How every number is made.

Series A · Public attention

Source: Wikimedia Foundation pageviews, English Wikipedia article “Pokémon Trading Card Game”, human traffic only, CC0.

Unit: mean daily human views per complete calendar month. A month is complete only when every calendar day is present. The latest complete month is 2026-09; incomplete months are excluded.

For month t, xₜ = sum of daily views / number of calendar days. Statistics use these unrounded monthly means. Displayed daily means round to the nearest whole view; percentages, z and percentiles round to one decimal. Standard deviations round to whole views; ACF coefficients round to two decimals.

Change vs a year earlier = 100 × (xₜ / xₜ₋₁₂ − 1). A zero or absent comparison is unavailable.

12-month mean μₜ = (xₜ₋₁₁ + … + xₜ) / 12. The band uses population σₜ = √(Σ(xᵢ − μₜ)² / 12), from those same twelve months, and spans μₜ ± σₜ. Windows with fewer than twelve consecutive complete months are unavailable.

z vs 12 months = (xₜ − μₜ) / σₜ. A zero standard deviation is unavailable.

Percentile = 100 × count of complete-month means ≤ xₜ / count of complete months. Ties are included.

Vs all-time high = 100 × (xₜ / max(x) − 1).

Marked earlier highs are complete months greater than every other mean within eighteen months on either side, using the available neighbors at the beginning and end of the series. Only highs above the latest complete month are marked; their count makes the headline.

The year × month matrix uses the same monthly means. Cell intensity = xₜ / max(x); blank cells have no complete month.

ACF at lag k = Σᵢ₌ₖⁿ⁻¹ (xᵢ − x̄)(xᵢ₋ₖ − x̄) / Σᵢ₌₀ⁿ⁻¹ (xᵢ − x̄)², for k = 1…36, where x̄ is the mean across all complete months. ACF is held when the monthly calendar has gaps or the denominator is zero.

Each exported entry’s public attention comparison uses the two values in its opened sealed item: average daily article views for the recorded month and the same month a year earlier. Unavailable items have no comparison. Home shows current-history charts and derived measures separately.

Complete-month count is the number of complete calendar months. History length in years is the whole number of twelve-month intervals between the first and last complete month. Missing days count absent calendar dates inside that stated period.

This measures awareness. It does not measure demand, sales or price. Every view here comes from one source, so the panels are not independent evidence.

Series S · Printed supply

Source: The Pokémon Company’s own published annual and cumulative printed-card totals. Status: held until verified.

Unit: cards printed per year; cumulative totals in cards. Each annual period and total must be checked against the published source before any quantity appears.

Printed supply does not measure availability, demand, sales or price.

Series R · The Record

Format: cycle-record/1. Each entry is a UTF-8 JSON file; its fingerprint is SHA-256 over its exact bytes. Public opened items retain their exact sealed bytes. Closed items remain fingerprints only.

OpenTimestamps is an open, independent timestamping standard.

A new entry commits to the previous chain hash. Corrections are new entries. Earlier files are never rewritten. A seal time is by our clock; independent timestamp status is reported separately for every entry.

Entries count the real chain files. Weeks missed count unique recorded missed weeks. Revisions are zero because earlier entry files retain their exact bytes; corrections add new entries. The 52-week strip starts at the first recorded week and marks sealed, missed and future weeks.

Current timestamp status: Bitcoin timestamp complete.

entry_hash =
  SHA256(exact entry.json bytes)
chain_hash =
  SHA256(prev_chain_hash + ":" + entry_hash)

The two hashes in the chain formula are lowercase hexadecimal text. The first entry uses 64 zeroes as its previous chain hash.

What we do not do

We do not forecast prices, issue trading signals, accept paid placement, track visitors or advertise.