PokeSplain
Market Index.

PMI 2.0 starts at exactly 100 on its recorded inception timestamp. Inception is the first PMI calculation using genuinely captured stored observations available at that time. Input capture times and materialization times are preserved separately. No old PMI levels, synthetic history or retrospective price backfill are used.

PMI Core 100 covers English, raw, Near Mint, USD card references from Scarlet & Violet onward within the existing 890-card prospective discovery roster. It is a modern-card reference-price benchmark, not a measure of all Pokémon assets. Japanese, vintage, graded and sealed indexes are not launched.

Use the latest fully committed stored observation cycle available at the cutoff. Require one active printing, unambiguous card/set/collector identity, a complete activity window, no identity conflict, a PKMNPrices-supplied TCGplayer Near Mint market reference of at least $5, provider timestamp no older than seven days and PokeSplain check no older than 24 hours. Require at least three unique admitted completed eBay sales on at least two distinct days, explicitly Near Mint and matching the printing, over the preceding 14 complete UTC calendar days. Exclude conflicting listing IDs. Rank observed sold-dollar volume descending; break ties by exact asset ID. Select 100. Fewer than 100 defers inception or rebalance. The frozen discovery roster is a documented coverage limitation, not the entire market.

Initial weights are proportional to admitted 14-day observed dollar volume, capped at 5% per asset; iteratively redistribute excess among uncapped constituents. Weights are fixed for the constituent period. Monthly rebalance occurs at the first eligible official close in the new UTC month, only with a valid outgoing close and 100 eligible incoming assets. Both baskets are retained. The outgoing index level becomes the new anchor, and new base prices are set at that cutoff. The index never resets to 100 after inception.

Within a period, PMI(t) = anchor level × Σ[fixed weight(i) × current reference price(i) / period base price(i)]. Period weights are allocation weights; current holdings drift with price. Sales-count changes alone cannot change the headline. Close-to-close return is PMI(t)/PMI(previous valid adjacent close) − 1; gaps remain unavailable. Contributions in index points are anchor × weight × price change / period base price. Dividing each contribution by the previous index level gives its percentage-point contribution to the index return.

Daily valuation cutoff is 22:00 UTC. The background worker records the close after 22:15 using only source observations materialized and checked by the cutoff. There is no intraday indicative index. Every observation, including unavailable closes, is immutable. All 100 assets must have a comparable fresh reference: same provider, source and printing; provider age ≤7 days and check age ≤24 hours. Otherwise no headline level is published for that close. No zero substitution, missing-value rebasing, indefinite forward fill, or reweighting around missing assets. Fresh/stale/unavailable counts and weights remain visible. Source references can represent older market information despite a recent successful check.

S&P 500 uses Yahoo Finance ^GSPC, the actual USD price index, not SPY and not a total-return index. Gold uses SPDR Gold Shares (GLD); silver uses iShares Silver Trust (SLV), explicitly exchange-traded proxies, not spot metal. Yahoo provides unadjusted session closes and session timestamps. Bitcoin uses Coinbase Exchange BTC-USD, the last trade in the completed 21:00–22:00 UTC hourly candle. All market/bucket timestamps and retrieval times are separate. Public endpoint availability is not guaranteed; source failure is shown and never replaced with synthetic values.

Only dates with a valid official PMI close and all four actual benchmark observations enter the comparative chart. US instruments use their actual exchange session close; PMI and Bitcoin use 22:00 UTC on that valuation date. This is same-date close alignment, not simultaneous ticks. Weekends and holidays are omitted from cross-asset comparisons, never forward-filled. Each series is rebased to 100 on the same first eligible date. Calendar-period controls require a common start date on or immediately before the target, no more than four calendar days earlier, and display the dates actually used. Two common dates are required. No period predating inception is enabled.

PMI excess return is PMI period return minus the selected benchmark period return, expressed in percentage points. If called alpha, it is simple excess return, not risk-adjusted CAPM alpha. The $10,000 view multiplies normalized historical growth by $10,000. PMI is not an investable security or ETF and excludes transaction costs, spreads, taxes, grading, marketplace fees, storage and liquidity constraints. ETF expenses may already affect their observed prices. It is not financial advice.

Compare valid constituent reference prices with the preceding observation, reporting advancing, declining, unchanged and unavailable. Percentages use available comparable constituents. The advance/decline ratio is unavailable when decliners are zero. After two official closes with complete breadth: at least 60% advancing = BROAD ADVANCE; at least 60% declining = BROAD DECLINE; otherwise MIXED. These are descriptive participation labels, not forecasts or trading advice.

Activity is separate from the headline. Count admitted exact-printing, explicitly Near Mint completed listings in the preceding seven complete UTC days and the preceding seven complete days. Dollar volume sums their reported USD sold prices over 14 days. Shipping and final settlement are not independently verified. Deduplicate marketplace listing IDs; exclude identity conflicts. Display collection coverage, active constituents and acceleration only with full window coverage and a positive comparison denominator. This is observed marketplace activity, not every Pokémon transaction.

7-, 30- and 90-day momentum require valid official PMI closes exactly that many calendar days apart. Missing endpoints yield unavailable. Volatility is the sample standard deviation of 30 daily log returns from 31 consecutive valid daily closes, unannualized. One day never produces a volatility statistic. Card-versus-PMI excess return can be derived from preserved exact-asset prices on aligned valid dates; no card-alpha ranking or recommendation is launched.

Methodology v1.0, constituent periods, source observations, official observations and rebalances live in new PMI 2.0 storage. A future method change needs an immutable effective-dated version explaining old and new rules. Source revisions are flagged separately; no official history is overwritten. A reviewed correction must append provenance linking the original record and replacement; automated correction is not enabled. Old PMI tables and cron remnants have no write path into this storage.