Perp DEX revenue: what 30 versus 90 days actually tells us

Why compare 30 and 90 days?

Annualizing the last 30 days assumes that particular month will repeat. A campaign, new market, volatility burst or accounting correction can make that a poor base case. A 90-day total gives context, but even it is not a full-cycle forecast. We use both windows to ask whether the recent month differs from the preceding two months; we do not multiply either number into a token price target.

For this edition, the 90-day interval is 25 June through 22 September 2026, inclusive, in UTC. The recent 30-day window ends on the same date. The earlier 60-day monthly pace is (90-day total minus recent 30-day total) divided by two. The comparison works only when both numbers describe the same product and metric. The dated figures below came from DefiLlama child fee/revenue series and were checked against its reported 30-day totals. They can be revised by the provider.

Sources: DefiLlama data definitions: fees, revenue and token-holder revenue ↗ · DefiLlama Ostium daily revenue ↗ · DefiLlama edgeX Perps V1 daily revenue ↗ · DefiLlama edgeX V2 daily revenue ↗ · DefiLlama Hyperliquid perps daily revenue ↗

Ostium: a recent income slowdown, with a correction caveat

Ostium's source recorded $1,465,320 of revenue over 90 days, including $36,007 in the most recent 30. The preceding 60 days therefore averaged $714,656.50 per 30 days. The recent month was about 95% below that earlier monthly pace. Fees tell the same directional story: $2,518,700 over 90 days, $126,359 in the recent 30, versus $1,196,170.50 per 30 days in the preceding 60 (about 89% lower). These are within-venue comparisons, not a ranking of Ostium against different business models.

DefiLlama records negative dated corrections in Ostium's fee series; we preserve them rather than deleting inconvenient days. The window arithmetic establishes a change in the source's recorded economics. It does not isolate a loss of users from lower volatility, changed incentives, a product migration or a provider revision. Ostium has no verified live token in our profile, so we describe a pre-TGE business question, not a return to an existing holder.

Sources: DefiLlama Ostium daily fees ↗ · DefiLlama Ostium daily revenue ↗ · DefiLlama data definitions: fees, revenue and token-holder revenue ↗

edgeX: combined V1 and V2 income improved; the buyback bridge remains incomplete

For edgeX, the pinned V1-plus-V2 fee children totalled $12,937,064 over 90 days and $5,050,280 in the most recent 30. The preceding 60 days averaged $3,943,392 per 30 days, so recent fees were about 28% higher. Source-reported revenue on the same two-child scope was $9,040,319 over 90 days and $3,647,716 in the recent 30, versus an earlier monthly pace of $2,696,301.50 (about 35% higher). This is a stronger recent reported run rate; the V2 share and its ramp-up need inspection before calling it durable organic growth.

The holder-revenue row is different: DefiLlama supplied $2,050,550 over 90 days and $614,162 over 30 days for edgeX Perps V1, with no V2 holder-revenue row in that data type at review. Dividing that V1 buyback series by V1-plus-V2 revenue would manufacture a platform payout percentage. edgeX's tokenomics describes supply and buybacks, but a fair-value model still needs the purchase-funding bridge, actual tradable float and vesting dates. Revenue growth alone does not close those gaps.

Sources: DefiLlama edgeX Perps V1 daily fees ↗ · DefiLlama edgeX V2 daily fees ↗ · DefiLlama edgeX Perps V1 daily revenue ↗ · DefiLlama edgeX V2 daily revenue ↗ · DefiLlama edgeX Perps V1 daily holder revenue ↗ · edgeX official tokenomics ↗

Hyperliquid: stronger recent perp economics are not the whole exchange

Hyperliquid's pinned perp child recorded $180,926,925 of fees over 90 days, including $73,914,789 in the last 30. The preceding 60 days averaged $53,506,068 per 30 days; the recent month was about 38% above that pace. Source-reported revenue was $135,614,399 over 90 days and $58,076,314 in the recent 30, versus an earlier monthly pace of $38,769,042.50 (about 50% higher). This does not prove that the next month will grow again.

The perps child includes native and builder-deployed perp activity under its reviewed adapter, while spot and other chain flows are outside this row. Its holder-revenue and revenue series identify the same Assistance Fund economic flow in this source and must not be added together. The documented Assistance Fund mechanism matters for HYPE, but gross trading fees are not a cash dividend to every holder. A token valuation must also test the current FDV, supply path and whether this income pace persists without assuming peak conditions.

Sources: DefiLlama Hyperliquid perps daily fees ↗ · DefiLlama Hyperliquid perps daily revenue ↗ · DefiLlama Hyperliquid perps daily holder revenue ↗ · Hyperliquid official trading fees and Assistance Fund ↗ · DefiLlama data definitions: fees, revenue and token-holder revenue ↗

What this comparison deliberately leaves out

These three venues had aligned source windows and a usable revenue series in the dated audit. They are not the entire tracked universe. GRVT and Variational lacked a pinned fee series in this source; Extended had fees but no verified revenue row; Paradex's recorded perp economics do not include its separate options/RFQ business. Aster's perp-fee row and platform-wide ASTER buyback row refer to different scopes, so they cannot form a single payout ratio. Missing rows are coverage gaps, not zero business activity.

Fees, revenue and holder-revenue definitions describe different destinations for a dollar of trading charges. DefiLlama says fees are what users pay, revenue is the portion retained after suppliers, and holder revenue is the portion attributed to tokenholders. Its attribution need not balance on each day because a revenue-funded buyback may be executed later. Emissions, subsidies, operating costs, dilution and a holder's actual claim require separate evidence. A comparable peer multiple starts with matched product and flow definitions, not three large numbers in a table.

Sources: DefiLlama data definitions: fees, revenue and token-holder revenue ↗ · DefiLlama Paradex perps daily fees ↗ · DefiLlama Aster perps daily fees ↗ · DefiLlama Aster Spot daily holder revenue ↗

When should the investment thesis actually change?

We treat a divergence as a research trigger: sustained income up while FDV is flat may improve the valuation setup, while a rapid FDV rise without retained income may weaken it. Neither is sufficient on its own. We would revise a numerical fair-value range only after repeated comparable observations, a verified token value-capture mechanism, forward supply and unlocks, and scenario assumptions that survive a downside case.

The next collection should record the same 30- and 90-day windows again, then explain how much of the change came from product expansion, trading conditions, incentives or a source revision. For edgeX, a V1/V2 buyback bridge matters more than another gross-volume headline. For Ostium, retention and the effect of provider corrections matter more than projecting one weak month indefinitely. For Hyperliquid, the relevant question is the durable perp income and what actually accrues to HYPE after builder shares and future supply.

Sources: DefiLlama data definitions: fees, revenue and token-holder revenue ↗ · Hyperliquid official trading fees and Assistance Fund ↗ · edgeX official tokenomics ↗

What we monitor next

• Repeat the child-series audit after a complete UTC day and record provider revisions separately from new trading income.
• Reconcile edgeX V1 and V2 token-funded purchases before estimating a platform-wide holder allocation.
• Check whether Ostium's weaker income persists and whether Hyperliquid's higher recent pace survives a quieter market.

Frequently asked questions

Is 30-day revenue a fair-value estimate?

No. It is a recent observation. A token estimate also needs durable income, holder rights, forward supply, dilution and a price or FDV comparison.

Why not add Hyperliquid revenue and holder revenue?

The reviewed perps rows refer to the same Assistance Fund flow under the source methodology. Adding them would count the same economic amount twice.

Does Aster have no buybacks because this article does not compare its perp revenue?

No. The available Aster Spot holder-revenue adapter reports platform-wide ASTER purchases, while the perps fee row has a different scope. This edition does not turn those into a payout percentage.

Compare the evidence

• Ostium metrics and sources →
• edgeX metrics and sources →
• Hyperliquid metrics and sources →
• Aster metrics and sources →

Source register

1. DefiLlama data definitions: fees, revenue and token-holder revenue · checked 2026-09-23
2. DefiLlama Ostium daily fees · checked 2026-09-23
3. DefiLlama Ostium daily revenue · checked 2026-09-23
4. DefiLlama edgeX Perps V1 daily fees · checked 2026-09-23
5. DefiLlama edgeX V2 daily fees · checked 2026-09-23
6. DefiLlama edgeX Perps V1 daily revenue · checked 2026-09-23
7. DefiLlama edgeX V2 daily revenue · checked 2026-09-23
8. DefiLlama edgeX Perps V1 daily holder revenue · checked 2026-09-23
9. edgeX official tokenomics · checked 2026-09-23
10. DefiLlama Hyperliquid perps daily fees · checked 2026-09-23
11. DefiLlama Hyperliquid perps daily revenue · checked 2026-09-23
12. DefiLlama Hyperliquid perps daily holder revenue · checked 2026-09-23
13. Hyperliquid official trading fees and Assistance Fund · checked 2026-09-23
14. DefiLlama Paradex perps daily fees · checked 2026-09-23
15. DefiLlama Aster perps daily fees · checked 2026-09-23
16. DefiLlama Aster Spot daily holder revenue · checked 2026-09-23