# CoinNudge sector trend grades 1.0

Frozen on 2026-09-26. Input and outcome window: 2026-06-26 to 2026-09-24 UTC. The source is the 107 included sector assignments in `sector-breadth-study-3.0`; 86 distinct symbols can occur in more than one editorial sector.

## Interpretation

A–E describe a **past price path**, not asset quality, token rights, current momentum or a buy recommendation. The rules in `grades.json` are applied in order. Returns are Binance USDT daily-open to daily-open changes; 7d and 30d are nested inside the 90d window. Maximum drawdown uses daily opens, not intraday lows. Median quote turnover is completed trading value, not order-book depth or guaranteed execution. This current-survivor, single-window sample is descriptive, not a probability estimate.

## Files and fields

- `source-observations.csv`: exact frozen input copied from sector release 3.0; its SHA-256 is in `grades.json`.
- `grades.csv`: all 107 assignments with sector, symbol, grade, labels, UTC dates, 7/30/90d decimal returns, BTC-relative 90d percentage points, daily-open max drawdown and 30d median USDT quote turnover. Decimal returns such as `0.25` mean +25%.
- `grades.json`: same rows grouped by sector, grade counts, exact rule text and data scope.
- `reproduce.py`: independent standard-library verifier for every grade, raw numeric value, row order, counts and manifest hash.
- `manifest.json`: SHA-256 of all other release files.

Run `python3 reproduce.py` from any location after downloading all files into one directory. Missing/new/off-venue candidates remain in the upstream sector candidate ledger and are not graded here.