Early access is open. Free sample datasets are live now. Download a sample

Learn

Reading Crypto Liquidations and Open Interest

7 min read · Updated 19 November 2026 · By the marketdata.co data team

A sharp drop in a perpetual future can be fresh selling or a wave of forced exits. Price alone can't tell you which. Open interest and liquidation data can, and once you know how to read them together you stop chasing moves that were only leverage being flushed out.

Details

What the two numbers mean

Open interest is the number of perpetual contracts currently open. Every contract has a long and a short, so it counts each pair once. When open interest rises, new positions are being opened. A fall means positions are being closed, by choice or by force.

Liquidations are positions the exchange closes because margin ran out. A long liquidation is a forced market sell; a short liquidation is a forced market buy. Exchanges publish liquidation orders as a stream, and crypto liquidation data is usually stored as one row per forced order with side, price and size.

Read them together with price:

Price Open interest Usual reading
Up Up New longs entering
Up Down Shorts closing or being squeezed
Down Up New shorts entering
Down Down Longs closing or being liquidated

The last row is where liquidation data earns its keep. If open interest falls fast and the liquidation stream shows a burst of long liquidations at the same minute, the move was forced. Forced moves often stall once the leverage is gone.

A real example from 1 November 2026

Our free sample holds 288 five-minute snapshots of BTC-USDT perpetual open interest for one day. Here is the largest five-minute drop of the day:

symbol timestamp open_interest open_interest_usd top_trader_long_short_ratio global_long_short_ratio taker_buy_sell_ratio
BTC-USDT-PERP 2026-11-01T19:00:00Z 97258.749 7573499881.13 2.36767576 2.26166037 0.871586
BTC-USDT-PERP 2026-11-01T19:05:00Z 96443.965 7491873289.56 2.37135652 2.26058096 0.281541
BTC-USDT-PERP 2026-11-01T19:10:00Z 96344.192 7484806856.93 2.36952426 2.25711304 1.119464

Source: Binance

Open interest fell by 814.784 BTC in five minutes, about $82 million of notional. In the same window taker_buy_sell_ratio dropped to 0.28, meaning aggressive sell volume was more than three times aggressive buy volume. Falling open interest plus one-sided taker selling is the footprint of longs exiting in a hurry, some by choice and some by liquidation. The next bar's ratio jumped back above 1, which tells you the selling burst was over.

For the full day, open interest ranged from 95,232.522 to 98,032.215 BTC. The 19:00 hour alone lost 1,181 BTC, the largest hourly fall in the file.

Details

Flag forced-selling bars in pandas

"""Example: find bars where OI drops sharply and takers are selling."""
import pandas as pd

oi = pd.read_csv("btc-perp-open-interest.csv", parse_dates=["timestamp"])
oi["oi_change"] = oi["open_interest"].diff()
oi["oi_change_pct"] = 100 * oi["oi_change"] / oi["open_interest"].shift()

z = (oi["oi_change"] - oi["oi_change"].mean()) / oi["oi_change"].std()
flush = oi[(z < -3) & (oi["taker_buy_sell_ratio"] < 0.5)]

print(flush[["timestamp", "open_interest", "oi_change", "oi_change_pct",
             "taker_buy_sell_ratio"]])

On the sample this returns two bars, 12:15 and 19:05. A third drop, at 23:35, is also beyond three sigma but is filtered out because its taker ratio was 2.55: buyers were the aggressors, which fits shorts closing rather than longs being flushed. With the liquidation stream joined on timestamp, you would also sum long liquidation notional per bar and check that it spikes on the same row. The three-sigma threshold is a starting point, not a rule; tune it on more than one day of data.

Watch out

Common mistakes

Treating all OI drops as liquidations. Traders close positions for many reasons, including taking profit. Without the liquidation stream you are guessing at the split between voluntary and forced exits.

Reading the long/short ratio as a sentiment poll. global_long_short_ratio counts accounts, not money. On 1 November it sat around 2.3, yet that says nothing about how large the short positions were.

Mixing coin and USD open interest. open_interest_usd rises when price rises even if no new contracts open. Use open_interest in coins to measure positioning and the USD figure only for size comparisons across assets.

Using one exchange as the whole market. Our sample is one venue. Liquidations on one exchange can spill into others, so aggregate venues before calling a move market-wide.

Details

Get the data

Free BTC perpetual open interest sample. 288 five-minute rows for 1 November 2026 with open interest in coins and USD, top trader and global long/short ratios, and taker buy/sell ratio.

FAQ

Frequently asked questions

What is crypto liquidation data?

It is a record of positions an exchange closed because the trader's margin fell below the maintenance level. Each row usually has a timestamp, side, price and size. Long liquidations are forced sells and short liquidations are forced buys.

Does falling open interest mean liquidations?

Not on its own. Falling open interest means positions are closing, which includes normal profit taking. You need the liquidation stream, or at least a burst of one-sided taker volume, to say the closing was forced.

Is rising open interest bullish?

Rising open interest with rising price shows new longs entering, which supports the move. With falling price, the same rise shows new shorts. Either way, a large build-up of leverage raises the risk of a sharp liquidation move later.

How often is open interest updated?

Exchanges update it continuously, but most historical data is stored as snapshots. Our sample uses 5-minute snapshots, and 1-minute history is planned for the Pro plan.

Keep reading