Why a generator hedged on its availability is really short, how large the true position is, and what a 10 per MWh rise does to it.
A power plant's hedge ratio is the volume sold forward divided by expected generation, not by available output. An illustrative 400 MW gas plant that sold a 350 MW baseload block reports itself 95.81 per cent hedged against its 8,000 available hours. It runs 4,200 hours, so it is 182.5 per cent hedged and net short 1,386,000 MWh, and a 10 per MWh rise costs it 13.86 million.
Worked in full in Energy Trading by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
Position reports at generators are often built on the asset register: capacity times available hours, against the forward book. That denominator is the most a plant could produce, not what it will produce. A gas plant runs only in the hours the spark spread pays, and those are fewer than half the year for many mid-merit units. Every MWh sold beyond expected output is a short position, and it does not show on a report that measures the hedge against availability.
This is a volume ratio for an asset hedged in its own market. It is not the minimum-variance ratio used when the hedge is a different contract from the exposure, which is worked in how to calculate the optimal hedge ratio for a futures hedge.
| Input | Value |
|---|---|
| Capacity | 400 MW |
| Available hours (after outages) | 8,000 |
| Expected running hours (spark spread positive) | 4,200 |
| Forward sale: baseload block | 350 MW |
| Hours in the year | 8,760 |
| Price move tested, parallel | 10.00 per MWh |
=(Cap*RunHours-BlockMW*8760)*Move.The report says the plant is 134,000 MWh long and gains 1,340,000 if prices rise 10. In fact the forward sale loses 30,660,000 on that move, the plant earns 16,800,000 more on what it actually generates, and the net is −13,860,000. The true short is 10.34 times the size of the open position the report shows, and of the opposite sign.
Splitting the year by hours shows where the short lives. In the 4,200 running hours the block delivers 1,470,000 MWh against 1,680,000 generated, so the plant is 210,000 MWh long in those hours. In the 4,560 idle hours the block still has to be delivered, 1,596,000 MWh with no generation behind it, all of it bought back at market. That is the position that loses money when prices rise.
The plant does not have to stay this short. A baseload block sold against a plant and bought back in the idle hours is a legitimate strategy, and the book shows it is worth exactly what the unhedged plant is worth when it is priced on its own hours. The problem is not the block; it is a report that calls the result 95.81 per cent hedged.
| Running hours | Generation (MWh) | True hedge ratio | Net position (MWh) | P&L on +10.00 |
|---|---|---|---|---|
| 3,000 | 1,200,000 | 255.5% | −1,866,000 | −18,660,000 |
| 4,200 | 1,680,000 | 182.5% | −1,386,000 | −13,860,000 |
| 6,000 | 2,400,000 | 127.8% | −666,000 | −6,660,000 |
| 8,000 | 3,200,000 | 95.8% | 134,000 | 1,340,000 |
Only at 8,000 running hours, a plant that runs every hour it is available, does the reported ratio become true. Every hour of merit-order reality between there and 4,200 adds to the short. To be fully hedged this plant should sell its expected generation, 1,680,000 MWh, which is the equivalent of a 191.8 MW baseload block, or better a shaped hedge concentrated in the hours it expects to run.
The table holds running hours fixed. In practice a rise in power against unchanged gas widens the spark spread and brings more hours into the money, so the plant's true sensitivity, its delta, is somewhat larger than its expected generation. That softens the loss; it does not change its sign. The spark spread that decides the hours is worked in how to calculate a clean spark spread.
Hedge ratio on availability. Availability is a technical number. The hedge has to be measured against expected dispatch, and expected dispatch moves with the curve. Recompute the running hours from the current forward curve and fuel prices every time the report is run, not once a year at budget time, and show the ratio on both denominators side by side so that the gap is visible to whoever signs the report.
Netting the cash with the economics. Even where the economics net, the cash does not. Sold on an exchange, the 3,066,000 MWh calls 30,660,000 of variation margin on the morning after a 10 rise; the 16,800,000 the plant gains is earned hour by hour over the year. The mechanics are in how to calculate variation margin on power futures.
Ignoring the fuel leg. A plant hedged on power but not on gas is still exposed to a rise in gas. One that bought gas sized for availability is long fuel it will not burn, a second mismatch running the other way from the power short. Count both legs on the same expected generation.
The reported and true shorts of the book's own plant are reconciled in the free workbook for this case, with the curve, the heat rate and the declared move as inputs.
Because a gas plant only runs when the spark spread is positive, so it generates far less than it could. The 400 MW plant is available 8,000 hours but runs 4,200, producing 1,680,000 MWh. Sell 3,066,000 MWh against availability and the extra 1,386,000 MWh is a naked short that has to be bought back in the hours the plant does not run.
Its expected generation, 1,680,000 MWh, the equivalent of a 191.8 MW baseload block, or better a shaped hedge sold in the hours the plant expects to run. The 350 MW block covers 1,470,000 MWh of the running hours, leaving the plant 210,000 MWh long there and 1,596,000 MWh short in the 4,560 idle hours.
Because the futures leg settles daily and the plant leg is earned only as power is delivered. A 10 per MWh rise calls 30.66 million of variation margin on the 3,066,000 MWh sold on the first morning; the 16.8 million the plant gains arrives over the year. The economic loss is 13.86 million, the cash need more than twice that.
This article is one calculation from Energy Trading. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: infrastructure, data centres and energy → · All 1340 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.