Research is a work in progress. Estimated release: 1 December 2026. Our goal is to make it a search engine for trading setups in recorded crypto markets.

Browse related readings

Research Readings Liquidations

Liquidation burst size

Also shown as Liquidation intensity

Compares the observed liquidation USD rate in the last 15 seconds with this symbol's adaptive baseline rate, then scales and caps the result.

Interactive illustration · not market data

Compare a burst with its baseline

Rate as a multiple of baseline 3×

Reading0.3min(3 / 10, 1) = 0.3

Baseline held fixed for this illustration. The real baseline adapts; zero is available only after warmup and does not prove no liquidations occurred.

Live monitoring: Binance Futures only; depends on the full setup
Observed: read at the minute itself
Unit: Ratio
Valid range: 0 to 1

How to read it, with examples

How to read it

The reading is min((rate / baseline) / 10, 1). A value of 0.1 is the baseline rate, 0.3 is three times it, and 1 is at least ten times it. The baseline updates on liquidation events and is floored at $100 per second. Five observed events are required before a value exists. After warmup, an empty 15-second window reads zero.

Constructed example · not historical data

Baseline rate is 0.1 on this scale

Rate = 1 × its baseline0.1
Rate = 3 × its baseline0.3
Rate = 12 × its baseline1.0 (capped)

The full bar is 10 times the baseline. Divide the rate-to-baseline ratio by 10, then cap at 1. These are constructed comparisons, not successive observations: the real baseline adapts.

Locate bursts relative to the symbol's own changing baseline, then compare observed hourly liquidation notional and event clustering for scale and timing context.

What it does not prove

This is not a percentile, a probability of a cascade or a fixed USD amount. Zero describes this observed window, not proof that no liquidations occurred.

More limits

The baseline changes, the ceiling hides larger ratios, and a minute sample can miss a burst between samples.

Use in Research

What this cannot see

Binance throttles liquidation publications, so these readings describe the observed feed and miss some liquidation records.

Registry definition

Liquidation USD rate over the trailing 15 seconds divided by an adaptive per-symbol baseline, that ratio divided by 10, then censored at 1.0. Not a rank or percentile: 1.0 means the rate reached ten times its own adaptive baseline, and that baseline moves, so the same value at two dates is not the same USD rate.

Field ID

feature.liq_intensity_norm

Valid query operators

at least, at most, between

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

Editable query preset

Starting point only. Review and edit before running.

{
  "schema_version": "research_query.v2",
  "normalization_version": "archive_normalization.v1",
  "feature_version": "feature_defs.v1",
  "target": "record_occurrences",
  "where": {
    "all": [
      [
        "feature.liq_intensity_norm",
        "gte",
        0.3
      ],
      [
        "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.