​As more batteries enter the NEM, owners need a way to distinguish whether performance reflects the market, the asset, or the operator. Riverina 1 and Riverina 2 are near-identical batteries, sharing a New South Wales connection point, owner, duration and market conditions, yet Riverina 1 earned around $14k/MW more in FY26.

Revenue on its own does not explain a gap like that. Benchmarking the two assets against each other and against the wider fleet does. Ko brings together fleet revenue, capture rates, value lost, cycling, local prices and bid behaviour to show where the gap opened.

​This article explores the gap across the New South Wales fleet, a like-for-like comparison of Riverina 1 and Riverina 2, and the single day that drove most of it.

Benchmark your asset against the wider fleet with Ko

Riverina 1 only stands out once its performance is placed against the New South Wales fleet. Ko first shows how each battery earned through FY26, and which markets contributed to revenue.

​Energy arbitrage dominated the fleet, usually 95% or more of revenue. Riverina 1 earned $76,285/MW, the most of any battery that ran the full year and almost entirely from energy. Wallgrove is the FCAS outlier, earning around $14,500/MW there, close to a fifth of its total.

Riverina 1 captured 72.3%, ahead of Darlington Point on 64.9% and Riverina 2 on 63.3%, then Wallgrove 53.7%, Capital 51.0% and Broken Hill 50.0%. Still-commissioning assets sit far below the field, Waratah lowest at 6.5% after an extraordinary $273,672/MW lost to unavailability.

Uncaptured revenue, value inside the adjusted potential that a battery simply did not realise, is the largest single loss for almost every full-year asset. The Riverina cluster shares an almost identical constraint signature of around $24,600/MW, pointing to a common transmission bottleneck on that corridor. Riverina 1 also carries a large loss-factor haircut of $15,579/MW, reflecting the region’s below-unity marginal loss factor.

Use Ko to identify when the performance gap opened

Once Ko has separated structural losses from trading outcomes, Riverina 1 becomes the most useful benchmark for Riverina 2. The two assets share the same connection point and duration, but have different operators.

Riverina 1 ran 322 equivalent full cycles to Riverina 2’s 272, leading every month from July to December before Riverina 2 took the lead from January to April. On revenue the two separate steadily from July, and by the end of January Riverina 1 was $14,825/MW ahead. From February the gap barely moves, finishing the year at $13,974/MW.

Trace the gap back to one trading day

January was the month where the comparison became most useful. Riverina 2 discharged more than Riverina 1 during the month, but still lost $2,907/MW of ground.

Most of that gap came from 10 January. It was the highest-value day of the year for New South Wales batteries, and it added $3,363/MW to Riverina 1’s lead.

Riverina 1 charged to roughly full by 10:15, held through the afternoon, then discharged steadily from about 17:40 and was near empty by 20:00. Riverina 2 settled at a high state of charge from around 15:30 and only discharged through the evening peak between 19:00 and 20:20, emptying by the end.

Both batteries saw identical local prices, because they share the same network constraint. From 16:00 they were heavily constrained at around -$1,000/MWh even as the New South Wales price climbed past $300. Through the spike the regional price reached $11,938/MWh while the Riverina local price barely moved, near -$1,000, before the constraint eased from 19:45. Local price governs whether a unit is dispatched; it is not the settlement price.

Riverina 1 stepped its offer down from $326.79/MWh to the -$1,000/MWh market floor at 17:40, and held there until it emptied around 20:05. Riverina 2 held a $1,298.89 offer and did not commit to the floor until 18:30. Both units sat at the floor together from around 19:00.

What asset owners should take away

Riverina 1 and Riverina 2 looked like a clean like-for-like comparison: same connection point, duration and regional prices. Once Ko separated the structural factors from the operating outcome, the remaining gap traced to how each asset traded rather than where it sat. Most of it built up over the first half of the year, and 10 January shows the mechanism most sharply.

Three points matter:

• Benchmark against similar assets. Revenue alone is too blunt. Capture rate, lost value and comparable assets show whether performance reflects the market, the asset or the operator.
• Separate structural losses from trading outcomes. Constraints, loss factors and availability set the opportunity; uncaptured value shows how much of it was left behind.
• Trace high-value days back to dispatch and bids. Riverina 1 reached the market floor 50 minutes earlier on 10 January, and in a constrained corridor that timing was enough to take the limited headroom first.

Ko makes the workflow repeatable: benchmark against the fleet, compare with similar batteries, then trace the gap to the day, price signal and bid behaviour behind it.