For each operating plant, we compute the generation-weighted average wholesale price it received over the trailing window from its real metered output: capture_p = Σ (output_h × price_h) ÷ Σ output_h Output is the plant's actual delivered energy, hour by hour. Price is the wholesale settlement price at the plant's location over the same hour. The sum runs across every hour in the window — so hours when the plant produced more weigh more heavily.

Per-plant captures are aggregated within {zone, fuel} by capacity weighting, using each plant's installed nameplate capacity: capture_z,f = Σ_p (capture_p × MW_p) ÷ Σ_p MW_p Capacity weighting reflects per-MW economics — what a typical megawatt of installed solar in a zone earns — which is the metric a developer evaluating new PPA pricing cares about.

The capture ratio compares per-zone capture to the flat 24/7 average wholesale price at the corresponding ERCOT hub: NORTH → HB_NORTH, SOUTH → HB_SOUTH, WEST → HB_WEST, HOUSTON → HB_HOUSTON, PANHANDLE → HB_PAN. These five zones each have their own ERCOT settlement hub, so the capture ratio for them is essentially a pure timing measurement: did the plant produce in expensive hours or cheap ones?

COASTAL is the exception — it's a MORA-defined planning zone that doesn't have its own ERCOT settlement hub. We use the nearest geographic hub as the comparator: COASTAL → HB_HOUSTON. The capture ratio for COASTAL therefore combines two effects: timing (when the plant produced) AND basis (how much the plant's local node prices at a discount or premium vs HB_HOUSTON). We'll decompose timing and basis as separate metrics in v2.

BESS / storage capture is in development and will ship in v2. The math is different — batteries charge during low-price hours and discharge during high-price hours, so output is bidirectional and the "capture price" concept becomes an arbitrage spread rather than a single weighted average. We're building a storage-specific energy-arbitrage metric to capture that. Gas and other thermal fuels are excluded because the matched universe is dominated by industrial cogeneration plants whose dispatch economics don't reflect merchant-market behavior. Curtailment is not separately accounted for — hours when a plant produced zero MWh don't contribute to its capture price, which is the economically correct treatment.

Refreshed daily. The plant census is updated monthly. Wholesale market data lags real time by 60 days due to ERCOT's settlement disclosure schedule. The trailing 12-month window therefore covers two months prior to today through fourteen months prior to today.