Methodology
How Pokélytics is built
We summarise stored prices; we do not predict future prices. Scores and labels use your stored price series only, not off-platform sold counts. This page is what the model can and cannot see.
Pokélytics · Independent market interpretation for Pokémon TCG collectors and hobby investors. · Last updated June 23, 2026
Three separate lenses
We never collapse these into a single "good / bad card" verdict. Each lens answers a different question, and the scanner lets you sort by whichever fits the moment.
- 01 · Market interest — how worth watching the card is right now on stored prices (recent moves, activity magnitude, tape freshness). This is the primary badge score.
- 02 · Pattern clarity (v2) — how clean and stable the historical pattern looks (momentum, pullback timing, volatility / jumpiness). Shown in drivers when you dig in.
- 03 · Activity — raw move magnitude in the series. High activity feeds interest but is not the whole story.
- 04 · Risk tags — shorthand from reasons (e.g. elevated volatility vs narrative trend).
Sparse history is shown, not hidden
Few listings do not turn into fake volume in our data. When history is thin or confidence is low, we say so, and the default scanner uses floors so noisy cards do not crowd the top of the list.
What we use — and what we don't
- Use for scores: stored listing prints for the selected variant lane (condition / printing), calendar horizons for % moves, and modeled scores from those features.
- Use for market context: SilphCo verified sold counts, sales volume, sales history, TVWAP, and grading/slab data on integrated cards — shown in buyer-activity and verified-market panels.
- Don't do: predict prices or issue buy/sell calls. We do not scrape TCGPlayer or eBay directly; listing and verified-market data come through JustTCG and SilphCo.
Simple vs technical view
By default, Pokélytics leads with plain-language tiers — Very active, Quiet lately, and so on — not raw 0–100 scores. The read and price behavior come first; numbers support the interpretation when you want them.
On the scanner (and in card views), turn on Technical detail to show numeric levels, pattern clarity, attention, and confidence columns. Same data — less jargon by default.
Inspect the data
On any card with a chart, export the stored series as CSV from the Chart panel — audit what the model saw. The export contains every price point we have for that variant, in the order the model read them.
Built for people who watch many cards
Scanner, portfolios, and alerts (Confirm / Pullback / Invalidate) tie the read to price levels you can track. Still not investment advice.
Frequently asked questions
Does Pokélytics predict future Pokémon card prices?
No. Pokélytics summarizes stored listing prices and historical patterns on the cards you track. It describes momentum, buyer activity, and price behavior — not where a card will trade tomorrow. Treat every read as context, not investment advice.
What does the read score mean?
The read score (market interest) measures how worth watching a card is right now on stored prices — recent moves, activity magnitude, and tape freshness. It is separate from pattern clarity, raw activity, and risk tags. None of these collapse into a single buy or sell verdict.
What data sources does Pokélytics use?
Pokélytics uses stored listing prints (JustTCG) for score horizons and modeled reads, plus SilphCo verified market data where integrated — sold counts, sales volume, verified sales history, and grading/slab context. Read scores and alert triggers are built from the stored price series; buyer-activity and verified-market panels come from SilphCo.
How does Pokélytics handle cards with thin price history?
When history is sparse or confidence is low, Pokélytics says so explicitly. Thin listings are not turned into fake volume, and the default scanner applies floors so noisy cards do not crowd the top of the list.