# Data dictionary (WP46)

## Daily panel (`<cache>/panels/<symbol>.parquet`, one row per session; built by `scripts/build_panels.py`)
| column | unit | definition |
|---|---|---|
| session | date | session identifier (ET date for equities/futures/FX, UTC date for crypto; see D-02) |
| close, open | price | session close (last 1-minute bar close in the window) and open; equities: vendor split+dividend-adjusted daily bars |
| r | percent | 100·log(close_t / close_{t−1}) — the modelled daily return |
| r_on | percent | 100·log(open_t / close_{t−1}) (overnight; 0 for FX/crypto, break return for futures) |
| r_oc | percent | 100·log(close_t / open_t) |
| rv_1m | percent² | Σ squared 1-minute log returns within the window |
| rv_5m | percent² | Σ squared 5-minute calendar returns (grid anchored at the window start, previous tick) |
| rv_5m_ss | percent² | average of the five 5-minute RV estimates with 1-minute grid offsets (primary intraday measure) |
| bv_5m | percent² | (π/2)·Σ|r_j||r_{j−1}| on the 5-minute grid |
| rsv_pos_5m, rsv_neg_5m | percent² | realized semivariances on the 5-minute grid |
| medrv_5m | percent² | median realized variance on the 5-minute grid |
| n_bars, n_5m | count | 1-minute bars used; 5-minute returns |
| expected_bars | count | full-session bar count (391/211 equities; modal count otherwise) |
| valid_rm | bool | realized measures usable (D-05, D-06) |

## Forecast table (`results/forecasts/<run>/<symbol>/<spec_id>.parquet`, one row per origin)
origin (date of the information set), fit_id, status ∈ {ok, stale_fit, failed, forecast_failed, no_fit, missing_input},
method ∈ {analytic, simulation, direct}, h1/h5/h22 (percent², cumulative variance forecasts of the 1/5/22-day returns),
sigma2_t, mu (one-step mean, percent), dp1/dp2 (distribution parameters), time_fc_s, admissible, plus the keys symbol,
model_id, dist, spec_id, tier, window, refit_every, expanding.

## Fits table (`<spec_id>.fits.parquet`, one row per refit attempt)
fit_id, origin, status ∈ {ok, inadmissible, failed}, error, loglik, n_iter, n_restarts, time_s, n_obs, admissible, params, param_names.

## Loss table (`results/metrics/losses/<symbol>.parquet`)
origin, spec_id, model_id, dist, tier, h, proxy, loss, value, status, forecast, target.

## Keys
`spec_id = <model_id>__<dist>__w<window>_r<refit>` (or `exp` for expanding). Run manifests (`_runs.jsonl`) carry git SHA, host,
package versions, wall/CPU time and peak memory per job; dataset versions and SHA-256 of every exported file are in
`results/manifests/ingest_manifest.json`.