The Weekly Take: DefiLlama Discounts Some Perp Volume by 41%. The Number Everyone Quotes Doesn’t Include the Discount.
The most-cited tracker for on-chain perpetuals publishes two volume columns — one it normalizes, one as protocols report it. But only for 24 hours. The 30-day figure the industry quotes, $555.7 billion, is the reported column alone. Apply the tracker’s own haircuts and about $38 billion disappears.
The most-cited tracker for on-chain perpetuals publishes two volume columns — one it normalizes, one as protocols report it. But only for 24 hours. The 30-day figure the industry quotes, $555.7 billion, is the reported column alone. Apply the tracker’s own haircuts and about $38 billion disappears.
Pull up DefiLlama’s perpetuals dashboard and you get two numbers for the same thing.
One column is headed Normalized Volume 24h. The next is Reported Volume 24h. For some protocols the two are identical. For others they are not close. On 3 September 2026, Aster reported $2.762 billion of 24-hour volume and was normalized to $2.205 billion — a 20.2% discount. Lighter: $1.416 billion reported, $1.155 billion normalized, an 18.4% discount. Antarctic, 31.0%. ApeX Protocol reported $1.632 billion and normalized to $962.92 million — a 41.0% discount.[1]
Hyperliquid, the market leader, is discounted by nothing at all. Its two columns match to the dollar.
The dashboard does not say what the discount removes. There is no explanation in the interface, and we could find no published methodology. The one help icon on the volume columns, on Lighter’s row, opens a tooltip reading “This protocol charges no fees for most of its users” — a note about fees, not about verification.[1]
So the tracker is applying a haircut to some venues and not others, in amounts ranging from zero to two-fifths, and declining to say why.
That is the first thing. Here is the second, and it is the one that matters.
There is no normalized column for 30 days. The dashboard’s longer windows — Reported Volume 7d, Reported Volume 30d — carry only the reported figure.[1] The 30-day number is the one the industry quotes. It is the number in this piece’s dek, the number in every “perp DEX market size” citation, and the number we ourselves pulled to write this. It is built entirely from the column the tracker has decided, for some protocols, not to trust.
What the haircut is worth
The dashboard gives the ratio for 24 hours. It does not apply that ratio to 30 days. So we did, as an illustration rather than a measurement, taking each protocol’s 24-hour normalized-to-reported ratio on 3 September 2026 and applying it to that protocol’s 30-day reported figure.
| Protocol | 30d reported | Share of total | 24h haircut |
|---|---|---|---|
| Hyperliquid | $209.81b | 37.8% | 0.0% |
| Aster | $51.41b | 9.3% | 20.2% |
| Lighter | $44.91b | 8.1% | 18.4% |
| ApeX Protocol | $42.74b | 7.7% | 41.0% |
| edgeX | $34.52b | 6.2% | 0.0% |
| Variational | $29.91b | 5.4% | no normalized figure |
| GMTrade | $19.41b | 3.5% | no normalized figure |
| Grvt | $14.54b | 2.6% | 0.3% |
| Pacifica | $10.53b | 1.9% | 2.7% |
| Antarctic | $5.54b | 1.0% | 31.0% |
Across the fourteen largest protocols — 86% of the headline total — the haircuts remove approximately $38.2 billion from the 30-day figure, taking $555.7 billion to roughly $517.5 billion.[2]
A separate $58.0 billion, 10.4% of the headline total, comes from protocols that carry no normalized figure at all. Variational, GMTrade, QFEX and Ondo Finance appear in the reported column and are simply blank in the normalized one. Whether that means the tracker has verified them, has not attempted to, or cannot, the dashboard does not say.
The consequence lands where the citations do. Hyperliquid is 37.8% of the reported total. Because its own volume is discounted by nothing while its four nearest competitors are discounted by between zero and 41%, applying the haircuts consistently would raise its share to roughly 40.5% — a move of nearly three percentage points, produced by nothing but the tracker’s own arithmetic applied to its own numbers.
That gap is larger than the differences the trade press has been arguing about.
How Aster got here
The discount on Aster is not arbitrary. It has a documented history, and it is the clearest available window into what the normalized column is for.
In early October 2025, DefiLlama removed Aster from its perpetuals dashboard. The stated reason was that Aster’s trading patterns tracked Binance’s perpetual futures closely enough to raise questions the tracker could not resolve. DefiLlama’s founder, 0xngmi, said on the record that the venue was “still a black box and we can’t verify the numbers,” adding that DefiLlama was “working on a solution that will include other metrics to make this better.”[3]
Aster’s CEO, Leonard, gave an explanation: opportunistic API traders were opening positions on Aster while hedging on Binance to maximize their rewards from Aster’s Season 2 airdrop campaign. He also argued that “some level of correlation is reasonable as many traders use Binance for liquidity and hedging.”[3] That is a coherent account, and it may be the whole of it.
DefiLlama relisted Aster on 19 October 2025, at Aster’s request. No reconciliation was published, then or since.[3]
What happened instead is the normalized column. Aster is back on the dashboard, its reported volume is displayed, and a haircut of about a fifth is applied beside it — with no statement of what the haircut removes or how it was derived. The verification question was not answered. It was converted into a number, and the number was put in a column most people never cite.
Context matters for scale. In September 2025, the month on-chain perp volume first crossed $1 trillion, Aster contributed $420 billion — 40% of that month’s total, on The Block’s dashboard.[4] Today Aster is 9.3% of a much smaller market. The incentive volume has rolled off, as incentive volume does.
What this does not show
The trade press is not the problem here, and we should say so plainly. We started this piece expecting to find stale figures quoted as current, and that is not what the record shows. Cryptobriefing’s 28 August 2026 piece states its denominator in its own text — 58% of $423 billion across the top eight venues.[5] Its earlier 80% figure is explicitly labelled as a 2025 peak, in an article that also gives 36.4% for January 2026 and 44% for mid-2026.[6] Forklog and CoinMarketCap both covered the decline from the October 2025 peak, with the right numbers attached.[7][8] The coverage dated its figures. Our complaint is with the source those figures are drawn from, not with the people drawing on it.
None of this is evidence of wash trading, by Aster or anyone. DefiLlama raised a verification question. Aster answered it. No regulator, auditor or court has found otherwise, and this piece alleges nothing. What it describes is a tracker applying an undisclosed discount and an industry quoting the undiscounted figure.
The haircut arithmetic is ours, not DefiLlama’s. DefiLlama publishes 24-hour ratios and 30-day reported totals. It does not publish a normalized 30-day figure, and it has not endorsed applying one to the other. Volume mix shifts across a month, so $517.5 billion is an illustration of what the tracker’s own scepticism is worth at current proportions, not a corrected total. The honest statement is narrower: nobody knows what the normalized 30-day number is, including, on the public record, DefiLlama.
High turnover is a flag, not a finding. ApeX Protocol traded $1.632 billion against $139.44 million of open interest on 3 September — 11.7 times its open positions in a day — and carries the largest haircut on the board. GMTrade traded 6.4 times its open interest and carries no normalized figure at all.[1] Fast turnover is normal in perpetuals and proves nothing on its own. It is the kind of thing a published methodology would let a reader evaluate, which is the point.
This is not a criticism of DefiLlama’s existence. It is free, it is the most complete on-chain perpetuals data anyone publishes, and this piece could not have been written without it. A tracker that discloses a discount at all is being more transparent than one that quietly applies it. The gap is between disclosing that a haircut exists and disclosing what it does.
Trackers move. Every figure here was pulled from the live dashboard on 3 September 2026, when the 30-day total was $555.709 billion, 24-hour volume was $21.067 billion, open interest was $22.718 billion, and the weekly change was −30.13%.[1] Rerun it and you will get different numbers. That is the point of attaching the date.
None of this is investment advice. It is a reading of a public dashboard and public statements. The dates are attached so you can check it.
What would change our mind
DefiLlama publishing its normalization methodology. If the tracker documented what the discount removes — which trades, on what test, for which protocols — every figure in this piece stops being an inference. The gap would narrow from opacity to a methodology reasonable people could argue about, which is a healthier place for a market to be and a smaller story for us.
A normalized 30-day column. This is the narrow fix, and it would resolve most of what the piece complains about. If the dashboard published normalized volume at the windows people actually cite, the industry would quote a discounted number by default instead of by accident.
An audited Aster trade log for the September 2025 window. If the trade set behind that $420 billion survived independent verification, the original verification question would be answered rather than absorbed, and the strongest illustration in this piece would lose its force.
A regulated reporting regime for on-chain perpetuals. Standardized volume disclosure is what makes CME’s futures numbers auditable and what on-chain perps have never had. Given the current legislative posture this is the slowest of the four, and it is the one that would make the tracker-of-trackers problem disappear entirely rather than merely documented.
Where this leaves us
There is a real market here. Open interest of $22.7 billion is capital genuinely at risk, and the trades clear on public chains where anyone can in principle recount them. That distinguishes on-chain perpetuals from the centralized-venue volume we wrote about on 7 August, which cannot be recounted by anyone outside the venue.
The problem is narrower and more fixable than a scandal. One tracker sits between the chain and every citation of this market’s size. That tracker has decided some of the volume it displays does not deserve to be counted at face value, has said so in a column, and has not said what the decision rests on. The window where it publishes that judgment is 24 hours. The window everyone quotes is 30 days, where the judgment is not applied at all.
Two columns, one of them discounted by as much as 41%. A 30-day total that uses only the other one. And no published account of the difference. The measurement is doing more work than anyone citing it has been told.
Sources
All dashboard figures pulled from DefiLlama on 3 September 2026 and stated as of that date; published 4 September 2026. Coverage citations carry their own publication dates. Where our arithmetic extends a published figure, the text says so.
- ↩ DefiLlama Perps dashboard, pulled 3 September 2026. Headline metrics: Perp Volume 24h $21.067b; Perp Volume 30d $555.709b; Open Interest $22.718b; weekly change −30.13%. Columns displayed: Name, Normalized Volume 24h, Reported Volume 24h, Open Interest, Reported Volume 7d, Reported Volume 30d — no normalized figure is published at the 7d or 30d windows. Selected rows (normalized 24h / reported 24h / open interest / reported 30d): Hyperliquid $7.038b / $7.038b / $13.518b / $209.814b; Aster $2.205b / $2.762b / $2.56b / $51.407b; Variational — / $1.234b / $1.567b / $29.909b; Lighter $1.155b / $1.416b / $1.262b / $44.908b; edgeX $1.272b / $1.272b / $1.175b / $34.515b; Grvt $488.68m / $490.23m / $427.46m / $14.538b; Antarctic $228.21m / $330.92m / $327.27m / $5.539b; QFEX — / $155.37m / $219.13m / $4.646b; GMTrade — / $1.207b / $188.55m / $19.41b; ApeX Protocol $962.92m / $1.632b / $139.44m / $42.736b; Pacifica $427.88m / $439.7m / $107.59m / $10.527b. The only help icon on the volume columns (Lighter) opens the tooltip “This protocol charges no fees for most of its users.” We found no published methodology for the normalized column.
- ↩ CoinageReport calculation, not DefiLlama’s. For each of the fourteen largest protocols by 30-day reported volume, we took the 3 September 2026 ratio of normalized to reported 24-hour volume and applied it to that protocol’s 30-day reported figure. Those fourteen protocols represent $479.8b of the $555.709b total (86.3%). Of that, $421.8b sits at protocols publishing both columns; applying their ratios yields $383.6b, a reduction of $38.2b (9.1%). A further $58.0b (10.4% of the headline total) sits at protocols with no normalized figure and is left untouched. Hyperliquid’s share of the resulting $517.5b is 40.54%, against 37.76% of the published total.
- ↩ CoinDesk via Yahoo Finance, “Aster Back on DeFiLlama, But Wash-Trading Fallout Still Not Resolved,” October 2025. DefiLlama delisted Aster in early October 2025 over similarities between its volume and Binance perpetual futures; 0xngmi on record: “still a black box and we can’t verify the numbers,” and “we’re working on a solution that will include other metrics to make this better.” Aster CEO Leonard attributed the pattern to traders opening positions on Aster while hedging on Binance for Season 2 airdrop rewards, and said “some level of correlation is reasonable as many traders use Binance for liquidity and hedging.” Relisted 19 October 2025 at Aster’s request; no reconciliation published.
- ↩ The Block, “Perp DEX monthly trading volume tops $1 trillion for first time,” covering September 2025, using The Block’s own data dashboard rather than DefiLlama: $1.05 trillion for the month; Aster $420 billion (40%); Hyperliquid $282.5 billion (26.9%); Lighter $164.4 billion (15.7%).
- ↩ Cryptobriefing, “Hyperliquid leads growth among top perp DEXs with 58% market share,” 28 August 2026. 58% of $423 billion across the top eight perp DEXs over 30 days; Hyperliquid’s own volume given as a $187–245 billion range; underlying data source not named.
- ↩ Cryptobriefing, “Hyperliquid captures 80% of decentralized perpetual trading volume,” 26 June 2026. The 80% figure is presented as a 2025 peak, with 36.4% given for January 2026 and 44% for mid-2026.
- ↩ Forklog, “Perp-DEX Trading Volumes Plummet Over 50% Since October,” using DefiLlama data: peak of $1.36 trillion monthly in October 2025, falling to $699 billion in March 2026.
- ↩ CoinMarketCap Academy, “Perp DEX Volumes Hit 9-Month Low After October Peak.” $1.36 trillion October 2025 peak; $699 billion March 2026; Hyperliquid approximately 34% of the top ten venues at $185.5 billion over 30 days.
This article is not investment advice. It describes a public dashboard and public statements. Comments are closed sitewide. Corrections to corrections@coinagereport.com.
Discover more from CoinageReport
Subscribe to get the latest posts sent to your email.
This article is for informational purposes only and is not financial advice.
