as the model assumes: it is an estimate, not a guarantee.

Take the same position, 7,200 MWh long. Say the price of next month's baseload moves by £3/MWh on a typical day: a standard deviation of £3/MWh. If daily moves follow a normal distribution, then on 95% of days the price falls by less than 1.65 standard deviations. So:

  • a 95% bad day is a fall of 1.65 x £3 = £4.95/MWh;
  • across 7,200 MWh, that is 7,200 x £4.95 = £35,640, about £35,600.

That is the one-day 95% VaR. A desk sets a VaR limit as well as a volume limit, and compares the two every morning.

VaR is useful, and it hides three things:

  • How bad the bad days are. VaR says the loss will exceed £35,600 on about one day in twenty. It says nothing about how far beyond. Real power and gas prices also spike more often than a normal distribution allows, so the worst days are worse than the example suggests. For this reason, the Basel Committee's standard for bank market risk asks for a different measure: expected shortfall, the average loss in the worst 2.5% of outcomes, from a ten-day base horizon.
  • Liquidity. VaR assumes the desk can close the position at the market price. In a squeeze, it cannot.
  • Broken relationships. A position hedged with a related product looks safe while the two prices move together. VaR measured on calm months assumes they always will.

So desks run stress tests beside VaR: what happens to the book if gas doubles, if a large plant trips, if the hedge stops tracking. VaR describes ordinary bad days; stress tests describe the rare ones.

Why hedges fail

A hedge is a second position that gains when the first one loses. Energy derivatives explained works one hedge four ways. In practice, hedges fail for four reasons:

  • Basis. The hedge settles against a different price from the one the desk actually pays: another hub, another index, another delivery point.
  • Volume. The desk hedged what it expected to need, and needed something else. A supplier whose customers use more power in a cold snap is under-hedged exactly when prices are highest.
  • Timing. The hedge and the exposure settle in different periods, so the cash arrives in one month and the cost in another.
  • The counterparty. The hedge is only worth what the other side can pay. That is the next risk.

Credit and counterparty exposure

What a counterparty owes the desk has two parts. One is the value of open trades that have moved in the desk's favour. Say the desk bought at £80 and the price is now £90: if the seller defaults, the desk must buy again at £90. The other is energy already delivered and not yet paid for, which can be a month's invoices.

Desks set a credit limit for each counterparty, and ask for collateral or a guarantee when exposure grows. On an exchange, the clearing house stands between the two sides and collects margin every day, which removes most credit risk. It replaces it with something else: the need for cash.

2022: margin calls, and the cash behind a hedge

A future moves cash every day. When prices rise, a company that has sold futures to hedge its production must pay the difference to the clearing house, in cash, the next day. Its physical gas or power will be worth more when it is delivered, but not until then. The hedge is working. The cash is still gone.

In 2022, prices rose so far and so fast that companies with sound hedges struggled to find cash for the margin calls. In Britain, the Treasury and the Bank of England opened a £40 billion Energy Markets Financing Scheme in October 2022. It guaranteed bank lending for that purpose. Our derivatives article walks through how a margin call arises, step by step.

The lesson for risk management: measure the cash a position could demand, not only what it could lose. A desk that reports VaR but not its potential margin calls is watching the wrong number.

Hedge effectiveness, and why finance cares

Without special treatment, a company's accounts can show a hedge's gains and losses in a different period from the thing it hedges. A well-hedged business then looks volatile. IFRS 9 hedge accounting lets the accounts show the effect of the company's risk management instead. It applies only if three things hold. The hedge is formally designated and documented when it is set up. There is an economic relationship between the hedge and what it hedges. And the hedge ratio matches the quantities actually hedged. That is why finance asks the desk for paperwork when a hedge is booked, not at the year end.

The risk reports a desk reads every morning

A risk team starts the day with a handful of reports:

  • Positions against limits: every limit, how much of it is used, and any breach with its explanation.
  • P&L and its explanation: what changed overnight, split into new trades, price moves and corrections.
  • VaR against the VaR limit, by desk and for the whole book.
  • Credit exposure by counterparty, against each credit limit.
  • Stress results: the book under the scenarios the firm has chosen.

Each report is only as right as the position underneath it. A trade booked late, a price that arrived after the run, a unit that was not converted: each flows straight into the VaR and the limit check. Jordan Dimov built real-time position and P&L across six commodity desks at Shell, the numbers a risk report starts from.

Building the position and P&L your risk reports depend on? It is the work A115 does. See the systems work

What the full guide adds

This article is the short version. The full guide, Risk Management Fundamentals, goes to the depth a desk works at:

  • position limits by trader, commodity, delivery period and book, with a worked example;
  • VaR by historical simulation, calculated step by step;
  • counterparty exposure with netting, and how exchange clearing swaps credit risk for margin calls;
  • stop-loss and trailing stop-loss limits;
  • the 2022 crisis traced through one company's case;
  • the daily risk report and how it rolls up from trader to desk to firm;
  • Python: a VaR calculator and a position limit checker.

The full guide is in the community on Skool, with the rest of the series, and it is updated regularly as the markets change. If you want to work on a trading desk, read it to understand the subject properly. To join, start at the Energy Trading Fundamentals course page.

Sources