Condition · recorded history

Large USD OI change with liquidations

Absolute USD open-interest change over roughly five minutes has a rolling z-score/normal-CDF score of at least 0.90, alongside at least $1M reported liquidations in the trailing hour. Falling OI also qualifies; price affects USD OI. No signed contract-growth requirement.

What is open interest?

Recorded window: 5 Oct 2026 to 3 Nov 2026 (UTC). Counted 4 Nov 2026 at 08:23 UTC.

Read it in context

On one market
Each market page compares that market with its own minutes, not with every market at once.
Compare markets
Test it, then plan it
Narrow it to your markets and dates in the Workbench. A tested result leads to Plan a trade, which measures what trading it would have paid after costs.
Edit this test in Workbench

By market

Results vary by market. Compare a market before drawing a conclusion.

Most matches:BTCUSDT ETHUSDT SOLUSDT ZECUSDT

How results varied by market

Range across 4 markets with at least 30 measured: reached +5% within 24 hours in 0 of 143 on BTCUSDT (0%) to 13 of 36 on ZECUSDT (36%).

Every market pooled, without context

What happened in the next 24 hours?

Both moves were less common than on minutes without it.

Other minutes are the same markets and dates without this condition. Each percentage is the share of measured minutes that reached that move from the starting price.

Price touched +5% or higher

When this held: 13.6% · 67 of 493 minutes
Other minutes: 24.4% · 5,062,810 of 20,743,924 minutes

Price touched −10% or lower

When this held: 3.2% · 16 of 493 minutes
Other minutes: 8.6% · 1,789,735 of 20,743,924 minutes

Price can touch both levels within the same 24 hours.

Every matching minute counts, so one long run adds many near-identical minutes. Busy markets weigh most.

Every condition is measured over the same 24 hours, whatever the length of its own reading. Other windows and move sizes shows shorter and longer windows.

This counts how often price touched a level, not what trading it would have paid.

Sample and counting

Rates use the 493 of 524 matches, pooled across 523 markets, with a full 24 hours after them. The 31 newer ones are left out, not counted as zero.

When it happened

About 17 a day, on 30 of 31 days: spread through the record rather than one event.

Other windows and move sizes

Matches that reached each move within each window. Counts first; the measured row is what each rate is over.
Move	1 hour	4 hours	24 hours	72 hours	7 days
Measured	500	500	493	468	383
Too recent	24	24	31	56	141
+0.1% or more	411 (82.2%)	448 (89.6%)	470 (95.3%)	450 (96.2%)	376 (98.2%)
+0.2% or more	340 (68.0%)	402 (80.4%)	442 (89.7%)	426 (91.0%)	365 (95.3%)
+0.5% or more	208 (41.6%)	299 (59.8%)	383 (77.7%)	390 (83.3%)	338 (88.3%)
+1% or more	116 (23.2%)	196 (39.2%)	303 (61.5%)	337 (72.0%)	303 (79.1%)
+2% or more	53 (10.6%)	106 (21.2%)	193 (39.1%)	243 (51.9%)	231 (60.3%)
+5% or more	11 (2.2%)	29 (5.8%)	67 (13.6%)	102 (21.8%)	104 (27.2%)
+10% or more	2 (0.4%)	5 (1.0%)	19 (3.9%)	40 (8.5%)	48 (12.5%)
+15% or more	2 (0.4%)	2 (0.4%)	6 (1.2%)	19 (4.1%)	32 (8.4%)
+20% or more	2 (0.4%)	2 (0.4%)	3 (0.6%)	11 (2.4%)	21 (5.5%)
+30% or more	2 (0.4%)	2 (0.4%)	3 (0.6%)	6 (1.3%)	7 (1.8%)
+50% or more	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (0.2%)	1 (0.3%)
−0.1% or lower	439 (87.8%)	467 (93.4%)	485 (98.4%)	464 (99.1%)	381 (99.5%)
−0.2% or lower	388 (77.6%)	440 (88.0%)	476 (96.6%)	461 (98.5%)	381 (99.5%)
−0.5% or lower	253 (50.6%)	349 (69.8%)	448 (90.9%)	449 (95.9%)	379 (99.0%)
−1% or lower	144 (28.8%)	231 (46.2%)	379 (76.9%)	408 (87.2%)	370 (96.6%)
−2% or lower	65 (13.0%)	117 (23.4%)	268 (54.4%)	341 (72.9%)	356 (93.0%)
−5% or lower	6 (1.2%)	32 (6.4%)	91 (18.5%)	188 (40.2%)	240 (62.7%)
−10% or lower	4 (0.8%)	7 (1.4%)	16 (3.2%)	45 (9.6%)	59 (15.4%)
−15% or lower	3 (0.6%)	4 (0.8%)	8 (1.6%)	19 (4.1%)	21 (5.5%)
−20% or lower	2 (0.4%)	3 (0.6%)	6 (1.2%)	7 (1.5%)	9 (2.3%)
−30% or lower	2 (0.4%)	2 (0.4%)	5 (1.0%)	5 (1.1%)	6 (1.6%)
−50% or lower	0 (0.0%)	0 (0.0%)	1 (0.2%)	1 (0.2%)	3 (0.8%)

Each window counts only matches old enough for it, so a longer window can show fewer.

Distributions and daily counts

Upside reached · 24hThe most price moved up within 24 hours of each match: a high-water mark, never a close. 493 with a complete 24 hours.

Drawdown · 24hThe worst price moved against within 24 hours of each match: the deepest point, never the close. 493 with a complete 24 hours.

Price 30 minutes laterForward 30-minute returns across 498 matches: close to an even split, which is what a coin flip looks like. 26 too recent.

How to read a distribution

Matches per UTC day

5 Oct11 matches
6 Oct8 matches
7 Oct28 matches
8 Oct63 matches
9 Oct8 matches
10 Oct2 matches
11 Oct10 matches
12 Oct31 matches
13 Oct25 matches
14 Oct17 matches
15 Oct33 matches
16 Oct18 matches
17 Oct6 matches
18 Oct18 matches
19 Oct13 matches
20 Oct9 matches
21 Oct14 matches
22 Oct20 matches
23 Oct18 matches
24 Oct21 matches
25 Oct9 matches
26 Oct10 matches
27 Oct6 matches
28 Oct13 matches
29 Oct44 matches
30 Oct35 matches
31 Oct1 match
1 Nov6 matches
2 Nov17 matches
3 Nov10 matches
4 Nov0 matches

How this is counted

Each rate counts only matches with the full window after them.
A minute where a reading was missing is left out and reported, not read as zero.
The condition picks the setup, never the outcome: a definition that filters on what followed is rejected (OUTCOME_IN_PREDICATE). Every match is measured, winners and losers alike.
"Other minutes" is every other eligible minute in the same markets and window. It is a reference, not a control group.
"Reached +5%" uses the highest price inside the window (maximum favourable excursion, MFE); "fell to −10%" uses the lowest (maximum adverse excursion, MAE). Neither is the close.
Checked 22,397,636 eligible market-minutes (eligible_symbol_buckets) on 523 markets, 2026-10-05 00:00 to 2026-11-04 00:00 UTC.
These counts pool every market, so markets that match often weigh more. Each market page compares that market with its own reference.
Live status is not connected yet, so this page is recorded history only.

What are MFE and MAE?

Reproducibility key (q#7a6d8db5)

The key that re-runs this exact count. Same key, same bytes; when the record grows, the dataset revision changes.

grammar: research_query.v2
normalization: archive_normalization.v1
feature library: feature_defs.v1
dataset revision: efa4fa580f35de27f8dca2c77e7722f8cf97bf0e505571eb4186853f08de9bd7
query hash: 7a6d8db5614fd40a161585b626c410ce6dde3ca2af3445eec299eab5416b1051
scanned window: 2026-10-05T00:00:00.000Z to 2026-11-04T00:00:00.000Z (UTC)

Re-run over the API

The exact search definition behind this count. Send it to the key-authed research API (or the MCP research tool) and compare the bytes; a cached rerun is free.

{
  "schema_version": "research_query.v2",
  "normalization_version": "archive_normalization.v1",
  "feature_version": "feature_defs.v1",
  "target": "record_occurrences",
  "where": {
    "all": [
      [
        "feature.oi_velocity_pctrank",
        "gte",
        0.9
      ],
      [
        "feature.liq_notional_usd_1h",
        "gte",
        1000000
      ],
      [
        "times.anchor_time",
        "between",
        [
          "2026-10-05T00:00:00.000Z",
          "2026-11-04T00:00:00.000Z"
        ]
      ]
    ]
  },
  "sort": [
    "times.anchor_time",
    "desc"
  ],
  "page": {
    "limit": 50,
    "cursor": null
  }
}

curl -s https://app.edgedepth.com/api/v1/research/query \
  -H 'authorization: Bearer $EDGEDEPTH_API_KEY' \
  -H 'content-type: application/json' \
  -d '{"schema_version":"research_query.v2","normalization_version":"archive_normalization.v1","feature_version":"feature_defs.v1","target":"record_occurrences","where":{"all":[["feature.oi_velocity_pctrank","gte",0.9],["feature.liq_notional_usd_1h","gte",1000000],["times.anchor_time","between",["2026-10-05T00:00:00.000Z","2026-11-04T00:00:00.000Z"]]]},"sort":["times.anchor_time","desc"],"page":{"limit":50,"cursor":null}}'