Top Holders

Fetch the balance-based top-holder leaderboard: every account (coldkey) with a nonzero free balance and/or delegated stake position, with free/delegated/total TAO columns matching the taostats-style Account/Free/Delegated/Total benchmark /api/v1/accounts explicitly cannot derive. Sort by total_tao (default), free_tao, delegated_tao, or cross-subnet stake flow over a window (net_flow_7d, net_flow_30d, net_flow_90d, #6886/#6887); limit caps the list (default 20, max 100). free_tao is sourced from a direct System::Account chain-state scan (not event-reconstructed, so it can't drift); delegated_tao is this account's own stake positions across every hotkey/subnet, VALUED IN TAO: every position is non-root and therefore denominated in its subnet's alpha token, so each is multiplied by that netuid's alpha_price_tao from the latest daily subnet_snapshots row before summing (#8803) -- a DAILY cadence, so the price can lag up to ~24h behind the live economics tier, and a netuid with no usable price is excluded from the sum rather than counted as zero. total_tao adds it to free_tao, which is valid because both are TAO; net_flow_* is StakeAdded minus StakeRemoved over the window -- a negative value is a real net outflow, not a missing value -- and null-valued rows sort last on the net_flow_* keys. TWO TIERS, SELECTED BY SORT (#9469/#9502). The net_flow_* sorts are LIVE: the top-holders-flow projection lane recomputes all three windows daily from chain.account_events, so captured_at advances and the ranking is a real one. The same lane composes the three holdings sorts from the store -- free_tao from account_balances, delegated_tao by valuing each nominator_positions row as share_fraction x hotkey_alpha.total_alpha x that netuid's alpha_price_tao (#9502 captured the pool totals; neurons.stake_tao covers only hotkeys holding a UID on that exact subnet, 22.8% of position rows, so recomputing from it would drop real top holders out of the ranking), and total_tao as free + delegated ranked across the FULL tables rather than summed over the other two sorts' capped rows, since the top of a sum is not contained in the union of the tops of its addends. A holdings sort is served live ONLY while its producer's most recent pass is recorded COMPLETE: ranking over a partially-loaded ledger returns the largest values PRESENT rather than the largest that EXIST, which is a well-formed leaderboard quietly missing real top holders, and for the pool ledger a missing total silently UNDERPRICES rather than dropping a row. While an input is unproven that sort DECLINES and answers from a fixed materialization taken 2026-08-02 whose captured_at and last_updated do not advance -- an account that has moved TAO since is misreported and one first funded since is absent, so read that ranking as historical and use GET /api/v1/accounts/{ss58}/balance for the live per-account balance. Which sorts are currently live is a property of the artifact, not a fixed list. On a net_flow_*-sorted page the three holdings columns are null rather than zero, because a zero would assert an empty wallet.

Query Parameters

Response Body