Whose Bias Is In Your Curve?

Four ways to be biased

Our view, in one line: there is too much solar to absorb in the middle of the day, so the rest of the day has to pay for the fleet — and there is so much short storage that the evening peak gets shaved and spread instead of spiking. The rest of this page is the evidence, in four steps.

One: there is too much solar for the middle of the day

In 2025 South Australia's daytime market was at or below zero 63% of the time, and Queensland 58%. Five years earlier it was 16% and 11%.

There is a simple reason the two go together. A renewable generator that is being curtailed is, by definition, willing to run and not being taken. It is on the margin — which means it is setting the price, at its own short-run marginal cost. For wind and solar that cost is zero, or negative once a certificate is attached, and there is no reason to bid above it: bidding higher only loses you dispatch you would otherwise have won.

So heavy midday curtailment and a $40 midday price cannot both be true. If a forecast shows one without the other, something in it is not letting renewables set the price.

At noon in 2030 there is 19.6 GW of utility solar and wind available. Operational demand is 14.4 GW, and must-run coal takes 4.8 GW of it — a third — before renewables get a look in. Storage charges 5.1 GW, about as hard as it can at that hour. 5.3 GW has nowhere to go.

By 2040 storage charging has plateaued near 9 GW while spill climbs to 15.7 GW. Storage stops growing as a midday sink well before the surplus stops growing.

Two: so the rest of the day has to pay for the fleet

The solar window falls to $8/MWh by 2030 and $4 by 2040. Everything the fleet failed to earn there has to be recovered somewhere, and the somewhere is the other seventeen hours. Midnight roughly doubles, from $76 to $142.

This is the whole seesaw, and it is not a solar story. If renewables set the price for a third of the day, the hours that are left are worth more. A model that never lets renewables set the price does not simply overstate solar — it understates wind, storage and peaking gas by exactly the same mechanism.

Solar earned 106% of the flat price in 2017 and 38% in 2025. Wind has barely moved, from 87% to 72%.

Three: short storage shaves the peak and spreads it

The old evening was a tall, narrow peak: $288/MWh at 5pm in 2025, against $22 at 10am and $86 overnight. There is so much short-duration storage proposed that the short window is exactly what gets crushed.

By 2030 the peak is $184 and it lands at 8pm; by 2040 it is $136 at 9pm. The value does not disappear — it reappears either side. Midnight goes from $97 to $142, and 3am from $86 to $114.

Read the ratio above each panel. The peak was 2.8 times the daily average in 2025 and is 2.0 times by 2030. That is the tall, thin spike giving way to a broad shoulder — worth far less to a 0-4 hour asset and far more to a deep one.

Four: until the storage runs out

Scarcity is not sprinkled evenly through the year. It is what a renewable drought looks like from the price side — and a drought lasts days, not minutes. The useful question is therefore not how often prices spike, but how long they stay up.

Events of an hour or less carried 76% of scarcity value in 2017 and 23% in 2025, and our house view keeps them under a quarter. Over-four-hour events were absent in most early years and run at 42 to 47% across 2028-30.

The year-to-year mix is lumpy — 2019 is the historic exception, where two long events carried nearly half the year's scarcity value on their own. That lumpiness is the point. Scarcity is concentrating into a handful of long episodes instead of spreading across hundreds of short ones.

That is the difference between a market whose scarcity comes from a unit tripping and one whose scarcity comes from the weather. A tripped unit is back within the hour. A wind drought is not.

It also decides what your storage is worth. A 0-4 hour asset was built for the left-hand side of this chart, and the left-hand side is shrinking.

And then the caps get tested

A cap contract is a promise to cover the market above $300/MWh. Written against a two-hour battery it is comfortable through a thirty-minute event and impossible through a six-hour one.

Storage bids just under gas to win dispatch, so while it has energy it caps what gas can charge. Once it is empty that discipline disappears — and the operators who sold caps are now forced buyers with nothing to deliver from. If you own gas length and your competitors are empty and short, the cap is simply the rational bid.

That last step is where a decade of merchant value sits. A model that sprinkles volatility across the year cannot produce it, because a sprinkle has no memory. It does not know the storage is empty, or that the drought is on its fourth day.

So: whose bias is in your curve?

Two questions. Both answerable from the file you already have.

1. What share of your daytime intervals sit at or below zero? Compare it with what the market actually did last year. Near zero means your midday is worth more than the market says — and every other hour less.
2. How long are your scarcity events? If volatility can be changed without anything physical changing, it is an assumption. Frequency is the easy half; duration is where storage revenue lives.

Every chart here is rebuilt from AEMO dispatch data and our own published run. Take the free Prices sample and check them yourself.