The HHI of Tokenomics: Measuring Concentration Risk

Antitrust regulators have measured market concentration for fifty years using the Herfindahl-Hirschman Index. The metric translates almost perfectly to tokenomics — and most founders ignore it.

The formula

HHI = sum over allocations of (allocation.percent ^ 2)

// Examples:
// 100% in one allocation:           100^2 = 10000
// Two equal halves (50/50):         50^2 + 50^2 = 5000
// Four equal quarters (25/25/25/25): 4 × 25^2 = 2500
// Ten equal tenths (10 × 10):       10 × 10^2 = 1000

The squaring is what makes it useful. If you doubled the number of allocations but kept the largest holder unchanged, HHI barely moves. The metric correctly rewards both spreading allocations widely and shrinking the largest one. It penalises lazy “we’ll add five small allocations to look diverse” designs that keep a 50% top holder.

The DOJ thresholds, applied to tokenomics

The US Horizontal Merger Guidelines ↗ define three concentration tiers. They map directly to tokenomics:

Our internal tooling treats HHI > 2,500 as a flag worth raising, matching DOJ’s “highly concentrated” threshold. It’s a crude single-number summary, but the squared term makes it remarkably good at flagging real problems.

HHI across the 13 real protocols

Computed across the launch-day allocations encoded in our dataset (each protocol’s top-level buckets, before sub-tranching like seed vs Series B within an “investors” line):

The takeaway is uncomfortable: most successful tokenomics fail the antitrust concentration test. That’s not a bug — it reflects that protocols are run by treasuries and DAOs that are fundamentally large single-entity holders. The signal isn’t whether HHI is high; the signal is whether the top holder is credibly distributed (a treasury voted by token-holders) or nominally distributed (a foundation controlled by founders).

Why HHI beats top-3 concentration

A common alternative metric is “top-3 percent of supply”: just sum the three largest allocations. It’s simple, but it has a hole. Consider two designs:

Design A is more concentrated than Design B (a single 60% holder dominates), but its top-3 sum is lower. Top-3 doesn’t see the difference between “one big dog and two medium ones” vs “three near-equal big ones.” HHI does. The squared term punishes the 60% holder much more than three 33%s.

Top-3 says “how big are the giants”. HHI says “how much does any one giant tower over the rest.”

What HHI doesn’t see

Two important things HHI misses, both of which matter when interpreting the score:

• Vesting status. An allocation’s 50% share doesn’t translate to 50% of circulating supply if 95% is locked. HHI of launch-day allocations and HHI of TGE-circulating supply can be very different. PENGU’s 70.7% TGE float means HHI of circulating ≈ HHI of total. Whereas Ethena’s 5% TGE float means the only allocation with circulating tokens is the airdrop, so HHI of circulating is essentially zero — everything is locked.
• Composition of the top holder. A 40% “DAO Treasury” allocation is structurally different from a 40% “Investors” allocation, even though HHI treats them identically. The first has hundreds of token-holders making decisions; the second has 6–12 funds.

For these reasons, HHI is best read as a starting question, not an answer. “Why is your HHI 3,500?” opens a useful conversation. “Your HHI is too high” doesn’t.

Two refinements

If you want to go beyond the basic HHI, two adjustments are common in academic tokenomics work:

// Re-weight by vesting at month t before squaring:
circulatingHHI(t) = sum over allocations of:
    (allocation.percent × unlockFraction(allocation.vesting, t)) ^ 2

// This drops insider weight in early months, climbs as cliffs unlock.
// Useful for stress-testing month-13 concentration after a cliff.

// Treat allocations controlled by the same entity as one share.
// Common case: "Team" + "Future hires" + "Advisors" all controlled by founders.
// Or: "Treasury" + "Foundation" + "Operations" all controlled by foundation board.

entityAwareHHI = sum over entities of:
    (sum of percents controlled by entity) ^ 2

Entity-aware HHI is the harder one — it requires reading governance docs and making judgement calls about who actually controls a multisig. But it’s also the more revealing one. A protocol with three foundation-controlled buckets totalling 50% should not get credit for “diversification” in HHI just because the buckets have different names.

Practical use

Three ways to use HHI in tokenomics design:

• As a sanity check. If your design’s HHI is above 3,000, ask yourself who owns the dominant bucket and whether they’re going to be comfortable being seen as that dominant.
• As a comparison frame. “We targeted an HHI similar to Uniswap and Arbitrum” is a defensible investor-deck claim. “Our concentration is healthy” is not.
• As a tool-callable check. The MCP’s compute_health_score returns HHI as one of its outputs. You can wire it into CI for token-launch infra: fail the build if HHI > some threshold.