The contract that assumes a crystal ball

A top-bottom spread contract (TBx) settles on the gap between a day's most and least expensive hours: TB4 pays the spread between the top four and bottom four priced hours, on a nominated capacity, whatever the asset does. Developers take these contracts because they turn merchant volatility into firm revenue a lender can finance against. The catch is structural: the index ranks hours after the day is over, so it prices a battery with perfect foresight, and no real battery has it. The gap between the index and what an asset earns is therefore the central number in structuring one of these contracts, and it has been hard to see. The modelling that exists mostly comes from platforms, advisers and counterparties with a commercial interest in the transaction.

This index makes the gap public. It is computed entirely from primary AEMO data: every bidirectional battery unit in the National Electricity Market, its actual five-minute dispatch, and the regional reference prices it faced, for the full quarter.

The index, defined in one paragraph

For each region and calendar day (NEM time, period-ending), the hourly price is the mean of the hour's twelve five-minute regional reference prices. The TBx daily settlement per nominated megawatt is one hour times the sum of the top x hourly prices minus the sum of the bottom x hourly prices; the quarterly index is the sum over complete days. An asset's actual revenue is the sum over five-minute intervals of its dispatch (megawatts, negative when charging) times the price, times one-twelfth of an hour: energy-only, before FCAS and other streams. The capture ratio is actual revenue divided by the index at the asset's registered capacity, matched to its duration (TB1 under 1.5 hours, TB2 to 3 hours, TB4 above). The reproduction script linked at the end of this page recomputes every index value from AEMO's public archive.

The Solar Analytica TBx index per region, Q1 2026, 90 complete days each, alongside the actual energy-only revenue per MW of each region's full-quarter battery fleet. Computed from AEMO five-minute regional reference prices and unit dispatch; index methodology as defined above.

South Australia's spread is another market entirely: its TB4 index ran 3.7 times Queensland's, which is why the same contract is worth radically different firm revenue in different regions. In every region, the perfect-foresight index also ran 2.7 to 3.8 times what the full-quarter fleet earned from wholesale energy. A TB4 swap written at nameplate on the average real asset would have owed the offtaker several times the asset's energy revenue this quarter.

The capture ledger

Per asset, the picture is a wide distribution, and duration is its backbone:

Selected full-quarter units from the 60-unit dataset (40 ran the full quarter; 20 commissioning units are flagged partial and excluded from all statistics). RT ratio is the observed AC discharge-to-charge energy ratio at the connection point, auxiliaries included. Units are identified by their AEMO dispatch identifiers; full ledger with every unit, both capture bases and sources in the public dataset.

The median full-quarter unit captured 46 per cent of its duration-matched index: perfect foresight, imperfect dispatch, round-trip losses and competing value streams cost the average asset more than half the index. The exception is WDBESS1 in Queensland, which returned 1.03 times TB1, evidence that five-minute volatility capture and second cycles can out-earn a daily top-bottom-one-hour ranking; that retained upside is what makes these contracts survivable. Meanwhile the largest unit in the NEM fleet, the 1,095 MW battery at Waratah, sat commissioning through the quarter with net negative energy revenue. Twenty of the 60 units, over a third of registered capacity, are still ramping.

The ledger also checks itself physically. The median observed round-trip ratio of the full-quarter fleet is 0.81 to 0.83 in every region independently, in the range expected for lithium-ion plant. Hornsdale, the oldest unit in this dataset, shows both the lowest round-trip ratio among the longer-serving units and a capture ratio well below the new fleet's median, consistent with an asset that takes a large share of its revenue from the frequency services this index deliberately excludes.

Structuring, in checkable numbers

The index turns structuring intuitions into checkable numbers. Duration-matched nomination is the first screen: a 1.5-hour asset facing a TB4 settlement is short the index by construction, and this quarter that mismatch alone separates a 0.35 capture from a 0.62 one. The foresight haircut can now be sized empirically, since the market's real capture distribution, median 0.46 with an upper decile near 0.7 for energy-focused assets, is a better basis for a swap's nominated quantity than any single optimised model. Region matters more than structure: the same TB4 contract that would have paid A$36,000 per MW in Queensland paid A$133,721 in South Australia this quarter, and the choice of region moved nearly four times more money than the choice between TB2 and TB4 anywhere outside it.

The distribution also puts a number on what a modelled capture claim asserts. With a fleet median of 0.46 and an upper decile near 0.7, a financing case modelled above 0.7 capture is a claim to top-decile performance across the operating NEM fleet, and one above 1.03 is a claim to beat every battery in the market this quarter. None of this says TBx contracts are mispriced as a structure. It says the gap they are priced across is now observable from public data. (Residential storage is a different market, covered in the behind-the-meter data and the lithium-ion decade.)

Sixty units, and the ones left out

The fleet is every unit registered bidirectional in AEMO's registration data effective in the quarter, 60 units, 11.2 GW and 20.4 GWh; pumped hydro and paired wind-farm load units are excluded by construction. Dispatch is AEMO MMSDM DISPATCH_UNIT_SCADA (five-minute, as-generated at the connection point); prices are AEMO five-minute regional reference prices. The limits are worth reading closely. Revenue here is wholesale energy only: FCAS and other ancillary streams, often the larger share for early assets like Hornsdale, are excluded, so capture ratios understate total asset economics by design. SCADA revenue is unadjusted for marginal loss factors and any settlement nuances, making it an estimate of energy market receipts, not a settlement statement. The TBx index is Solar Analytica's own transparently defined benchmark; real contracts differ in definitions, floors and windows, and no figure here describes any actual contract or any party's financial position. Commissioning units are flagged and excluded from statistics. Every input is public, the methodology is published above in full, and the per-unit dataset is queryable. Q1 carries summer volatility, though: the index is built to be republished each quarter, and the multi-quarter record will be the one to rely on.