Methodology & data sources
Last reviewed: 2026-07-21
TCGOdds separates three kinds of evidence: catalog facts, market observations and community opening samples. They answer different questions and are never blended into a single invented probability. This page documents the same rules used by the build pipeline and the visible tables on the site.
Current dataset coverage
Catalog snapshot: 2026-07-27.
Card and set catalog
Card names, product identifiers, rarities, collector numbers, set dates and sealed products sync from TCGCSV's public mirror of the TCGplayer catalog. Riftbound is category 89. Every generated set file records the exact endpoint, source title and access date used for that build.
Pages are generated only for sets that exist in the explicit data registry. A missing dataset therefore produces no public page rather than an empty template. Browse the current records in the complete Riftbound card database.
| Set | Generated-data sources | Accessed |
|---|---|---|
| Vendetta | 2026-07-27 | |
| Unleashed | 2026-07-27 | |
| Spiritforged | 2026-07-27 | |
| Origins | 2026-07-27 | |
| Origins: Proving Grounds | 2026-07-27 |
These links and access dates are rendered directly from each generated set file. They update with the data sync rather than being maintained as a separate editorial list.
Market prices and trends
We store TCGplayer low, mid, high and market values separately for Normal and Foil printings. A daily snapshot is stamped with TCGCSV's upstream update date, not the local clock, so a delayed job cannot label yesterday's prices as today's. Trend charts compare those dated snapshots; they do not interpolate missing days.
Market price is a reference point, not a guaranteed sale price or investment forecast. Implausible negative values and extreme outliers above the integrity ceiling stop the build instead of reaching a page.
Observed pull-rate evidence
Riot does not publish official Riftbound booster odds. We therefore report what was observed in citable openings, always with the pack count and source link. Full box or case videos are preferred because the denominator is visible. Creator terminology such as “Overnumbered,” “Signature” or “textured rune” is preserved rather than forced into a catalog rarity when the mapping is not documented.
| Set | Observed packs | Cited sources |
|---|---|---|
| Unleashed | 192 | 3 |
| Spiritforged | 72 | 3 |
| Origins | 96 | 3 |
Reported claims from sellers or preview material appear in a separate “Reported pack odds” block with a caveat. They never enter the observed-sample calculation. If evidence is insufficient, the page says data collection is in progress.
Every observed source carries a product-origin label. Riot-provided review product, pre-release third-party product and verified post-release retail product remain separate in the visible sample summary, including strict per-pack samples. When a creator does not disclose how the sealed product was obtained, we say “Origin not disclosed” instead of assuming it was retail.
How the denominator is handled
Strict rarity samples can be pooled only when the source gives a pack count and compatible rarity tallies. The displayed per-pack rate is total observed hits divided by total observed packs; “one every N packs” is its reciprocal. Premium box-opening tallies remain visible as individual boxes when the creator uses treatment labels that cannot be safely mapped to catalog rarities. This prevents a clear but differently labelled video from being discarded without pretending the categories are interchangeable.
Every result is a finite community sample. A 96-pack observation can show what occurred in those openings, but it cannot prove the factory's exact collation rate. More packs and independent sources narrow that uncertainty; they do not turn the estimate into an official publisher statement.
Official previews and upcoming sets
Upcoming-set pages use officially announced material and catalog listings only. Each confirmed fact or revealed-card record cites a source ID and verification date. Rumors, anonymous leaks and unreleased artwork are excluded even if they appear plausible. A preorder listing can confirm that a product is offered, but it does not by itself confirm every card, treatment or pack configuration.
Refresh schedule
| Dataset | Cadence | What triggers review |
|---|---|---|
| Market prices | Daily | Missing date, invalid value or upstream delay |
| Cards and sets | Daily sync; launch-day check | New group, rarity or catalog field |
| Pull-rate observations | When a source is reviewed | New cited opening with a usable denominator |
| Official previews | During reveal season | New official announcement or gallery update |
Build-time quality controls
- Every dataset must contain a source list and capture date.
- Declared card and sealed-product totals must match their actual records.
- Unknown rarities, duplicate product IDs and dangling set references stop the build.
- Pull-rate samples require a source URL and a positive integer pack count.
- Official-preview facts and revealed cards must cite a known source ID.
Corrections, independence and limitations
TCGOdds is independent and is not endorsed by Riot Games, UVS Games, TCGplayer or any marketplace. Card images are reduced-resolution identifiers, not downloads. Pull-rate results describe finite samples and should not be presented as official manufacturing odds. Prices can change after the displayed snapshot.
Found a bad count, stale source or treatment we classified incorrectly? Send the page, disputed value and a checkable source to [email protected]. We correct the underlying data or rule so the fix survives the next sync.
Methodology change log
- 2026-07-21 — Published the source hierarchy, refresh schedule, sample denominator rules and visible integrity checklist.
- 2026-06-16 — Added source-linked official-preview records with separate confirmation and reveal fields.
- 2026-06-11 — Established daily price snapshots, explicit dataset registration and build-blocking integrity checks.