order-book imbalance rank

Historical percentile rank of absolute book imbalance.

Observed: read at the minute itself
Unit: percentile
Valid range: 0 to 1

In plain English

Describes how unusually large the imbalance is in either direction relative to this symbol's recent observations. Taking the absolute imbalance removes the bid-versus-ask sign.

How to read it

The percentile scale is a normal-CDF transform of a rolling z-score, not a counted share of historical minutes. A value of 0.5 corresponds to the rolling mean magnitude; 0.9 is about 1.28 rolling standard deviations above it. Low values mean smaller-than-usual magnitude. Statistics update at most once per 10 seconds; 100 updates are required. With degenerate variance, the last valid rank holds, or remains absent if none exists.

Why a researcher might use it

Find unusually uneven best quotes for the same symbol without imposing one raw imbalance threshold on every market. Read the signed imbalance alongside it.

What it does not prove

A value of 0.9 does not mean bids dominate, a 90% chance of a move, or that exactly 90% of recorded observations were lower. It is separate from the snapshot API's counted full-record percentile metadata.

Use in Research

Technical details

Field ID

feature.book_imbalance_pctrank

Valid query operators

at least, at most, between

Window operators: lowest, highest, average, latest or change in, over a trailing window.

Editable query preset

This preset is a starting point, not a recommendation. Edit it before running the search.

{
  "schema_version": "research_query.v2",
  "normalization_version": "archive_normalization.v1",
  "feature_version": "feature_defs.v1",
  "target": "record_occurrences",
  "where": {
    "all": [
      [
        "feature.book_imbalance_pctrank",
        "gte",
        0.9
      ],
      [
        "times.anchor_time",
        "gte",
        "2025-07-15T00:00:00.000Z"
      ]
    ]
  },
  "sort": [
    "times.anchor_time",
    "desc"
  ],
  "page": {
    "limit": 5,
    "cursor": null
  }
}

The search reports how often the record held this condition and what followed. It does not decide whether an idea works.