What is the Nakamoto Coefficient for Cardano?

The Nakamoto Coefficient represents the minimum number of independent operators or stake pools required to compromise consensus security or exceed a majority threshold (greater than 50% for chain reorganization in Ouroboros Praos, or greater than 33.3% for block liveness stall). Cardano Observer calculates this metric live on every snapshot across the complete active stake distribution.

Why are there two different Nakamoto numbers: Entities versus Pools?

Single pool IDs measure distinct registered on-chain pools, but multiple pools may be operated by a single entity. The Entity-level Minimum Attack Vector (MAV) clusters pools sharing common operator groups, owner stake keys, or reward addresses to reflect realistic collusion thresholds.

What is the difference between the 33.3% and 50% thresholds on Cardano?

Unlike BFT consensus networks that halt when 33.3% of voting power fails, Cardano uses probabilistic longest-chain consensus (Ouroboros Praos). Controlling 33.3% can temporarily stall block production, whereas rewriting historical blocks or double-spending strictly requires greater than 50% of the active stake.

How does multi-pool operator clustering work on Cardano?

Cardano Observer clusters pools by analyzing multi-pool operator groups, shared owner verification keys, and shared reward accounts on-chain, supplemented by curated operator registries to ensure transparency and prevent sybil obfuscation.

Consensus & Decentralization

Cardano Nakamoto Coefficient Live Tracker

The smallest set of operators whose combined stake crosses a consensus threshold. One count clusters pools by operator, the other counts registered pool IDs — the gap between the two is what multi-pool operation hides.

Nakamoto – Entities · 50 % majority set

-- awaiting snapshot

Loads with snapshot

Nakamoto – Pools

Loads with snapshot

1/3 Liveness

-- Entities

Loads with snapshot

Gini Index (0-1)

0 = even · 1 = one pool

Active Delegated Stake

-- ADA

Loading pool telemetry...

Consensus security thresholds

The bar is the active stake the smallest qualifying set holds; the tick is the line it has to cross.

33.3 % liveness (1/3) --

Reading the snapshot…

50.0 % majority (NC) --

Reading the snapshot…

66.7 % supermajority (2/3) --

Reading the snapshot…

66.7 % of stake --

90 % of stake --

Pool IDs per entity (50 % set) --

Cumulative stake curve

How fast the top of the distribution reaches each consensus line.

Lens

Operator entities (clustered) Single pool IDs 1/3 liveness line 50 % majority line Equal split across entities

Reading the snapshot…

Loading snapshot...

Power concentration tiers

Loading tiers…

Reading the snapshot…

How the entities are built

Loading clustering breakdown...

Snapshot: loading...

Reading the numbers

-- largest entity share --

-- counts at the 50 % line entities vs. pool IDs: the gap is what clustering hides

-- distance past each line read as stability, not as a quality score

-- stake outside the 50 % set held by entities no majority set needs

-- HHI concentration < 1,500 = unconcentrated market

-- Shannon entropy equipartition efficiency

Decentralization trend

One sample every six hours; the axis is logarithmic so both counts fit.

Sampled Every 6 h

NC entities (50 %) Entities for 1/3 NC pools (50 %)

Loading history...

Live Minimum Attack Vector Breakdown (Top Entities)

All registered pools, clustered into entities; the threshold sets are cumulative rank groups.

Live On-Chain Telemetry

33.3% Liveness Stall: live liveness entity count

50.0% Minimum Attack Vector (MAV): live MAV entity count

Decentralized Tail: entities below the MAV threshold

#	Entity / Operator	Active Stake (ADA)	Share	Cumulative %	Threshold Set	Tickers (first 6)
Fetching live stake distribution & clustering entities...

How the Nakamoto Coefficient is Calculated

Cardano hands block production to every registered stake pool in proportion to its delegated stake (Ouroboros Praos), so the coefficient is computed from the active stake of the whole pool set.

Why a higher number is safer

Neither 50 % nor 33 % is "good" without context, but the direction is unambiguous: the more independent operators it takes to reach a threshold, the harder collusion becomes. The same snapshot always yields two different counts (entities vs. pools) because clustering changes the count, not the stake.

Pool saturation (the k parameter)

Cardano enforces a soft saturation limit per pool, derived from the k parameter, to move delegation away from over-saturated pools towards smaller ones. The limit in ADA depends on circulating supply and is shown live in the dashboard KPI.

The three calculation steps, and why trackers disagree (33 % vs. 50 %)

Step 1 (Data Collection & Clustering): Measure the active delegated stake per registered pool, and aggregate pools managed by the same operator (operator group, shared owner key or reward account) into single entities for the Minimum Attack Vector (MAV).
Step 2 (Descending Sort): Sort all entities/pools in descending order by active stake.
Step 3 (Threshold Accumulation): Count the top k entities or pools whose cumulative stake reaches the security threshold.

Platform / Methodology	Aggregation	Threshold	Cardano Value	Meaning
Cross-Chain Trackers (33% Liveness)	Entity-Level	33.3%	Live liveness count	PoS Liveness / Censorship stall threshold
Cardano Observer MAV	Entity-Level (Clustered)	50.0%	Live MAV count	Smallest entity set capable of a majority (Ouroboros consensus)
Cardano Observer (Raw)	Single Pool IDs	50.0%	Live pool count	On-chain registered pools required for >50%

The entity coefficient is only as good as its clustering: a shared owner key can merge operators that act independently, and distinct keys can hide one operator behind many pools. A coefficient is also only meaningful with its threshold and aggregation — published rankings usually quote 33 % across chains, while this page reports > 50 % for the coefficient and the 1/3 line separately. Threshold, clustering and inclusion rules dominate the result (SoK: Measuring Blockchain Decentralization, arXiv 2501.18279).

Note on Ouroboros consensus: Unlike BFT networks (which halt when 33% of validators fail), Cardano's Ouroboros Praos uses probabilistic longest-chain consensus. Rewriting the chain or executing a double-spend strictly requires > 50% of the active stake; the 1/3 line describes where block production can be stalled, not rewritten.

Learn more in our technical Cardano Metrics Glossary or read our FAQ guide.