> For the complete documentation index, see [llms.txt](https://oracledocs.magpie.gg/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://oracledocs.magpie.gg/the-oracle/how-it-works.md).

# How the price is made

A card's "price" is really a pile of signals of very different quality. The fusion engine's job is to weigh them honestly — and to tell you how sure it is, not just what it thinks.

## The hierarchy of evidence

Not all price signals are equal. The fusion engine weights them by nature:

1. **REALIZED** (weight 1.0) — actual settled sales: public on-chain sales of tokenized cards (indexed per-mint, all venues, full history back to 2023), and first-party marketplace deals. Each sale is stored with its settlement timestamp and converted at **that day's** FX/SOL rate — an old sale is never re-priced at today's spot.
2. **MARKET aggregates** (≈0.85) — same-language market aggregates used as anchors and cross-checks.
3. **ASKING** (≈0.65) — live sell-side listings (language-aware, EU): a floor estimate, never allowed to claim "high confidence" on its own.

## The fusion, honestly

* **Weighted median, not mean** — a single wash trade or fat-finger cannot move the value. Weights = source trust × recency decay (half-life 30 days).
* **The band is weighted too** (p25–p75 of the same weighted distribution), so stale low-trust sales can't inflate the spread the value already discounted.
* **Wash-trade guards**: self-trades (buyer = seller) are dropped at ingest; the median + band absorb single outliers.
* **Language-aware**: a Japanese card is priced by Japanese-market signals — English aggregates are down-weighted as a wrong-language proxy.
* **Graded is its own market**: slab values come from realized graded sales, broken down per (grading company, grade) — a PSA 10 and a CGC 9.5 are not the same asset. The headline gem value is the grade ≥ 9 band.

## Confidence, defined

| Tier     | Meaning                                                                                                    |
| -------- | ---------------------------------------------------------------------------------------------------------- |
| `high`   | multiple corroborating signals, or enough fresh realized sales (effective liquidity ≥ 4) with a tight band |
| `medium` | real signal, limited corroboration — the default for healthy-but-thin series                               |
| `low`    | thin, dispersed, or diverging signals — displayed, never hidden                                            |

Confidence is computed from **effective liquidity** (recency-weighted sales, so 5 stale sales count less than 5 fresh ones), band dispersion, and corroboration between independent sources. Asking depth alone can never reach `high`.

## Coverage model (tiers of depth)

* **Dense tier (\~5,000 series)**: owned/traded cards plus the most valuable cards of each game — hourly ticks + daily points (this is what feeds 1h/4h candles).
* **Catalog tier (\~170,000 cards)**: daily baseline, refreshed perpetually in rotation; values computed on demand with the full fusion.
* **Realized layer**: every mapped on-chain/marketplace sale lands as an intraday tick at its settlement time — the backfill *is* the chart history.

Everything above runs unattended: scheduled sweeps rotate persistent cursors over the venues and the catalog; new sets and late-added cards (promos, secret rares) are detected and ingested automatically.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://oracledocs.magpie.gg/the-oracle/how-it-works.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.