τTaosisTaoScale

Reference

Definitions

Every figure Taosis shows, what it means, the formula the code actually runs and where it is read from — the chain for all of them but the USD rate and the GitHub development counts, whose sources are named. A blank cell anywhere on the site means the figure could not be verified — never a zero standing in for one. The same text is available to machines at /llms.txt.

Every yield is shown twice, labelled: APR (simple) is one period’s return times the periods in a year; APY (compounded) compounds the same return. The difference, with worked numbers: APR vs APY on Bittensor.

Concepts

Subnet
One of the independent markets on the Bittensor network, each with its own α token, its own price and its own set of miners and validators.
α (alpha)
A subnet's own token. Bought and sold against TAO through the subnet's liquidity pool; staked to a validator on that subnet.
τ (TAO)
Bittensor's base token, in which every α is priced.
Pool
The subnet's constant-product liquidity pool holding τ and α; the price is the ratio of its reserves, and every buy or sell moves it.
Root (SN0)
Not a market: staking to root is one-to-one with τ at any size. Taosis hides it from the screener by default and blanks every pool-shaped figure for it.
Emission
τ the chain mints each block and splits across subnets — part injected into pools, part spent buying α when there is more τ than α to pair it with.
Conviction lock
α locked on a subnet for a chosen time; the longer and larger the lock, the more "conviction" it accrues, and conviction is what decides who can own a subnet.
Deregistration
When the network is at its subnet limit, registering a new subnet dissolves the existing non-immune subnet with the lowest moving price.
Fail-closed display
Taosis's rule: a figure it cannot verify is shown as "—", never as a zero standing in for one. A zero is always a verified zero.
APR (simple)
A yield annualized without compounding: one period's return multiplied by the number of periods in a year. On a subnet the period is one epoch (tempo blocks); on root it is the ~24-hour window Taosis measures.
APY (compounded)
The same return compounded: (1 + one period's return)^(periods a year) − 1. Always at least the APR. Taosis shows both side by side, labelled, and computes the APY from the APR with one shared rule — never measured separately.
Root yield
What root stakers earn on a validator. Root dividends go into the validator's basket fund, not straight into root stake; the chain tracks the fund's return in its BasketTwr series, and Taosis measures the yield from two readings of it about 24 hours apart. In τ.
On-chain event
A news item Taosis writes itself from the chain — a subnet registered or deregistered, an owner change, a halving, a large trade — linked to the block it happened in. No third-party source.

Price

Price (τ)
The spot price of one α of the subnet, in TAO: the ratio of the pool's two reserves.price = τ in pool ÷ α in poolSource: SubnetTAO and SubnetAlphaIn, read from the chain every few minutes.Root (SN0) has no pool: staking to it is one-to-one with TAO at any size, so root has no price.
EMA price (τ)
The chain's own moving-average price — the figure the runtime uses for emission and for the deregistration queue.Exponential moving average maintained by the chain (SubnetMovingPrice), never above 1 τ.Source: SubnetMovingPrice, read from the chain.A moving price of 0 means "not applicable" (no emission yet), not a measured zero; Taosis shows a dash there.
1h / 24h / 7d / 30d / 90d change
Percentage change of the spot price against the price sample taken that long ago.(price now − price then) ÷ price thenSource: Taosis's price_history table: the pool price sampled from the chain every ~5 minutes since 24 Aug 2026, and read back from the chain's archive node to dTAO's launch on 13 Feb 2025.Each window needs a sample near its start, after the subnet's current registration; 90d takes the nearest within 13 hours either side, because that far back the archive's samples may be one a day. Without one the figure is a dash, never 0, and the 30d and 90d columns are offered once some subnet has the figure. No third-party price feed is ever used.
7d sparkline
The last seven days of hourly closes, drawn as a line.Source: Taosis's price_history table.
Price chart (candles)
The α price as candles — open, high, low and close for each 15 minutes, hour, 4 hours, day or week — with the τ of the indexed buys and sells beneath each one.close = the bucket's last price sample; open = the previous candle's close when that candle is the one just before, else the bucket's first sample; high and low include the openSource: Taosis's price_history table — the pool price sampled from the chain every ~5 minutes since 24 Aug 2026, and read back from the chain's archive node to dTAO's launch on 13 Feb 2025 — and its index of on-chain fills for the volume.4H, 1D and 1W hold the whole history (from dTAO's launch, or from the subnet's current registration when that came later); 15m the last 7 days and 1H the last 90. Volume starts on 27 Aug 2026, when the fill index begins: an older candle has no volume bar, never a zero one. The USD view starts at the first CoinGecko TAO/USD rate Taosis stored.
Why the price moved (24h price attribution)
The last ~24 hours' price move split into its causes — τ traded into and out of the pool, τ and α emission injected into it — each shown as the move it alone would have produced, plus the residual that no named cause explains.ln(P₁/P₀) = ln(τ₁/τ₀) − ln(α₁/α₀), each side split into named parts; residual = the move − Σ named parts; each part's effect = e^(its log contribution) − 1Source: Taosis's own snapshots of the pool (SubnetTAO, SubnetAlphaIn and the emission injected) and its index of on-chain fills.Exact arithmetic on the pool ratio, not a fitted model. Flagged partial when the named causes explain under 90% of the move; unavailable, and saying why, when Taosis has no snapshot at one end of the window.

Size & flow

Market cap (τ)
The value of every α that circulates, at the chain's moving price.EMA price × (α in pool + α staked out − α burned). Root: spot price × α staked out.Source: SubnetMovingPrice, SubnetAlphaIn, SubnetAlphaOut and the burn counter, read from the chain.Subtracting burned α is what makes the figure agree with taoswap and taostats (under 0.6% apart on every subnet checked).
FDV (τ)
Fully diluted valuation: the price applied to the 21 million α maximum supply every subnet shares.EMA price × 21,000,000Source: SubnetMovingPrice.Not shown for root.
Volume 24h (τ)
τ of every buy and sell leg on the subnet over the last 24 hours — buys plus sells, including the legs the chain writes for stake moves and transfers.Σ τ of every buy and sell fill Taosis indexed in the windowSource: Taosis's own index of on-chain StakeAdded and StakeRemoved events.The chain writes stake moves and transfers as StakeAdded / StakeRemoved events too, so their buy and sell legs are counted, on both sides. Across subnets those legs are real pool trades; on the same subnet they carry no fee and never touch the pool — about a tenth of network volume on 28 Sep 2026. A verified 0 means the window is covered and had no fill; blank (—) means Taosis's block index cannot vouch for the window: a gap in it, the index more than 15 minutes behind, or the subnet registered again inside the window.
τ in 24h / τ out 24h
The two sides of the 24-hour volume, apart: τ of every buy fill (in) and τ of every sell fill (out) on the subnet over the last 24 hours, the buy and sell legs of stake moves and transfers included.in = Σ τ of buy fills; out = Σ τ of sell fills; in + out = Volume 24hSource: Taosis's own index of on-chain StakeAdded (in) and StakeRemoved (out) events.Stake moves and transfers are written by the chain as these events too, so they appear on both sides. Same coverage rule as Volume 24h: blank when the index cannot vouch for the window, 0 when it can and nothing traded.
Volume vs prior 24h
How the last 24 hours' volume compares with the 24 hours before them.(volume 24h − volume of the previous 24h) ÷ volume of the previous 24hSource: Taosis's index of on-chain fills, over two consecutive 24-hour windows.Blank when the previous window had no fill (a division by zero is not a percentage) or when either window is not covered by the index.
Turnover 24h
How much of the subnet's market cap changed hands in the last 24 hours.Volume 24h ÷ market capSource: Taosis's index of on-chain fills and the market cap above.Not shown for root. Moves and transfers count in the volume — same-subnet ones included, though they never touch the pool — so a busy week of restaking reads as turnover.
Buys, sells, buyers, sellers (24h)
The number of buy and sell fills on the subnet over the last 24 hours, and how many distinct coldkeys made at least one buy (buyers) or one sell (sellers).Source: Taosis's index of on-chain StakeAdded and StakeRemoved events.Coldkeys, not people. Blank when Taosis's block index has a gap in the window; a verified 0 when the window is covered and nobody traded.
TAO flow 24h (net staking)
τ that users staked into the subnet minus τ they unstaked over the last ~24 hours — stake, unstake, stake moves and swaps between subnets, and limit orders. The protocol's own τ is not in it: no emission injection, no chain buys of α, no root-basket trades.SubnetTaoFlow at the latest sample − SubnetTaoFlow at the sample nearest 24 hours earlierSource: SubnetTaoFlow, the chain's cumulative per-subnet counter, sampled by Taosis.Never rescaled to 7,200 blocks: the counter moves in lumps, so the window's real length in blocks is shown with it. Blank when no sample lies within 20 minutes of 24 hours ago, when the subnet was registered again inside the window, when the chain's flow EMA — which resets the counter — was written inside the window, or across a runtime upgrade Taosis has not audited.
TAO flow 7d (net staking)
The same user flow as the 24-hour figure — τ staked minus τ unstaked — over the last ~7 days.SubnetTaoFlow at the latest sample − SubnetTaoFlow at the sample nearest 7 days earlierSource: SubnetTaoFlow, sampled by Taosis.Same blanking rules as the 24-hour figure, over a longer window: a counter reset or unaudited upgrade inside the week blanks it for up to a week.
Net protocol flow 24h (τ)
Net τ the protocol itself moved into the pool over ~24 hours: emission injected, chain buys of α and root-basket buys, minus root-basket sells — validator rebalancing included.SubnetProtocolFlow at the latest sample − at the sample nearest 24 hours earlier (the TAO flow's window)Source: SubnetProtocolFlow, sampled by Taosis.Equals emission captured plus net basket flow. Same blanking rules as TAO flow, and blank for a subnet the chain has not started emitting to.
Emission captured 24h (measured, τ)
τ emission the subnet actually received over the last ~24 hours — measured, not projected from one block.Σ over Taosis's samples of (SubnetTaoInEmission + SubnetExcessTao) × blocks until the next sampleSource: SubnetTaoInEmission and SubnetExcessTao, sampled by Taosis.Blank when the samples do not cover the whole window, when two samples sit more than 300 blocks apart (the rate in between was not observed), or when the swap balancer held τ back or released it inside the window.
Net basket flow 24h (τ)
The net of the protocol's root-basket trades on this pool over ~24 hours: claim redemptions, validator rebalancing and deposit buys. Negative means these trades took τ out.net protocol flow 24h − emission captured 24hSource: SubnetProtocolFlow, SubnetTaoInEmission and SubnetExcessTao, sampled by Taosis.Not "root sells": buys and rebalancing are in it too, so it can have either sign. Blank when either of its two parts is.

Supply

Supply (α)
Every α that exists for the subnet: in the pool plus staked out.α in pool + α staked outSource: SubnetAlphaIn + SubnetAlphaOut.Burned α is still counted here, which is why supply × price sits above market cap on burn-heavy subnets.
Staked α
α held outside the pool — everything staked to validators, the owner's and locked α included.Source: SubnetAlphaOut.
α in pool
The α side of the subnet's liquidity pool.Source: SubnetAlphaIn.
Burned α
α destroyed for good on the subnet, with its τ value at the moving price.Source: The chain's AlphaBurned counter.A lower bound: the counter covers burns on one path since 2026-05-28.

Liquidity

TAO in pool
The τ side of the liquidity pool — the deeper it is, the less a trade moves the price.Source: SubnetTAO.Root's "reserve" is τ staked to root, not a pool; it is excluded from the network total.
ADR
Alpha dominance ratio: α held outside the pool for every α inside it. Above 1, holders own more than the pool could absorb, so a rush for the exit sells at a discount.max(α staked out − α burned, 0) ÷ α in poolSource: SubnetAlphaOut, AlphaBurned, SubnetAlphaIn.Still includes the owner's locked α and the protocol's own α.
Depth ±2% / ±5% (τ)
TAO of net buying that would move the price up by that much, and TAO of net selling that would move it down by that much.Up: τ reserve × (√(1+δ) − 1) ÷ (1 − swap fee). Down: τ reserve × (1 − √(1−δ)).Source: SubnetTAO and the subnet's own swap fee, read from the chain.The down figure is always a little larger than the up figure at the same distance: fee treatment plus the curve's convexity, not a bug. Constant-product pool assumed.
Swap fee
The fee the pool takes on every swap, as set on the chain for this subnet.Source: Swap.FeeRate per subnet.Read per subnet, never a global constant — all subnets currently read 0.0504%.
Sell-off price (τ per α)
The average τ per α if every α anybody holds outside the pool were sold into the pool at once, swap fee included, at the current reserves.sell-off value ÷ S, with S = α staked out − α burned − the protocol's own αSource: SubnetTAO, SubnetAlphaIn, SubnetAlphaOut, AlphaBurned, SubnetProtocolAlpha and Swap.FeeRate, read from the chain.S is the chain's circulating α: burned α and the α the protocol bought for itself are left out, because nobody can sell them. A stress figure, not a price anybody will trade at. Not shown for root.
Sell-off value (τ)
The τ that single sale of every circulating α would fetch in total.τ_pool · S(1 − fee) ÷ (α_pool + S(1 − fee)), S = α staked out − α burned − protocol αSource: The same chain reads as the sell-off price; constant-product pool, fee on the α sold.Blank when any input is unknown or when S is not positive.
Sell-off discount
The sell-off price against today's spot price: how far below spot the whole float would clear.sell-off price ÷ (τ in pool ÷ α in pool) − 1Source: The sell-off price and the spot price.Always negative. SN64 read −48.15% on 2026-09-25.

Emission & yield

Emission %
The subnet's share of all the τ the chain emits each block: τ injected into its pool plus τ the protocol spends buying its α.(τ injected + τ spent on chain buys) ÷ the same sum over every non-root subnet, from one blockSource: SubnetTaoInEmission and SubnetExcessTao, sampled by Taosis.Blank when the block's split could not be read. 0 is a verified zero: the chain allocated the subnet no τ that block — its miner burn is 100% (the runtime weights every share by 1 − burn), its share sits far below the emission gate (each share s is multiplied by 1 ÷ (1 + (θ ÷ s)^h), θ being the share ranked EmissionBarRank; see How emission is split), governance switched its emission off, or the chain has not started emitting to it (no FirstEmissionBlockNumber). The pairing cap is not a reason: a subnet whose allocation fits under it reads 0 in Chain buys, not here. Approximate while the swap balancer is holding back or releasing τ.
Chain buys (τ/day)
τ the protocol swaps into the subnet's pool each block to buy its α: the part of the subnet's τ emission above the α-injection cap (root proportion × α emission). The α bought accrues to the protocol (SubnetProtocolAlpha).SubnetExcessTao on one block × 7,200 blocks per daySource: SubnetExcessTao, from the same sample as the pool injection.One block's rate × 7,200, not a measured day — the measured figure is Emission captured 24h on the subnet page. Most subnets read a real 0.
Chain buys %
The subnet's share of every τ the protocol spends buying α network-wide.Source: SubnetExcessTao across non-root subnets.Spiky: the denominator moves too.
Root cut %
The share of the subnet's validator emission paid to people staking τ on root instead of to α holders on the subnet.γ·R ÷ (γ·R + α in pool + α staked out), with γ the chain's TaoWeight and R the τ staked on rootSource: TaoWeight, root's SubnetTAO, SubnetAlphaIn, SubnetAlphaOut — computed, never read from the chain's RootProp cache, which is stale on emptied subnets.A stake-weight split, not what root stakers were actually paid.
Miner burn %
The share of the subnet's miner emission captured by the owner's own neurons and destroyed instead of paid out.Σ over mechanisms of (mechanism weight × owner neurons' incentive ÷ all incentive)Source: Incentive per mechanism, MechanismEmissionSplit, the owner's registered hotkeys — read from the chain every 30 minutes.The owner does not pocket it; the chain scales the subnet's whole τ allocation by (1 − burn). The chain's own MinerBurned value is kept as an audit figure and can read 0 where the truth is 100%.
Default APR (simple)
The annualized yield of the subnet's default validator — among its paid validators holding at least 5% of the largest paid stake, the most-staked one whose yield is within 2 points of the best — from the dividends the chain paid its nominators last epoch. Simple: one epoch's return multiplied by the epochs in a year, with no compounding.APR = r × E, with r = D ÷ (TotalHotkeyAlphaLastEpoch − D), D = AlphaDividendsPerSubnet (the nominators' part, net of take) and E = 7,200·365 ÷ tempo epochs a year (7,300 at tempo 360)Source: AlphaDividendsPerSubnet and TotalHotkeyAlphaLastEpoch, read from the chain at one block every 30 minutes.In α: your return in τ also depends on the α price. Blank when no validator on the subnet was paid, or when the last reading is more than 2 hours old — not a claim of 0%. Parent validators paid through child keys are included. Not sourced from any third-party API.
Default APY (compounded)
The same validator's yield if every epoch's dividends stay staked and earn in turn — which is what the chain does: dividends are added to the stake every epoch. It is the APR (simple) compounded once per epoch, never a separate measurement.APY = (1 + r)^E − 1 = (1 + APR ÷ E)^E − 1, with E = 7,200·365 ÷ tempo (7,300 at tempo 360)Source: The Default APR (simple) and the subnet's tempo (SubnetHyperparams), read from the chain.Always at least the APR: 34.5% APR at tempo 360 is 41.2% APY. Holds the current epoch's rate for a whole year, so it is an extrapolation, not a forecast; in α, like the APR. Blank when the APR or the tempo is unknown; a verified 0% APR is a 0% APY.
Validator share
The share of the subnet's α emission its owner gives validators. Root stakers are paid out of it first; what is left goes to the validators' α stakers, so this share sets the ceiling of the APR.SubnetEmissionWeights[netuid] slot 3 ÷ u64::MAX; α stakers' share = validator share − min(validator share, 41% × root proportion)Source: SubtensorModule.SubnetEmissionWeights, read from the chain every 5 minutes.Set by each subnet's owner since the per-subnet emission weights (default 41%). SN73 cut it to 6% on 27 Oct 2026 and its nominators' APR fell from 38% to 0.24%. Blank until the sync has read it.
Emission per day (current rate)
Where the subnet's α emission goes each day at the current per-block rate: the owner's cut, the miners' slot, the part withheld from the owner's own miner slots, the validators' slot, and within it the root stakers' part and what is left for α stakers — each in α and in its τ value at the spot price.per block × 7,200, split by the subnet's own SubnetEmissionWeights [burn, recycle, owner, validator, miner] (default 0 / 0 / 18 / 41 / 41%). Withheld from miners = the incentive of the owner's own UIDs. Root stakers' part = total × min(validator share, 41% × the subnet's root proportion), paid off the top of the validators' slot; α stakers get the restSource: DynamicInfo.alpha_out_emission, SubnetEmissionWeights, the miner-burn share, the root proportion and Σ SubnetMovingPrice, read from the chain.Per-block rates × 7,200 blocks, not a measured day; τ values are the α at the spot price, not τ emitted. Each owner sets the split (17 subnets off the default on 2026-10-30; SN73 gives validators 6%): a validator share near root's cut leaves α stakers almost nothing. The chain pays root stakers only while the sum of all subnets' EMA prices is above 1; otherwise their part is recycled, as it is at an epoch where no validator qualifies for root dividends. The withheld miner α is burned or recycled per the subnet's RecycleOrBurn setting.
Injected into the pool (τ/block)
τ the chain adds to the subnet's pool each block, paired with α up to the injection cap — the pairing leg only, not the subnet's whole take.SubnetTaoInEmission; ≈ per day = one block's rate × 7,200Source: SubnetTaoInEmission, from the same sample as the chain buys and the emission share.See Emission % for the whole allocation. A verified 0 for a subnet the chain has not started emitting to, which receives nothing. While the swap balancer is holding back or releasing τ, this is not the block's own emission.

Validators

APR (simple), main subnet
A validator's yield to its nominators on its main subnet — the one holding its largest α stake by τ value, at least 500 τ — as the median of the last 24 hours of epoch samples. Simple: one epoch's return × the epochs in a year.median over 24 h of r × E, with r = D ÷ (TotalHotkeyAlphaLastEpoch − D), D = AlphaDividendsPerSubnet and E = 7,200·365 ÷ tempoSource: AlphaDividendsPerSubnet and TotalHotkeyAlphaLastEpoch, read every 30 minutes; the main subnet from TotalHotkeyAlpha × each subnet's τ price.In α. Epochs that paid nothing are left out of the median. Blank means no subnet with at least 500 τ of this validator's stake paid dividends in the window — not a claim of 0%.
APY (compounded), main subnet
The main-subnet APR (simple) compounded once per epoch at that subnet's tempo: what a nominator who never withdraws would earn in α over a year at the same rate.(1 + APR ÷ E)^E − 1, E = 7,200·365 ÷ the main subnet's tempoSource: The main-subnet APR and SubnetHyperparams.tempo.Blank when the APR or the tempo is unknown.
Root APR (simple)
What root stakers on this validator earned over the last ~24 hours, annualized without compounding. Root dividends, net of the validator's take, go into the validator's basket fund and reach stakers when they claim; the chain tracks the fund's return in one series, BasketTwr, which grows by every dividend at its value when paid.g × 7,200·365 ÷ span, with g = BasketTwr(end) ÷ BasketTwr(start) − 1 over a window of span blocks (~7,200) ending at the block shownSource: SubtensorModule.BasketTwr and BetaBaseline, sampled from the chain every hour by Taosis and compared with the sample nearest 24 hours earlier.In τ. 0.00% is a verified zero: no root dividend was paid into the fund in the window. Blank, with the reason on hover, when the chain holds no fund for the validator, when no reading from about 24 hours ago exists yet, when the fund started or restarted inside the window, or when no reading under 3 hours old exists.
Root APY (compounded)
The same root return compounded over a year, once per window: what a root staker who claims and restakes every day would earn at the same daily rate.(1 + APR × span ÷ (7,200·365))^(7,200·365 ÷ span) − 1Source: The Root APR (simple) and its exact window in blocks.Compounds over the ~24-hour window, not per epoch: root dividends reach stakers through claims, not through the epoch. Blank whenever the Root APR is.
Take
The share of the dividends its stake earns that the validator keeps before paying its nominators.Source: SubtensorModule.Delegates (a PerU16: parts of 65,535), read every hour.The chain caps it at MaxDelegateTake (18% on 2026-09-24). Every APR (simple) and APY (compounded) on the site is already net of it. The directory's Avg take tile is the plain mean over the rows on the page.
Total stake (τ)
Everything staked to the validator, valued in τ: its root stake plus its α stake on every subnet.root stake (τ) + α stake (τ)Source: TotalHotkeyAlpha on every subnet, root one-to-one with τ and α at each subnet's τ price, read every hour.Moves with α prices even when nobody stakes. Approximate (~) when part of its α could not be priced; blank when none of it could.
Root stake (τ)
τ staked to the validator on root (SN0).TotalHotkeyAlpha(hotkey, 0) — root α is one-to-one with τSource: SubtensorModule.TotalHotkeyAlpha, read every hour.0 is a verified zero: nothing staked on root.
α stake (τ)
α staked on the validator on every subnet, valued at each subnet's current τ price.Σ over subnets > 0 of TotalHotkeyAlpha(hotkey, netuid) × τ in pool ÷ α in poolSource: TotalHotkeyAlpha, SubnetTAO and SubnetAlphaIn, read every hour.Moves with α prices even when nobody stakes or unstakes. Approximate (~) when some of its subnets have no price: the sum then covers the priced ones only, and the root/α split bar is hidden. Blank when α is held but none of it can be priced.
Stake weight (τ)
A network-wide weight Taosis builds the way the chain weighs stake on a subnet: α stake (valued in τ, all subnets) plus root stake × TaoWeight. The chain itself weighs each subnet separately, in that subnet's α; no single chain value equals this total.α stake (τ) + root stake (τ) × TaoWeight (0.18 on 2026-09-25)Source: TotalHotkeyAlpha, subnet prices and SubtensorModule.TaoWeight, read in the same hourly pass.A Taosis aggregate used for Dominance, not a weight the chain uses in consensus. Blank when TaoWeight is unknown or the validator's α stake is only partly priced.
Dominance
A validator's share of the total stake weight of every delegate in the directory.stake weight ÷ Σ stake weight over every delegateSource: The stake weights above.Blank on every row when any delegate's weight is unknown: a share of an incomplete total would overstate everyone.
24h stake change
The change in a validator's total stake since Taosis's reading nearest 24 hours ago (22 to 26 hours back), in τ and as a percentage, with the root and α parts apart.total stake now − total stake at the reference reading; % = change ÷ the reference totalSource: Taosis's hourly readings of TotalHotkeyAlpha and subnet prices.The α part moves with α prices, so it is not only stake coming or going; the root part is exact τ. Blank during the first day after a validator appears, or when its stake could not be read in full.
Nominators
Coldkeys holding a non-zero stake on the validator, on any subnet.Source: The runtime's DelegateInfo (DelegateInfoRuntimeApi.get_delegate), counted for the 256 largest validators in hourly batches, oldest count first; the basket escrow account is left out.Blank for every other validator, and when the count is more than 24 hours old. Never summed across validators: one coldkey can stake on several.

Metagraph

Consensus
How far the subnet's validators agree on this UID: the stake-weighted median of the weights they set on it at the last epoch (Yuma consensus), from 0 to 100%.Source: The chain's Consensus vector, from the runtime's metagraph call for the subnet's first mechanism, read about every 30 minutes.The subnet page lists the top UIDs by incentive or dividend, not every UID.
Incentive
This UID's share of the subnet's miner emission at the last epoch, as the validators' weights decided it. The shares add up to 100% across the mechanism's UIDs.Source: The chain's Incentive vector, from the runtime's metagraph call for the subnet's first mechanism, read about every 30 minutes.A subnet running several mechanisms splits its emission between them; the screener's Miners and Miner burn % read every mechanism.
Dividend
This UID's share of the subnet's validator dividends at the last epoch — non-zero only for UIDs that validate.Source: The chain's Dividends vector, from the runtime's metagraph call, read about every 30 minutes.A share, not an amount: the α the UID received is its Emission, and what its nominators earn is the validator's APR (simple) and APY (compounded).
Emission (UID, α)
The α this UID was paid at the last epoch: its incentive as a miner and its dividends as a validator.Source: The chain's Emission vector, from the runtime's metagraph call, read about every 30 minutes.
Trust
Yuma consensus's trust score for the UID: the share of the stake-weighted weight set on it that survives consensus clipping.Source: The runtime's metagraph call.The runtime currently returns no trust vector, so the column is hidden rather than shown as a row of dashes; it comes back on its own if the chain exposes it again.

Hyperparameters

Tempo
Blocks between two epochs of the subnet: how often the chain runs Yuma consensus there and pays out its miners' and validators' α.Source: Tempo, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.360 blocks (about 72 minutes at 12 s a block) on most subnets. It sets the E in every APR (simple) and APY (compounded): 7,200·365 ÷ tempo epochs a year.
Immunity
Blocks during which a newly registered UID (miner or validator) is shielded from pruning. When the subnet is full, the next registration replaces the lowest-emission UID outside its immunity — unless only MinNonImmuneUids (10 by default) or fewer such UIDs are left, in which case the lowest-emission immune UID is replaced instead (with none immune, the registration is refused). The owner's own immune hotkeys (the owner hotkey first, up to ImmuneOwnerUidsLimit, 1 by default) are never replaced.Source: ImmunityPeriod, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.A UID's protection inside the subnet, not the subnet's own immunity from deregistration (Immune, under Lifecycle).
Max validators
How many UIDs can hold a validator permit on the subnet: the permits go to the UIDs with the most stake.Source: MaxAllowedValidators, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
UIDs
UIDs registered on the subnet now — miners and validators together.Source: SubnetworkN, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.A count of slots taken, not of active or paid neurons: Miners (under Lifecycle) counts the miners actually being paid.
Reg. burn
τ burned to register one new UID on the subnet right now. The chain moves it with registration demand, between the subnet's MinBurn and MaxBurn.Source: Burn, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.The price of a UID inside the subnet, not of a new subnet (Subnet registration cost, under Network).
Registration
Whether new UIDs can register on the subnet: open or closed.Source: NetworkRegistrationAllowed, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
Weights rate
Minimum blocks between two weight submissions by the same validator.Source: WeightsSetRateLimit, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
Activity cutoff
Blocks a validator may go without setting weights before the epoch treats it as inactive: an inactive validator's stake is left out of the consensus, so its weights stop counting.ActivityCutoffFactorMilli × tempo ÷ 1,000 blocks (at least 1)Source: The effective cutoff the runtime's epoch uses, as the metagraph call reports it, read about every 30 minutes.
Commit-reveal
Whether validators must commit their weights encrypted and reveal them later, so that no one can copy them before they count. The figure in brackets is the reveal period, in epochs.Source: CommitRevealWeightsEnabled and RevealPeriodEpochs, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
Serving rate
Minimum blocks between two updates of a neuron's serving endpoint (its axon: IP and port) on the chain. 0 means no limit.Source: ServingRateLimit, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
κ (kappa)
The stake majority behind Yuma consensus: each UID's consensus weight is the κ-quantile, by stake, of the weights validators set on it, and weights above it are clipped.shown raw, as the chain stores it (a u16); the fraction is κ ÷ 65,535 — 32,767 is one halfSource: Kappa, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.
ρ (rho)
The steepness of the trust sigmoid in earlier versions of Yuma consensus.Source: Rho, from the runtime's metagraph call (SubnetInfoRuntimeApi), read about every 30 minutes.Stored and reported by the chain, but the runtime's current epoch does not use it (Taosis reads the runtime source): it changes nothing today.

Ownership & conviction

Takeover %
How far the strongest challenger — the hotkey with the most matured conviction among those another coldkey than the owner's owns — has climbed toward the bar that lets it claim ownership of the subnet.challenger hotkey conviction ÷ (18% × (α staked out − protocol α − α burned)), from the runtime's lock.rsSource: The chain's conviction lock maps, read every 10 minutes.100% is not eligibility. The runtime also requires the subnet to be at least one year old, and hands the subnet only to the top-scoring hotkey of all: the owner's own locks count on the owner's hotkey, so a challenger over the bar changes nothing while the owner's side out-scores it (the subnet page shows that side). The bar moves as α is minted, so a gauge can slip back while nobody unlocks. The 10% figure published elsewhere is not the chain's.
Largest lock %
How much of the subnet's whole α supply the single largest conviction lock has locked up and matured.leader's matured conviction ÷ (α in pool + α staked out), projected forward to the current blockSource: The chain's lock maps and its maturity rate.The leader is usually the owner. Projected between Taosis's syncs, so it keeps maturing.
Locked α / Locked τ
α locked in conviction on the subnet — every lock, owner included — and what that α would actually fetch if sold through the pool in one go, after fee and slippage.Locked τ (realizable) = constant-product exit of the locked α against the poolSource: Conviction lock maps, SubnetTAO, SubnetAlphaIn, swap fee.The spot valuation is a fiction on thin subnets: a large stack can take a 20% haircut on exit.
Owner lock (α)
α locked in conviction on the subnet owner's hotkey — every lock on that hotkey, whoever placed it: the owner or any holder backing it.OwnerLock + DecayingOwnerLock locked mass (the perpetual and the unlocking leg, summed)Source: The chain's OwnerLock and DecayingOwnerLock maps, read every 10 minutes.The runtime counts a lock as the owner's by its hotkey (lock.rs), so a large figure shows support for that hotkey, not that the owner locked anything: at block 9,137,487 SN1's whole owner lock, 192,824.59 α, was one outside coldkey's. What the owner itself does is in Owner net sold and on its address. The τ under it prices the α at spot, not at what it would fetch (Locked τ).
Owner conviction (α)
The matured conviction of those same locks on the owner's hotkey: how much of the locked α already counts in full.Source: The conviction field of the OwnerLock and DecayingOwnerLock maps (fixed-point U64F64), read every 10 minutes.Grows toward the locked amount at the chain's maturity rate (about a 30-day half-life at current values); an unlocking lock also decays. Whoever placed them, as for Owner lock. They count in the chain's takeover contest, on the owner's hotkey: a challenger must out-score them as well as clear the 18% bar (Takeover %).
Hotkey locks Σ (α)
α locked in conviction on every other hotkey of the subnet, summed.Σ locked mass over the subnet's HotkeyLock and DecayingHotkeyLock entriesSource: The chain's HotkeyLock and DecayingHotkeyLock maps, read every 10 minutes.A sum across hotkeys, so not a takeover figure: a takeover needs one hotkey's matured conviction on its own (Takeover %). The τ under it prices the α at spot.
Registration lock (τ)
τ locked when the subnet was registered.Source: SubnetLocked, read every 10 minutes.What that registration paid — not today's price to register a subnet, which is Subnet registration cost.
Still missing to takeover (α, τ)
How much more matured conviction the strongest challenger hotkey (see Takeover %) needs to reach the takeover bar, and what that much α is worth at today's price.max(0, 18% × (α staked out − protocol α − α burned) − challenger hotkey conviction); τ = that α × spot priceSource: The chain's lock maps (every 10 minutes), SubnetAlphaOut, SubnetProtocolAlpha, AlphaBurned and the spot price.A distance, not a cost to act: buying that much α would move the pool, and new locks must mature before they count. Reaching the bar is not enough while the owner's hotkey out-scores the challenger (Takeover %). The bar moves as α is minted. Subnets whose α staked out cannot be read are left off the board, never ranked at 0; those whose eligible α reads 0 on chain are named as having nothing to take over.
Owner
The coldkey that owns the subnet on the chain.Source: SubnetOwner.
Owner net sold 7d / 30d
The owner coldkey's own pool trades on its subnet over the window: α sold minus α bought, and τ received minus τ spent. A negative figure reads "net bought".Σ α sold (the swap fee included) − Σ α bought; Σ τ received − Σ τ spentSource: Taosis's per-leg index of on-chain fills, classified by the same ledger as the holder P&L.Stake the owner moved between its own hotkeys is not a trade and is not counted; stake it transferred to another wallet is shown apart as Transferred out. Blank until Taosis's index covers the whole window after 24 Sep 2026 16:20 UTC (block 9,138,689, where coldkey-swap markers begin) — from 1 Oct 2026 for 7 days, 24 Oct 2026 for 30 days — and whenever the owner key arrived by a coldkey swap inside the window or the subnet was registered again inside it.
Owner transferred out 7d / 30d (α)
Same-subnet stake the owner coldkey sent to other coldkeys minus the stake it received from them. Nothing is sold in the pool; a negative figure reads "net received".Source: Taosis's per-leg index of on-chain stake transfers.Same coverage rules as Owner net sold.
Owner held (α)
The α the owner coldkey holds on its subnet, as the chain itself reports it at a stated block, every hotkey summed.Source: StakeInfoRuntimeApi, read every 6 hours by the holder profit pass.Blank when the owner is outside the subnet's 100 largest holders (not read), when the read is more than 8 hours old, or when a coldkey swap into the key happened since. 0 on a subnet with no α staked out at all (SubnetAlphaOut is 0).
Owner share of holdings sold 7d / 30d
How much of what the owner held it sold, net, over the window ending at the chain read of its holding.net α sold ÷ (α held + net α sold)Source: Owner net sold and Owner held above.Negative for a net buyer. Blank when the holding is unknown or the window is not covered.

Concentration

Holders
Distinct wallets (coldkeys) holding α of the subnet, above a 0.000001 α dust floor — the owner's coldkey included, like every other.Source: Taosis's scan of the chain's stake maps, converted from shares to α, refreshed every 10 minutes from indexed fills.Coldkeys, not people. Published only for subnets whose scan covered at least 98% of the α and reconciled with the chain's total — and 0, verified, on a subnet with no α staked out at all (SubnetAlphaOut is 0).
HHI
How lopsided the ownership is once the owner's own stake and the protocol's root-basket escrow are set aside: 0% evenly spread, 100% one wallet.Σ share² over holders, min-normalised: (HHI − 1/N) ÷ (1 − 1/N)Source: Taosis's holder scan, owner coldkey and root-basket escrow excluded. The escrow (5EYCAe5j…tp7s, the runtime's modlsubtensrbeta/esc account) holds validators' basket positions for root stakers: a protocol account, not a wallet.Min-normalised, so not comparable to the raw HHI most explorers publish.
Nakamoto
The fewest wallets that would have to act together to hold more than half the α, owner and root-basket escrow excluded.Source: Taosis's holder scan.Coldkeys, not independent parties. Shown as a dash below two holders.
Top 10 %
Combined share of the ten largest holders, owner and root-basket escrow excluded.Source: Taosis's holder scan.Blank below twenty holders, where it would read 100% for a reason that has nothing to do with capture.
Gini
Inequality of the same holder set, owner and root-basket escrow excluded: 0 perfectly equal, 1 one wallet holds everything.Source: Taosis's holder scan.
Verified / Unverified (holder balances)
Whether the holder balances Taosis publishes add up to the α the chain says is staked on the subnet. Verified: within 2%. Unverified: further apart, and HHI, Nakamoto, Top 10 and Gini show — until they match again.drift = Σ holder α ÷ TotalAlphaStaked − 1; verified when |drift| < 2% and the scan covered at least 98%Source: SubtensorModule.TotalAlphaStaked, checked on every watcher pass (about every 10 minutes), against Taosis's holder balances.Emission reaches holders without a trade, so balances fall behind between exact chain reads; a subnet that drifts is read again from the chain within the hour.
Concentration history
One point per UTC day of the holder count, Nakamoto, Top 10 share, HHI and total α: the subnet's published state at the end of that day.Source: Taosis's holder scan, recorded once a day from the day the history table was deployed; the chart states its first day.A missing day was not measured — never 0. Unverified days are marked. Days before the subnet's current registration are dropped: a reused netuid does not inherit its predecessor's holders.

Holder profit

Staking (holder)
α a holder earned by staking since the subnet's start block — the α the chain added that no trade explains — valued at today's pool price.α now − α at start − net α of every indexed buy, sell, fee and transfer; in τ: × today's pool priceSource: The holder's α read from the chain (StakeInfo) at the start block and at the last pass; Taosis's per-leg trade index in between; SubnetTAO ÷ SubnetAlphaIn.For the owner, a validator or a miner coldkey it also holds the owner cut, take and dividends, or incentive — the row says which. Below −0.001 α a trade is missing from the index, and the row is shown as unverified instead.
Trading P&L (holder)
Profit or loss from buying and selling the subnet's α since the start block: the total minus what staking earned.total − staking; realized by weighted average cost (WAC) over the α held at the start plus every indexed trade; unrealized = trading − realizedSource: Taosis's per-leg trade index and the chain's pool price.Counted from 1 Sep 2026 03:25 UTC (block 8,969,695), when the per-leg index starts — gains made before are not counted; α already held then is valued at the chain price then. A transfer to another coldkey leaves at its τ value.
Total P&L (holder)
Everything the position gained or lost since the start block, trading and staking together.α now × price now − α at start × price at start − τ paid in + τ taken outSource: Chain pool price at both ends; each trade's τ from the index. USD: every trade at its own TAO/USD rate (CoinGecko, used only for USD).Computed for the top 100 holders of each subnet every 6 hours. USD is blank when a trade's TAO/USD rate is missing or more than 3 hours old.
P&L status (holder)
How far a holder's profit can be trusted. Full history: it held no α at the start, so every α came through a trade Taosis indexed. Since 1 Sep: it already held α at the start, valued at the chain price then. Unverified: something could not be matched — a missing trade, a coldkey swap, a failed or stale read — and its profit is not shown.Source: The holder P&L job's own checks.A row computed more than 12 hours ago is shown as unverified.
TAO/USD move (holder)
The part of a holder's USD profit that comes from TAO's own price against the dollar, not from trading or staking: every τ that went in or came out is valued at the TAO/USD of its own day, and today's τ at the TAO/USD of the last computation.USD P&L − τ P&L × TAO/USD of the last computation = Σ τ flow × (TAO/USD then − TAO/USD at the flow)Source: The holder P&L job; TAO/USD from CoinGecko, used only for USD.It is why a τ loss can read as a USD gain. Staking has no TAO/USD part: it is valued at the same rate as the rest. Not a trading result.

Portfolio

Total value (τ)
What the wallet holds, valued in τ at spot: free and reserved τ, root stake, and every α position at its subnet's spot price.free + reserved + root + Σ α × (τ in pool ÷ α in pool)Source: The wallet's balance and stake and each subnet's SubnetTAO and SubnetAlphaIn, read from the chain at one block; prices at most 30 minutes old.Blank when any held subnet has no fresh price, rather than valuing that position at zero. Selling would fetch less: see Realizable if sold now.
24h price effect (τ)
How much the wallet's α positions gained or lost in τ from price moves alone over the last 24 hours.Σ over positions of α held × (spot price now − spot price 24 h ago)Source: The positions read from the chain and Taosis's own price samples.Price moves only: staking rewards, buys and sells are not in it.
Holder rank
Where the wallet stands among the subnet's holders by α held: 1 is the largest.Source: Taosis's holder scan (see Holders).Among coldkeys holding at least 0.000001 α, the subnet owner included. Blank below that dust floor or before the scan has counted the wallet.
Realizable if sold now (τ)
What the wallet would hold in τ if every α position were sold into its own pool right now: free τ, plus root stake, plus what each position's sale would fetch after the swap fee and slippage.free τ + root τ + Σ over positions of τ reserve − τ reserve × α in pool ÷ (α in pool + α sold × (1 − swap fee))Source: The wallet's balance and stake, and each subnet's SubnetTAO, SubnetAlphaIn and Swap.FeeRate, read from the chain at one block.An estimate: each position is sold alone against today's reserves, so selling several, or after someone else, fetches a different amount. Reserved τ is left out. Beside it, Total value prices every α at spot and reads higher.
Avg buy (WAC / FIFO, τ per α)
The average price paid for the α still held, over the buys Taosis indexed. WAC blends every buy into one running average; FIFO matches each sale against the oldest lot first, so the lots left decide the average.cost basis of the α remaining ÷ α remainingSource: Taosis's index of the chain's StakeAdded and StakeRemoved events, replayed by its ledger: a move between your own hotkeys carries its basis, a transfer in arrives at its τ value, a burn realises the loss.Only as complete as the index: marked ~N% when the indexed fills explain under 98% of the α held, and not shown under 10%, or for a position that predates the index.
Realized P&L (τ)
Profit or loss already locked in by sales: each sale's τ proceeds minus the cost basis of the α it sold, by the chosen method (WAC or FIFO).Source: The same indexed ledger as Avg buy.α sold with no indexed purchase behind it is left out rather than counted as profit, and the coverage says so.
Unrealized P&L (τ)
The indexed α still held, at today's spot price, minus what it cost.α remaining × spot price − its cost basisSource: The same indexed ledger, and SubnetTAO ÷ SubnetAlphaIn.Covers the tranche Taosis indexed, never extrapolated to the part it did not see. Blank when the price is unknown.
Staking earned since 1 Sep (τ)
α the wallet's stake gained that no trade explains, valued at today's price.α now − α at 1 Sep 2026 03:25 UTC (block 8,969,695) − every indexed buy, sell, transfer and burn since; × today's spot priceSource: The wallet's α read from the chain at block 8,969,695 (an archive node) and now; Taosis's per-leg trade index in between.Root is not included. The wallet total stays blank until every subnet it holds is verified; the verified part is shown apart.
Total P&L since 1 Sep (τ)
Everything the wallet's positions gained or lost since 1 Sep 2026 03:25 UTC, trading and staking together.value now − value at 1 Sep − τ put in + τ taken out; trading = total − stakingSource: The same chain reads and index as Staking earned.Blank on the same terms as Staking earned.
Correlation r
How closely two subnets' prices move together: the Pearson correlation of their hourly log returns.r = Pearson(ln(Pₜ ÷ Pₜ₋₁ₕ)) over the hours both series share in the last 90 daysSource: Taosis's price_history: its own samples of the pool price, and the chain archive's before they began.Blank under 48 common hours — too little overlap is not zero correlation — and when either series is flat. Only consecutive hours count, so a gap in sampling is never read as one hour's move.

Lifecycle

Dereg #
Position in the deregistration queue: #1 is dissolved by the next new subnet registration.Non-immune subnets ranked by the chain's EMA price ascending, oldest first on tiesSource: SubnetMovingPrice, NetworkRegisteredAt, NetworkImmunityPeriod.
Immune
Inside its registration immunity (about 120 days), during which no new registration can deregister it — with the block it lapses.Source: NetworkRegisteredAt + NetworkImmunityPeriod.
Cost to kill rank 1 (τ)
What registering a new subnet costs right now. With every slot taken, that registration dissolves the subnet at the front of the deregistration queue, so it is the price of removing rank 1.Source: The runtime's own registration price (SubnetRegistrationRuntimeApi), sampled about every 5 minutes.Blank when no sample from the last 30 minutes exists. "In 1 day" beside it is the same price 7,200 blocks later on the chain's straight-line decay, shown only when Taosis's formula reproduces the runtime's price at the sampled block.
P(dereg) 30d / 90d
The probability that the subnet is dissolved within 30 or 90 days if new subnets keep registering at the rate Taosis has observed: the chance that at least as many registrations arrive as its rank in the queue.P(Poisson(λ·t) ≥ rank), with λ = subnet registrations observed ÷ days covered by Taosis's block indexSource: Registration blocks read from the chain (NetworkRegisteredAt) that fall inside Taosis's own block index, and the queue rank."No estimate" — never 0% — when no registration has been observed, when the index covers under half a day, or when the subnet is immune. A model of arrivals at the observed rate, not a forecast.
Age
How long ago the subnet was registered on the chain.Source: NetworkRegisteredAt.
Miners
Miners actually being paid right now: neurons with a non-zero incentive score, counted across every mechanism, with the owner's own neurons taken out.Source: Incentive per mechanism and the owner's registered hotkeys, read every 30 minutes.
Mechs
Separate incentive mechanisms the subnet runs, each with its own miner set and slice of the emission.Source: MechanismCountCurrent (default 1).Almost every subnet has exactly one.

Network

Issued supply
Every τ that exists so far, out of the 21,000,000 τ maximum.issued share = TotalIssuance ÷ 21,000,000 τSource: SubtensorModule.TotalIssuance, read at a stated block.
Block reward
The τ the chain mints per block at the current issuance.1 τ ÷ 2^k, with k = floor(log2(1 ÷ (1 − issuance ÷ 21,000,000))); per day = × 7,200Source: Worked out from TotalIssuance with the runtime's halving rule (block_emission.rs).The chain's BlockEmission item is out of date (still 1 τ) and is not used.
Measured issuance (τ per block)
How fast TotalIssuance actually grew between two readings, per block, over the last 200 blocks, 24 hours or 7 days.(TotalIssuance now − TotalIssuance then) ÷ blocks between themSource: TotalIssuance, sampled by Taosis about every 5 minutes.Net of what the chain recycled or burned in the same blocks, so it runs below the block reward; the gap is shown as Recycled or burned. Blank when no sample sits within 30 minutes of the window's start.
Recycled or burned (τ per block)
τ the chain took back out of circulation per block over the window: what it minted less what TotalIssuance actually grew by.block reward − measured issuance, over the same windowSource: Taosis's TotalIssuance samples and the runtime's halving rule.Blank for a window that crossed a halving, where the block reward changed inside it.
Unissued
TAO the chain has not minted yet.21,000,000 τ − TotalIssuanceSource: SubtensorModule.TotalIssuance, read at a stated block.
Supply split
Issued τ in three parts, all read at the same block: staked on root, in subnet pools, and in wallets (not staked).root = SubnetTAO[0]; pools = Σ SubnetTAO[netuid > 0]; wallets = issued − root − poolsSource: TotalIssuance and SubnetTAO, read at one block.Wallets is a difference: blank unless all three were read, and blank if the reads disagree. It includes reserved balances and registration locks.
Next halving
The issuance at which the block reward halves next, how much τ is left to mint before it, and two estimated dates.the k-th halving happens at 21,000,000 × (1 − 2^−k) τ issued: 10.5M, 15.75M, 18.375M…Source: The runtime's halving rule and TotalIssuance; block time measured on the chain clock.A projection. At the full reward is the earliest the threshold can be reached; at the measured rate uses the net issuance above. Neither is a promise.
Subnet registration cost
The τ it costs to register a new subnet now: the runtime's own price.max(min lock, last lock × 2 − (last lock ÷ (NetworkLockReductionInterval × block reward in τ)) × blocks since the last registration)Source: SubnetRegistrationRuntimeApi at a stated block; decay inputs NetworkLastLockCost, NetworkMinLockCost, NetworkLockReductionInterval, LastRateLimitedBlock[NetworkLastRegistered] and the block reward (from TotalIssuance).The runtime scales the reduction interval by the current block reward: 115,200 × 0.5 τ = 57,600 blocks today, so the price falls by one last lock every 57,600 blocks until it reaches the floor. Right after a registration the price is twice what was paid. The decay and floor are only projected when Taosis's own formula reproduces the runtime's price at the same block.
Sum of α prices
The spot prices of every subnet's α added together, in τ.Source: SubnetTAO ÷ SubnetAlphaIn for every subnet, sampled by Taosis.A market gauge, not a price. The chain uses the sum of the EMA prices, not this one, to decide whether root stakers are paid.
Staking return simulator
What a stake on one subnet and validator becomes over 1, 3, 6 and 12 months if its yield stays at the rate measured now: α at APR (simple), α at APY (compounded), and their τ value at today's price and if sold into today's pool.APR (simple): α₀ × (1 + APR × t); APY (compounded): α₀ × (1 + APY)^t; entry α₀ from today's pool with its swap feeSource: The validator's median yield over the last 24 hours, and the latest pool reserves read from the chain.A projection at the validator's yield measured now, held constant, with the α price held at today's value. Not a forecast. The site cannot stake for you.

Development

Commits 7d / 30d
Commits on one branch of the GitHub repository named in the subnet's on-chain identity — the default branch, or the one the identity's link names — over the last 7 and 30 days before the last read.Source: GitHub's public commits feed of that repository, read every 6 hours.The feed lists the latest 20 commits only, so a busy repository reads "20+": at least that many. One repository and one branch — a team shipping from other repositories can read low. Blank when the identity names no repository, or an organisation rather than a repository.
Dormant
No commit on that branch in the 30 days before the last read.Source: The same commits feed.Measures one repository and one branch only.

Chain storage names refer to the Bittensor runtime (SubtensorModule and related pallets) on finney. Formulas are the ones in Taosis’s own code; where a competitor computes a figure differently, the caveat says so.