Data Model

The ByteTree API exposes both adjusted and raw price values. This page explains the three internal layers that produce them — and exactly what the adjustment contains — so consumers can decide which surface to read.

Bottom line. Default-trust the adjusted values for any chart, dashboard, or computed metric. Raw is for forensic or audit users who need to reconstruct the historical market-printed price.

Three layers

1. RawPriceData — forensic archive

What the upstream data provider returned at original ingest. Append-only and immutable. Not exposed via the public API. Used internally to reconstruct exactly what we received and when.

2. PriceData raw fields — daily landing pad

The daily-fetch landing pad for the provider’s current view of historical OHLCV. Mutable: the provider retroactively rewrites the raw close (and the legacy adjClose) for split-affected symbols whenever a split is processed, so a raw historical close for a stock can change after the fact.

Exposed publicly through:

• GET /api/prices/:symbol?raw=true — full raw OHLCV
• The *Raw fields on metric responses: closeRawUsd, caprRaw, marketCapRawUsd

3. Own-computed adjusted OHLCV — default display

Seeded from a fully split- and dividend-adjusted price history, then maintained forward against our internal splits ledger via a pure function. Idempotent — the same inputs always produce the same outputs — and immune to upstream rewrites since the 2026-05-05 cutover.

This is the default for any chart, score, moving average, drawdown, or return computation. Exposed publicly through:

• GET /api/prices/:symbol — default OHLCV is adjusted
• GET /api/prices — default close is adjusted
• GET /api/metrics/* — closeUsd, capr, marketCapUsd are adjusted; every time-series metric is computed on adjusted close

What the adjustment contains

• Splits — fully adjusted, past and future. New splits rebase the series via the internal splits ledger as they occur.
• Dividends — fully adjusted as of the series’ seed point. Distributions after the seed are not retro-applied; they flow through as ordinary price movement, exactly as on an exchange print. A parity audit against the upstream fully-adjusted series watches for material drift.

In practice: long-history values for dividend-paying assets sit below the market-printed price of the day (cumulative distributions are folded back into history), while the most recent bars match the print. If you reconcile against a split-only-adjusted vendor series, expect historical values on dividend payers to differ by roughly the compounded yield.

Which view do you want?

Why two close fields exist

Internal compute pipelines (ByteTrend scores, drawdowns, MAs, returns, deviations) operate on adjusted close so historical signal isn’t corrupted by split discontinuities or dividend-distribution steps. Quoted-state fields (closeUsd, market cap) used to live on raw close — that meant a stock’s reported “today’s close” disagreed with the close used to compute its MA. The 2026-05-05 cutover unified both onto the adjusted basis by default; the raw view is preserved alongside as *Raw fields and ?raw=true for the small audience that needs it.

FX

FX pairs aren’t subject to splits or dividends, so the adjusted and raw views are identical. The /api/fx/* namespace and the polymorphic FX rows on /api/assets and /api/metrics therefore expose only the default (adjusted-equivalent) values.

Related

• GET /api/prices/:symbol — ?raw=true reference
• Latest Metrics — Valuation — closeUsd / capr / marketCapUsd and their *Raw counterparts