Methodology

Active positions: backing still in the pool is locked principal, not an expense. The result does not mark unsold inventory to market or measure its opportunity cost. Accrued fees and eligible pending rewards are accounted for separately.

Allocated positions: selection removes the listing from the available pool, but its final economic outcome may not be known. Allocated backing remains separately disclosed exposure. Only settled backing enters the resolved-outcome expense. The account’s gross backing-loss counter is reconciled against settled plus allocated backing; it is not deducted again as a separate cost.

Settled positions: every settled listing contributes its recorded backing to the expense. If the spinner keeps, relists or receives the NFT through finalization, returned backing is principal and the depositor’s lost NFT receives the selected reference mark. If the spinner accepts ETH or FWA, no lost-NFT mark is charged for an NFT returned to the depositor. An unresolved NFT delivery can qualify the estimate, as explained below.

Withdrawn positions: returned principal and the original NFT are not new income. Any recorded early crown exit fee and covered gas remain expenses. A failed NFT return is kept as a separate delivery exposure.

Standard estimate: lost depositor NFTs use recorded allocation backing, without a universal 5% discount. The tracker does not reconstruct the wallet’s historical NFT purchase cost. Historical oracle and current collection-floor settings are alternative reference values, not proof of acquisition cost. Historical oracle uses the last eligible collection update before listing creation, no older than seven days by its observation time; cleared or unavailable references remain missing. Allocation backing is frozen at selection and is not an NFT resale price. The special Punk floor substitution applies to spinner prizes, not depositor NFT cost.

Credited ETH earnings already include credited crown payouts. Pending listing fees and the current crown holder’s uncredited pot are added from the reconciled balance bundle. Claiming ETH previously accrued does not earn it again. The fee-earnings USD display uses the latest available ETH/USD quote.

Depositor FWA includes cumulative credited rewards plus pending uncredited tokens. Claiming those tokens does not add a second reward; the calculation is based on earned amounts rather than the wallet’s current FWA balance. Credited tokens and pending balances are anchored together so a newer credit is not added to the older pending amount it already contains.

Earned FWA is valued using the latest available FWA/ETH quote, including tokens earned or claimed earlier. This is a current reference mark, not historical ETH proceeds, acquisition-date pricing or reconstruction of later token sales. Profiles may retain a saved quote if the current source is unavailable; its use is disclosed in Calculation details.

The depositor ranking uses the stricter reference calculation, reusing retained verified deposit snapshots where available. Reconciled allocated backing remains disclosed exposure outside resolved-outcome expenses; a numeric result does not mean every allocation has settled. Profiles can additionally show supported delivery scenarios or partial deposit estimates, sometimes from a different snapshot. The profile estimator checks required monetary inputs; the direct ranking reference calculation assumes its ready source bundle contains those amounts. A ready flag alone is not proof of field completeness.

Spinner reference PnL = ETH cashouts + marked FWA received + marked NFTs kept + buyer rewards − paid spin fees − VRF fees − gas.

Paid acquisition fees exclude pending escrow and refunded acquisitions; refunds are not subtracted a second time. VRF and covered transaction gas remain costs when paid, including pending and refunded requests. Spinner return % = reference PnL ÷ (acquisition fees paid + VRF fees paid + covered gas) × 100. This measures net return on spin spend, including recycled spending, not on ETH deposited into a wallet. Zero-cost records do not receive a percentage. Partial estimates carry the same partial qualification into their return; delivery intervals propagate both endpoints, not a guessed midpoint. Where transaction gas is unmeasured, the reference result and return exclude that cost and are labelled before unmeasured gas.

Automatic CryptoPunk winners. Only confirmed main-pool NFT economic settlements enter the winner directory (keep, keep-and-relist, depositor backing reclaim delivering the NFT, or timeout NFT finalization), regardless of backing or present wallet ownership. ETH/FWA exits do not receive an NFT floor mark. A persistent Punk-specific event index resumes from its saved block and exposes already confirmed winners while older coverage is still being recovered. A wallet appears once in the general ranking and once in the winner directory, with every kept Punk grouped in its card. The general overview shows up to five highest usable Spins PnL estimates; Punk winners receive no extra ranking slot. Under Standard estimate, kept Punk backing is removed from the total NFT mark and replaced by the number of kept Punks multiplied by the current native collection floor. Floor and backing are never added as two prizes. This applies to all confirmed kept Punks, not only those backed at 1 ETH. The backing at allocation remains separate information. Floor-valued awards are estimates, not realized sales. The native collection floor is shared and cached; a failed or invalid refresh retains the last valid quote and its date. Profile, ranking and winner card all apply the selected basis and gas. The simulator’s hypothetical mark never enters these results.

The default hourly chart, timeframe tiles and top-spinners ranking use the same successful requests by request time in the rolling interval ending at the activity snapshot. Chart intervals adapt to the selected window: 1 minute for 15m, 2 minutes for 30m, 5 minutes for 1h, 30 minutes for 6h and 1 hour for 24h. Intervals are anchored at the window’s start, which need not be a clock hour. The average is successful spins divided by window hours, rounded up; it is unavailable for incomplete coverage. Calendar history is a separate opt-in view from midnight to midnight UTC, so its daily total differs from the rolling 24-hour total. Its average excludes the current partial hour and uncovered hours, and is also rounded up. The ranking remains on the selected rolling timeframe while calendar history is open. Historical results reflect their stored source snapshot; unresolved requests can become fulfilled later. Ranked ETH spending sums acquisition fees, excluding VRF and gas.

With Recorded backing selected, kept NFTs including Punks use the exact allocation backing. Historical oracle uses the last eligible collection update before settlement, subject to the seven-day observation-age limit and source coverage. Current collection floors applies the available floor to each kept NFT in that collection. Neither reference proves a sale price. ETH and FWA payout choices always use their recorded cash/token receipts rather than also receiving an NFT mark.

The wallet’s biggest-wins list ranks settled payout contributions using ETH received, current FWA marks, or the selected NFT basis. It does not allocate aggregate gas, VRF or rewards to individual wins; those appear separately. An individual win with an unknown mark or unresolved delivery does not receive a numeric payout contribution; it is not treated as a zero-value prize. The wallet can still retain a supported partial estimate. Mean PnL per wallet, mean per spin and the result of the largest spinners are distinct statistics. None is inferred from the activity ranking.

Reference PnL marks FWA settlement proceeds and final purchaser rewards at the latest available FWA/ETH quote; it does not attempt to reconstruct later wallet trades. Holdings displayed separately are not added to PnL. Finalized closed-epoch rewards with verified reward-unit and pot coverage are counted once, including claimed epochs. Closed epochs awaiting finalization or complete source coverage are not yet included. Current-epoch provisional rewards are excluded. Actual economic PnL remains different from this protocol reference.

The total is derived directly from the two component numbers; it does not wait for matching blocks or a separate cumulative recovery. Gas and reward values are already included in the components and are not applied again. Zero and negative results are valid. Missing, null and non-finite values are not zero.

Existing component ranges propagate by adding their endpoints: [a, b] + [c, d] = [a + c, b + d]. A point is added to both endpoints. No midpoint or new uncertainty margin is invented. Incomplete and unvalued-outcome qualifications carry into the cumulative result.

For an identified terminal NFT delivery claim, the full-recovery estimate P assumes recovery at its selected mark M. The existing delivery scenario runs from P − sum(M) to P. A kept spinner NFT is already included in P, so its claim is not added twice. A depositor’s original and returned NFT normally cancel; failure to recover the returned NFT lowers that result by M. This is a conditional delivery scenario, not a confidence interval or a guaranteed bound on final profit. Unmeasured future claim or settlement gas can reduce the result further.

An allocated outcome without a final choice can leave a known subtotal while the remaining outcome is explicitly unvalued. Recorded backing describes its exposure and is not substituted for unknown proceeds. Missing claim marks also remain disclosed. Partial estimates can be shown without declaring them complete; pending onchain actions are distinguished from outcomes that the tracker has not yet reconstructed.

The latest usable component survives incomplete refreshes, timeouts and rate limits. If a component is genuinely absent, a previous usable cumulative estimate can remain with its original date; otherwise no total is invented. The profile’s Updated time refers to the oldest dated component currently combined, not the time of a refresh attempt. Component blocks, timestamps and saved valuation sources appear in Calculation details.

Spinner return on spend = displayed net Spins PnL ÷ (acquisition fees paid + VRF paid + covered normalized gas) × 100. The costs belong to that component’s snapshot. Recycled spending remains part of the denominator. This is not return on wallet funding, and there is no percentage for zero spend. Existing ranges propagate into percentages; missing gas remains excluded and disclosed.

The depositor wallet table currently shows ETH PnL rather than a depositor wallet ROI percentage. The Depositors EV percentage elsewhere is a different, forward-looking all-cashout fee-coverage scenario with expected backing consumed as its denominator. Spinner EV is expected payout divided by modeled all-in spin cost. Neither should be read as a wallet’s realized return or an APY.

Rankings sort the current reference result; an existing interval uses its lower endpoint. Profitable and Negative filters require the whole interval to have that sign. Changing valuation or gas settings can change the displayed result and order. All of these calculations concern the main pool only.

A numeric PnL is a usable estimate, not proof of an exact economic result. Its arithmetic can be correct under the selected assumptions while NFT reference values differ from actual purchase costs or sale proceeds.

Unresolved outcomes or NFT claims indicate an input or delivery exposure that needs clarification, not automatically a calculation error. One wallet can have several unresolved outcomes. The stricter depositor ranking may lack a complete result while the profile already shows a supported partial estimate or a retained snapshot. A count of unreconciled wallets is therefore not a count of incorrect PnLs or empty profiles.

Observed snapshot — 9 October 2026, 15:03 UTC (17:03 in Italy): 12 of 371 depositor wallets, or 3.23% — less than 3.3%, did not have a fully reconciled result under the stricter ranking checks because of 16 unresolved NFT outcomes or delivery claims. Some profiles could already show usable estimates. This is a dated snapshot count, not an average, a live error rate or evidence of 12 incorrect calculations. The denominator is depositor wallets, not all tracker wallets. Six additional depositors had numeric PnL without measured gas; that separate limitation is not included in the 3.23% figure. These observations do not certify every other estimate as economically exact.

Unmeasured gas is a separate limitation. Where no usable gas measurement exists, the result is shown before that cost; this does not mean the transaction was free. Holding everything else constant, adding an omitted positive cost lowers PnL. Measured units priced at the selected gwei are still a normalized estimate.

An older retained snapshot is not inherently wrong, but later activity may change the result. Check the displayed update time and coverage. Counts of affected wallets describe a particular snapshot and settings, not a permanent error rate; overlapping limitations must not be counted twice. None of these categories establishes a guaranteed percentage error.

There is no guaranteed universal maximum percentage error for estimated PnL. Close to break-even, a small difference in ETH can create a large relative percentage difference or reverse the sign. For example, a 0.1 ETH difference around a 0.05 ETH result is material even if the wallet’s total activity is much larger.

Complete history and precise valuation are different questions. Even with every event indexed, gas may be normalized, NFT acquisition cost may be unknown and NFT reference prices may differ from a realizable sale. A collection floor estimates a prize’s value; it is not a completed sale or a guaranteed executable price for that NFT.

Main uncertainty sources include missing or delayed source coverage, outcomes not yet linked to their settlement, NFT delivery claims, unknown NFT purchase cost, the selected NFT reference, FWA reward valuation and unmeasured or modeled gas. Latest NFT/FWA quotes can also differ in observation time from the accounting component. These limitations do not imply a fixed 5% discount, a maximum error band or a guaranteed direction of error.

Selection probability is listing weight / total active weight. The integer weight is floor(10³⁶ / backing in wei), so it is inverse to backing. Spin price follows weighted expected backing plus the current surcharge; VRF and gas are additional. EV recomputes the spin fee from the same pool aggregate snapshot as its odds, with VRF and request gas both normalized to the selected gwei. The separate latest-quote comparison uses the latest official VRF quote and excludes unknown transaction gas.

Full option EV uses the best modeled value of keeping an NFT, taking ETH, or taking FWA for each possible outcome. The headline spinner percentage is gross expected payout divided by spin fee + VRF + modeled request gas. 100% is break-even; subtract 100 percentage points for net EV (91.74% expected return means −8.26% net). Values are computed from the current scenario, not fixed targets. The “always take FWA” break-even price multiplier is modeled all-in spin cost divided by expected FWA value at entry, after conversion loss. It is an expected average; individual outcomes differ and no return is guaranteed. The budget simulator uses integer-wei division for the number of affordable spins, leaves any remainder unspent, does not reinvest payouts, and freezes its sampled outcomes and valuations for the four-second reveal. Draws are independent with replacement from the observed inventory; the simulated pool and probabilities stay fixed. If inventory coverage is incomplete, sampling is normalized to the observed listings and can differ from full-pool EV. The default best-choice scenario uses FWA-reported indicative collection floors, including the native Punk floor, with 1% FWA conversion loss and unchanged FWA price; the simulator can vary these assumptions. Headline spinner EV and the FWA break-even scenario exclude purchaser token rewards and resale costs. Additional settlement or claim gas is excluded by default; the simulator adds any extra gas units entered by the user. Wallet PnL separately includes verified final rewards and gas within its disclosed coverage. Incomplete inventory or missing NFT marks are labelled as a partial valuation. FWA execution costs are not silently treated as zero.

Chase pulls. The tiles show every CryptoPunk listing that is Active with positive selection weight and backed by exactly 1 ETH, regardless of depositor. Allocated, settled, staged and withdrawn listings are excluded. Count, images and combined probability use one retained inventory snapshot: the sum of the shown listing weights divided by the same total active weight. The contract total active weight is the denominator, not TVL; changes to any active inventory weight can change the displayed odds. A paused pool is labelled separately. Crown is selectable. The winner directory is independent and waits for the NFT-kept settlement; removal from Chase pulls is not proof of a kept win. A zero state requires complete inventory coverage, and failed refreshes keep the previous bundle with its timestamp.

Punks remain in the pool at actual backing, including 1 ETH positions. Their backing is not their collectible market value; the native Punk floor values the NFT option separately. The live EV uses the indexed pool, including executed backing reductions. FwaPunkLister reductions require a successful transaction; time passing does not lower backing automatically. The simulator reads the V2 Lister’s current step, interval and minimum. “One reduction” changes each matching managed position once; “Lister minimum” assumes each survives until its minimum. Both are conditional fixed-inventory scenarios, not forecasts. Unrelated Punks keep their backing. Every scenario recalculates inverse-backing weights, total weight, weighted backing, all listing probabilities and the spin fee; the normalized VRF estimate and chosen gas assumption are held constant. Projections pause if Lister positions and pool backings do not match. Non-atomic pool coverage remains labelled approximate. Executing Lister reductions also depends on crown protection, pending acquisitions and protocol state; these reduction constraints do not make the crown unselectable in a draw.

The depositor headline is an all-ETH-cashout fee coverage ratio: distributable acquisition fees plus any redistributed retained backing, divided by expected backing consumed. Its net margin is that ratio minus 100 percentage points. With a 5% surcharge and 1% owner fee, and retained backing routed to the protocol, coverage is 1.05 × 0.99 = 103.95%, or +3.95% net. This uses expected backing consumed, not total locked capital; it is neither APY nor a promised investment return. The before-gas baseline excludes depositor gas and token rewards. After-gas coverage = (distributable fees + redistributed retained backing − modeled replacement-deposit gas) ÷ expected backing consumed × 100. It keeps the same backing denominator, so after-gas net ETH = distributable fees + retained backing − backing consumed − gas. A fixed pool is assumed to replace one selected listing with one new deposit per draw. This replacement model does not multiply deposit gas by the expected spin waiting time. Entry gas uses a six-receipt median of 408,715 units from five BitMON and one Nakamigos staged deposits, normalized to 0.3 / 0.4 / 0.5 gwei. This small calibration does not cover every deposit path or ordinary active-state listing; approvals, later claims, withdrawals and future FWA rewards are excluded. The listing simulator uses a fixed-pool model, current NFT opportunity value, an explicit keep probability and per-listing fee share divided by selection probability, including the terminal draw.

Inventory refreshes must reconcile unique listing IDs, listing count, inverse-backing weights, total weight, weighted backing and total backing before replacing the cached snapshot. Failed checks retain the previous snapshot and timestamp. These checks do not make pagination atomic; equal-backing replacements can still be indistinguishable from aggregates, so individual odds and floor scenarios remain approximate. The EV breakdown separates the best cash option, CryptoPunk NFT uplift and other NFT uplift. Floors update independently of backing and can change expected return without a formula change. The official aggregate tier snapshot is used for tier probabilities.

Default normalized gas price: 0.3 gwei, selected by the dashboard owner. The 0.3–0.5 gwei buttons are a sensitivity range, not a measured weekly or monthly average. The overview shows how that range affects gross spinner return while holding the expected payout constant; depositor coverage is shown both before and after modeled replacement entry gas, using the same selected gwei. Wallet reference PnL uses measured role-classified direct-core gas units × chosen gwei × 10⁻⁹ ETH, including failed calls to those classified methods. Mixed or maintenance calls are tracked separately and not assigned to either role; approvals and routed transactions outside that direct-core coverage are excluded. The EV request-gas model uses 402,589 units: the median of five single-action requests in a six-transaction calibration sample; the sixth transaction contains five draws. It is not a protocol-wide average, and separate settlement or claim transactions are excluded. VRF is computed from the latest available parameters of the configured VRF service: floor(serviceGasEstimate × gasPriceWei × (10000 + marginBps) / 10000) + flatWei. Both costs use the same selected gas assumption. The service address is integration configuration, not automatically rediscovered: a protocol service migration requires an integration review. This scenario is not the executable live quote: congestion can materially lower live expected return. The latest official quote is shown separately, before transaction gas.

Gas belongs to its actual payer. Batched actions share one transaction receipt, and failed transactions require separate coverage. A sample is not a wallet’s measured history. When role gas is unavailable, the displayed result is before that unmeasured cost; a displayed gas deduction is limited to the disclosed coverage cutoff.

Sum of past selection odds Σpₜ and survival Π(1−pₜ) use the pool at each eligible processed draw. These describe the observed pool path, not an alternative evolving inventory. The sum is cumulative selection exposure, not the expected number of times one listing would be won: a listing leaves the pool at its first allocation. For example, two opportunities at 50% give a sum of 1 and a 75% chance of at least one selection along that path, not one guaranteed hit. Current odds × all past spins is not that history. Historical rows replay ordered V2 structural events, validate request weight anchors and winning ranges, and exclude known noncanonical records. The displayed mean wait 1/p assumes unchanged odds.

V2 previously split purchased tokens 10% to depositors, 10% to purchasers and 80% to burn. The staking & revenue section reconstructs ETH routing and purchased-token splits from their separate onchain configuration histories. These use different denominators. Routing, funding, reward eligibility and claims are separate stages. Historical rewards and burns use actual V2 events. The shared-token V1 receiver and V1 emission schedules are excluded from V2 accounting. The supply-burn panel is explicitly token-wide: initial supply minus current ERC-20 totalSupply plus tokens held at the dead address. Both components are read directly from the contract; an upstream aggregate burn total is never added to a dead-address balance. V2 buyback burn is a separate subset, not an extra amount to add. The displayed circulating-supply share staked uses total vault assets ÷ (1 billion FWA − token-wide burn). That supply includes staked, custodied and treasury tokens; it is not an independently certified free-float estimate. Queued exits remain included in total staked assets, and source reads can differ by block.