Subnet Holders

Fetch who owns one subnet's alpha (#9557) — the top coldkeys by alpha held on that netuid, with each holder's share of the subnet total, how many hotkeys they hold it through, and whole-subnet aggregates (distinct holder count, total measured alpha, top5/top10/top20 concentration). The reverse index of /accounts/{ss58}/positions, which reads the same ledger one coldkey (one account) at a time. Distinct from /subnets/{netuid}/concentration, which computes its scalars off neurons.stake_tao and therefore sees REGISTERED UIDs only: this reads nominator_positions, keyed on (coldkey, hotkey, netuid) whether or not the hotkey holds a UID on the subnet, so alpha staked to UNREGISTERED hotkeys is included — on netuid 74, 92 hotkeys carry positions and 10 are registered there, which is the part no other public source reaches without a full-chain map scan. Valued as share_fraction x hotkey_alpha.total_alpha against ONE proven pool pass, and ranked in ALPHA rather than TAO: within a single subnet alpha is already a common unit, so there is no subnet_snapshots price join and none of its up-to-24h staleness — multiply by the subnet's alpha_price_tao for TAO. limit caps the returned rows (default 20, max 100); holder_count, total_alpha and the three concentration shares are computed across the FULL holder set and then sliced, never over the capped rows, because the top of a sum is not contained in the union of the tops of its addends. TWO STATES DECLINE rather than answer, both with holders:[] plus a degraded.reason and NULL counts: pool_totals_unproven while no hotkey_alpha pass is recorded complete (a partially loaded pool ledger silently UNDERPRICES holders rather than visibly dropping them, so the ranking would be plausible and wrong), and root_not_in_alpha_map for netuid 0, which SubtensorModule::Alpha does not cover at all. A zero in any count is therefore a measured zero, never a decline. Mainnet-only: neither source table carries a network dimension.

Path Parameters

Query Parameters

Response Body