A liquidation map and cumulative volume delta answer different questions. A liquidation map estimates where leveraged positions may become vulnerable under a model. CVD summarizes the balance of buyer-initiated and seller-initiated volume in a selected market and period.
Reading them together can add context, but one indicator does not validate the other and neither predicts the next price move. The useful task is to compare what each dataset appears to show, identify alternative explanations, and decide what additional evidence would be needed.
Educational-use notice: CVD is a descriptive calculation and liquidation maps are model-based estimates. Results vary with venue coverage, trade classification, aggregation, contract rules, and the assumptions used by the map. They are not exchange liquidation quotes, probabilities, or financial advice.
This guide covers:
- what CVD measures and leaves out;
- how it differs from a liquidation map;
- conditional interpretations of absorption, divergence, and continuation;
- why spot, perpetual, and exchange-level CVD can disagree;
- a repeatable way to document observations without treating them as trade instructions.
What Is CVD?
Cumulative volume delta is a running total of volume delta. For each interval, a data provider classifies executed trades as buyer-initiated or seller-initiated, subtracts the latter from the former, and adds the result to the prior total.
For a broader introduction to volume, see the volume analysis glossary.
In simplified form:
CVD change = buyer-initiated volume − seller-initiated volume
A rising line means the selected feed classified more volume as buyer-initiated over that window. A falling line means it classified more volume as seller-initiated. This does not reveal every participant's intent, profitability, identity, or future action.
What CVD can describe
- the balance of aggressive buying and selling in the selected feed;
- whether price and aggressive flow moved together or apart;
- changes in order-flow intensity;
- differences between venues or contract types when separate feeds are compared.
What CVD cannot establish on its own
- whether a move will continue or reverse;
- whether passive liquidity came from a particular class of trader;
- the location of all resting orders;
- the intent behind a trade;
- a market-wide total when only a subset of venues is included.
Trade-side classification also matters. Providers may use different rules to infer the aggressor, and missing or delayed records can change the shape of the series.
What a Liquidation Map Adds
A liquidation map estimates concentrations of potential liquidation levels. Its output depends on a proprietary model, the declared market and venue coverage, the selected view, and the observation time. A bright area is therefore an estimated concentration, not a guaranteed target, support level, or reversal point. The Kingfisher explains how to interpret LiqMap outputs without publishing the private engine behind them.
The two tools are complementary in a limited, testable sense:
| Tool | Question it can help investigate | What it does not prove |
|---|---|---|
| Liquidation map | Where does the model estimate concentrated liquidation exposure? | That price will visit, reject, or pass through a level |
| CVD | How has aggressive executed flow evolved in the selected feed? | Who is trading or what price will do next |
| Price and volume | How did the market actually respond? | That the same response will repeat |
The word “confirm” should be understood as cross-checking observations, not certifying an outcome.
Three Conditional Patterns to Study
These are analytical hypotheses. Each has plausible alternatives, so none should be converted directly into an entry or exit rule.
1. Apparent absorption near an estimated cluster
An analyst may label a period as apparent absorption when aggressive flow increases in one direction while price makes comparatively little progress.
Questions to record:
- Is the CVD change broad or caused by one brief burst?
- Does the price response persist across more than one interval?
- Is the behavior visible on one venue or several?
- Did liquidity, volatility, or a news event change at the same time?
- Is the nearby map area still present after the product view refreshes?
Possible explanations include passive limit orders, hidden liquidity, execution across other venues, position transfers, or simple measurement noise. Participant identity or sophistication cannot be inferred from CVD alone.
2. Price–CVD divergence
A divergence occurs when price and CVD move differently over a chosen comparison window. For example, price may print a lower low while the selected CVD series does not.
That mismatch can prompt investigation, but it is not automatically exhaustion or reversal. The result can change when the start time, venue, aggregation, or contract is changed. Spot buying on one exchange can also coexist with perpetual selling elsewhere.
A careful review compares:
- identical start and end times;
- the same instrument and contract definition;
- spot and perpetual feeds separately;
- price change, total volume, open interest, and volatility;
- behavior before and after the observation, without selecting only favorable examples.
3. Price and CVD moving together
When price and CVD move in the same direction, the selected feed shows aggressive flow aligned with the observed move. This can describe current participation, but it does not establish that momentum will persist or that an estimated liquidation cluster will be reached.
Useful follow-up questions include whether participation is broadening, whether the move is concentrated on one venue, and whether price response weakens while aggressive volume continues.
Historical Chart-Reading Example
The archived chart below pairs a CVD divergence annotation with an estimated liquidation area near 18,850 during the 2022 market.
Archived illustration: CVD divergence and an estimated liquidation area.
In hindsight, an analyst could describe the sequence as follows:
- seller-initiated volume increased in the selected CVD feed;
- price stopped extending lower over the observed interval;
- price later moved away from the area.
The chart is useful for learning how the two displays can be compared. It does not show that CVD caused the later move, that the map forecast it, or that the same visual pattern has a stable success rate. A proper evaluation would require a defined sample, fixed rules, complete timestamps, costs, and out-of-sample testing.
Spot and Perpetual CVD Are Different Views
Spot and perpetual markets represent different activity. Perpetual contracts make leveraged long and short exposure accessible, while spot trades reflect direct asset exchange on the covered venue. Neither feed is universally “better.”
Differences can arise from:
- participant mix;
- funding and basis conditions;
- exchange-specific liquidity;
- hedging between spot, futures, options, and perpetuals;
- contract denomination;
- time-zone and aggregation choices.
Comparing both can reveal disagreement, but the disagreement remains descriptive. It does not identify which side is informed.
Multi-Exchange Analysis and Coverage
An aggregated CVD series can reduce dependence on a single venue, but aggregation introduces its own choices. A provider must decide which exchanges, pairs, currencies, and contract sizes to include and how to normalize them.
Before interpreting an aggregate, check:
- included venues and instruments;
- whether spot and derivatives are separated;
- how inverse and linear contracts are normalized;
- the handling of outages and missing trades;
- update time and historical revisions;
- whether one high-volume venue dominates the result.
Agreement across venues describes broader alignment in the included sample. Disagreement can indicate fragmented activity or simply different market structures.
A Non-Prescriptive Review Workflow
The following workflow is designed for research notes, replay, or post-event review.
1. Define the observation first
Write down the asset, venue, contract, interval, map version, and CVD source before looking at the subsequent price path. This reduces hindsight selection.
2. Describe, do not predict
Use neutral statements such as:
- “seller-initiated volume increased while price remained within the prior range”;
- “the model displayed an estimated cluster above the observed price”;
- “spot and perpetual CVD disagreed during this interval.”
Avoid converting those descriptions into a required reversal or an inevitable cluster visit.
3. List competing explanations
For each observation, record at least one alternative. A divergence may reflect cross-venue hedging. A stalled price may reflect passive orders, an index difference, or a short observation window.
4. Define what would change the interpretation
Examples include a different venue showing the opposite flow, the estimated cluster disappearing after a refresh, or price responding differently on a higher aggregation interval.
5. Review a complete sample
Keep both favorable and unfavorable cases. Record costs and latency if evaluating an execution hypothesis. A few memorable screenshots cannot establish reliability.
Common Interpretation Errors
| Error | Why it matters | Better question |
|---|---|---|
| Treating every divergence as reversal evidence | Divergence depends on window and venue | Does it persist under consistent definitions? |
| Calling stalled price “institutional absorption” | CVD does not identify participants | What other mechanisms could produce the same pattern? |
| Treating a cluster as a price target | A map is an estimate, not an order book | What does the product say this output represents? |
| Mixing spot and perpetual data | The feeds represent different activity | Are instruments and units comparable? |
| Ignoring coverage | One venue can dominate an aggregate | Which markets are included and how are they weighted? |
| Selecting only successful screenshots | Hindsight inflates confidence | What happened across all predefined observations? |
Combining CVD With Other Context
Open interest, funding, volatility, and options-derived measures can add context, but adding indicators does not automatically increase accuracy. Correlated inputs may repeat the same information, and every additional rule raises the risk of fitting a story to past data.
For related concepts, see:
FAQ
Which CVD timeframe is best?
There is no universally best interval. A useful choice depends on the question being studied, and comparisons need consistent definitions. Short intervals expose more microstructure noise; longer intervals can hide brief changes.
Can CVD be used without a liquidation map?
Yes, because CVD is an independent description of classified trade flow. A liquidation map adds a separate model-based view, but it does not make CVD predictive.
What is the difference between divergence and absorption?
Divergence compares the direction or extent of price and CVD over a window. Apparent absorption describes strong aggressive flow with limited price progress. Both are labels for observations, not proof of participant identity or future direction.
Why do exchanges show different CVD?
Each venue has its own participants, products, liquidity, and trade stream. Data providers may also classify and normalize trades differently.
How can an absorption interpretation be checked?
Use consistent timestamps and instruments, compare other venues, inspect total volume and price response, and retain cases where the interpretation did not precede the expected outcome.
Further Reading
The Kingfisher application and its available datasets may change over time. Consult the current product interface for coverage and access details rather than relying on an article as a product specification.
