POST / MARKET
Demand and dispatch prices: the five-minute distinction
Regional demand can help describe the price environment without explaining the next five-minute price move. That distinction is visible in a same-region comparison of AEMO dispatch prices and regional demand: in each of the five NEM regions, the reported correlation between price and demand levels is higher than the correlation between their interval-to-interval changes. These are descriptive associations, not estimates of what demand caused or forecasts of the next dispatch price. D011 D012 D013 D014 D015
EXPLORE PUBLISHED EVIDENCE
Post revision 3 · each series keeps its own unit and source.
New South Wales: daily mean (MW)
Mean 7,514.09 MW · Range 5,213.7–9,061.34 MW · Coverage 288 / 288
Data details & source versions
The provenance below covers the whole chart (Sep 11–24, 2026), not only the selected day. Day-to-version attribution is not supplied by this published figure.
- Evidence
- D006
- Dataset
- aemo.load.dispatchis_regionsum
- Entity
- aemo.region.NSW1
- Whole-period coverage
- complete
- Source rights
- allowed_with_attribution
- Rows SHA-256
- a59002ea544c7188f82230cce95b05c818b874c92db384c25f671e6368e855e0
- Bundle SHA-256
- be43958c09c3992de4306d7a36144aeddbaca2e3b7c012339689f0816c196928
- PIT basis
- measured
Versions for the whole chart (14)
aemo.load.dispatchis_regionsum:2026-09-11/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-12/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-13/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-14/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-15/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-16/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-17/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-18/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-19/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-20/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-21/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-22/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-23/NSW1:v1aemo.load.dispatchis_regionsum:2026-09-24/NSW1:v1
View daily values and coverage
| Date | Mean (MW) | Min | Max | Rows |
|---|---|---|---|---|
| 2026-09-11 | 7,514.09 | 5,213.7 | 9,061.34 | 288 / 288 |
| 2026-09-12 | 6,552.15 | 3,926.03 | 8,356.12 | 288 / 288 |
| 2026-09-13 | 6,255.42 | 3,439.23 | 8,527.96 | 288 / 288 |
| 2026-09-14 | 6,884.9 | 5,088.53 | 8,800.75 | 288 / 288 |
| 2026-09-15 | 6,692.18 | 4,234.78 | 8,867.42 | 288 / 288 |
| 2026-09-16 | 6,890.56 | 4,156.34 | 9,474.94 | 288 / 288 |
| 2026-09-17 | 7,217.69 | 4,540.72 | 9,348.73 | 288 / 288 |
| 2026-09-18 | 6,849.38 | 4,089.25 | 8,655.14 | 288 / 288 |
| 2026-09-19 | 6,194.04 | 3,455.88 | 8,104.13 | 288 / 288 |
| 2026-09-20 | 6,075.78 | 3,308.81 | 8,433.54 | 288 / 288 |
| 2026-09-21 | 6,818.23 | 4,979.1 | 8,915.19 | 288 / 288 |
| 2026-09-22 | 7,627.99 | 6,406.86 | 9,482.14 | 288 / 288 |
| 2026-09-23 | 7,497.21 | 6,665.61 | 9,009.33 | 288 / 288 |
| 2026-09-24 | 7,082.11 | 5,122.58 | 8,981.61 | 288 / 288 |
First, define the window actually compared
The two AEMO series do not have the same latest window in the supplied, measured point-in-time dataset. The price series’ requested window runs from 12 September 2026 at 04:15 UTC to 26 September at 04:15 UTC, end exclusive. The demand series’ requested window runs from 10 September at 14:00 UTC to 24 September at 14:00 UTC, end exclusive; its measured coverage is partial D008. Their aligned comparison therefore runs only from 12 September at 04:15 UTC to 24 September at 14:00 UTC, end exclusive. It must not be presented as a complete overlapping 14 days. The later price observations cannot extend a comparison for which the corresponding demand observations are not in this bundle. D001 D006 D011
This is a Velryn-canonical, measured point-in-time dataset as of 26 September 2026, 04:55 UTC, derived from AEMO’s NEMWEB DispatchIS reports—not an external snapshot or a latency-adjusted simulation. Its demand endpoint is older than that as-of time. The measure pairs AEMO regional reference prices in AUD/MWh with regional demand in MW at matching five-minute intervals within each region. A correlation of levels asks whether intervals with higher demand also tended to have higher prices. A correlation of changes instead asks whether the two series tended to move together from one available interval to the next. Neither question, by itself, establishes causation or predictive performance. D011 D012 D013 D014 D015
Across this overlap, price-demand level correlations ranged from 0.571 to 0.711 across the five regions; correlations between consecutive interval changes ranged from 0.112 to 0.228. The distinction appears in every region. The figure compares the two measures region by region. D011 D012 D013 D014 D015
Four regions are complete but South Australia is not. This table shows the matched observations behind each regional estimate.
| Region | Matched five-minute pairs | Missing of 3,573 expected |
|---|---|---|
| New South Wales | 3,573 | 0 D011 |
| Queensland | 3,573 | 0 D012 |
| South Australia | 3,538 | 35 D013 |
| Tasmania | 3,573 | 0 D014 |
| Victoria | 3,573 | 0 D015 |
What changes for a reader watching prices?
The useful distinction is context versus movement, not simply that one region has a larger coefficient than another. Across all five regions, higher-demand intervals tend to coincide with higher-price intervals more strongly than demand changes coincide with price changes. A reader using demand to frame the broad regional price setting should therefore not carry that relationship unchanged into a five-minute price-move explanation. The same evidence that makes demand useful as context gives substantially less support to treating its latest change as a stand-alone account of the latest price change. D011 D012 D013 D014 D015
That is a boundary on interpretation, not evidence identifying the missing mechanism. These paired price and demand series cannot determine whether a particular move reflected generation, network conditions, bidding, or some combination. Nor do contemporaneous correlations test whether a demand signal available before an interval would have predicted its price. For that decision, the next step is to examine other interval-specific evidence and test a forecasting rule separately—not to assign an unexplained price move to a named cause. D011 D012 D013 D014 D015
South Australia also cautions against calling the comparison uniformly complete. Its aligned series has 35 missing matched intervals, and its separate requested demand window has 51 missing rows; those counts describe different windows. Its reported correlations use fewer observations than those of the other regions. The direction of the level-versus-change distinction still appears in its observed pairs, but this bundle cannot say what including the absent intervals would do to either coefficient. D013 D008
PRIMARY SOURCES
www.nemweb.com.au ↗ · Source published date unavailable
Editorial reference date: Sep 26, 2026, 12:00 AM UTC. This is separate from the publication time above; check source dates and later changes before acting.
Analyze this Post in Terminal →