NSW early-signal twin

Zenith-fit ranking formula

Ranking weights are an assumption (ours), stated against Zenith's stated limits (16 Oct validation: load ≤5 MW · CMD ≤4 MW · connection <10 MW). The bands, gates and multiplier values are ours. Findbot's own v2 spec_score is shown on each card for reference only.

zenith_score = rank_score (Findbot) × load_fit × timing × headroom × upgrade

• load_fit, re-tagged from est_mw (a proxy estimate), not from Findbot's label: in range (our band) 1.5 ≤ est ≤ 10 MW ×1.0 (Zenith limits: load ≤5 MW · CMD ≤4 MW · connection <10 MW) · small 0.5 ≤ est < 1.5 MW ×0.6 · est < 0.5 MW (< 500 kVA): not ranked; listed under Pending MW recalibration while est_mw still comes from the old 200 ha ≈ 5 MW proxy, and excluded only once the lead's est_mw has a recalibrated basis · Phase 1 bridging play (CMD ≤4 MW while waiting for transmission; BESS sizing unverified, hypothesis; Findbot transmission-scale, or est > 10 MW) ×0.4: kept in the ranking but below the in-range (our band) estates, with no lift · unclassified (no est_mw, not transmission-scale) ×0.5, so an unknown can't outrank sized sites · not yet checked (file has no load-fit fields, e.g. v1) ×1.0
• timing_window: intercept ×1.0 · too-late ×0.3 (badged) · too-early: removed from the ranking and listed in the Watchlist
• headroom: zone-substation headroom < 5 MVA (or flagged < 5) ×1.0 · otherwise ×0.7 when known
• committed HV upgrade within 18 months: ×0.5, badged penalised: upgrade lands ≤18 mo
• Neutral-on-missing: any factor not yet checked counts as ×1.0 and is flagged not yet checked. With empty fields the Findbot order is unchanged.

The ranked list has fewer than 20 rows because of the Pending MW recalibration list and the Watchlist; nothing is backfilled. Sort: hard leads first, then zenith_score (highest first). Soft leads stay labelled soft. MW figures are estimates (basis shown). Context: Zenith's stated limits (16 Oct validation): load ≤5 MW · CMD ≤4 MW · connection <10 MW. The ranking band is ours, not Zenith's.