How to Read a Crypto Liquidation Heatmap

Updated 05 September 2026

First identify the map. “Liquidation heatmap” is used for three different datasets: resting order-book liquidity, modelled future liquidation levels, and forced liquidations that already happened. They answer different questions and cannot be read interchangeably. This guide is educational, not a trading signal.

If you want the event tape rather than the explanation, open Yieldo’s live crypto liquidations. It groups observed Binance and Bybit forced-close events by source-reported price and time: average execution price for Binance, bankruptcy price for Bybit. It does not estimate where open positions will liquidate next.

The three maps people call a liquidation heatmap

1. Order-book liquidity map

This visualises limit orders currently advertised in an exchange order book. A bright band means more displayed bids or asks at that price, not more leveraged positions waiting to be liquidated. Orders can be moved or cancelled before price reaches them. Use this map to study displayed market depth and execution conditions, not hidden leverage.

2. Estimated liquidation-level map

This is a model of potential future liquidation zones. The provider estimates where positions might be forced closed from assumptions about entry prices, leverage, margin mode and exchange rules. Those private account inputs are not fully visible from public market data, so intensity is an estimate, not a queue of guaranteed orders. Compare the provider’s methodology and scale before comparing colours across products.

3. Observed liquidation-event map

This groups forced-close events that an exchange has already published. It is historical even when it updates live. A cluster can confirm that forced deleveraging occurred near a price, but it cannot tell you which unliquidated positions remain or where price goes next. Yieldo’s live page is this third type.

Map	What a bright area means	Time direction	Main limitation
Order book	More displayed limit-order size	Current	Orders can change or disappear
Estimated levels	More modelled liquidation exposure	Potential future	Depends on hidden inputs and assumptions
Observed events	More published forced-close notional	Past, streamed live	Public feeds may be sampled or use different price fields

How to read an estimated liquidation-level map

1. Find current price first. Treat it as the reference line separating levels above and below the market, not as evidence that price will travel toward the brightest area.
2. Use that provider’s legend and scale. On many estimated maps, zones below current price are interpreted as modelled long-liquidation exposure and zones above it as modelled short-liquidation exposure. This is a conditional reading of the provider’s model, not a universal colour rule; verify whether intensity represents relative weight, estimated notional or another score.
3. Fix the market definition. Check the exchange or exchange group, exact contract, asset and lookback before comparing two zones. A BTC perpetual on one venue and an aggregate across venues are not the same cohort.
4. Read every zone as an estimate. The model cannot fully observe private entry prices, leverage, added collateral or cross-margin relationships. Exchange risk rules can also change the eventual liquidation threshold.

A bright estimated zone therefore means “the provider’s assumptions place more potential liquidation exposure here.” It does not mean every position will trigger, that a cascade is guaranteed, or that price must reverse after reaching the zone.

How to read an observed liquidation map in five steps

1. Confirm the label and source. Look for words such as observed, executed or historical. Check which exchanges, contracts and retention window are covered.
2. Select one asset and window. Start with BTC, ETH or SOL and compare the same asset over 1h, 4h and 24h. A cell is meaningful only relative to the same chart’s scale.
3. Read the position side. A long liquidation means a leveraged long was force-closed through selling; a short liquidation means a short was force-closed through buying. UI colours are conventions, so use the legend rather than memory.
4. Read price and notional together. A large USD cell near one price shows concentrated realised deleveraging. Several adjacent cells indicate a broader cascade; neither pattern predicts a reversal.
5. Add independent context. Compare the event with price structure, open interest and funding. See live funding for BTC, ETH and SOL. Never let the heatmap replace position sizing and a risk limit.

A clearly synthetic reading example

Illustration only — these are invented teaching values, not market data: suppose an observed-events chart contains $4 million of long liquidations from $99,800 to $100,100 and $0.7 million of short liquidations in the same 15-minute interval. The valid statement is: “the covered venues reported substantially more forced selling than forced buying in that bin.” It does not prove that $100,000 is support, that all longs have been flushed, or that price must rebound. Check the live page for actual values.

Why Binance and Bybit cells are not directly identical

Venue feeds have different coverage and field semantics. Binance documents its all-market liquidation snapshot as the latest liquidation order per symbol within each 1,000 ms interval. Its payload distinguishes order price p, average execution price ap and accumulated filled quantity z. Yieldo retains only terminal FILLED or EXPIRED snapshots with positive ap and z, then uses ap × z as an observed-notional estimate. This deliberately undercounts rather than adding a partial snapshot and its later cumulative total twice; it is not an exact execution tape. Binance stream documentation.

Bybit’s all-liquidation topic publishes all liquidations with a 500 ms push frequency. Its p field is explicitly the bankruptcy price and v is executed size, so that point should not be described as an ordinary trade execution price. Bybit stream documentation.

Common reading mistakes

• Treating a past-event cluster as a forecast of the next liquidation level.
• Comparing colour intensity between providers without checking whether scales are absolute, relative or modelled.
• Calling a Binance snapshot exhaustive when its documented 1,000 ms sampling can omit events during a busy interval.
• Calling Bybit’s bankruptcy-price field an execution price.
• Adding volumes from different windows, assets or contract types as if they described one cohort.
• Opening a trade from a bright cell alone. A cluster documents stress; it does not choose direction or timing.

Practical next step

Open the live observed-liquidations page, choose BTC, ETH or SOL, and state what the chart actually proves in one sentence before forming an interpretation. If your sentence contains “will”, return to the source label: you may be reading historical evidence as a prediction.