IOSG says 86% of Papertrade traders’ net losses were paid to PAPER stakers

IOSG breaks down Papertrade’s cash flows

IOSG researcher Mario Chow published a long-form analysis of Papertrade, a synthetic perpetuals protocol on HyperEVM. The data in the report runs through Oct. 19, 2026 at 10:00 UTC, or 212 hours after launch.

Papertrade went live on Oct. 10. It offers only BTC and ETH markets and allows leverage of up to 1,000x. The team is described as Jez, identified as Colin Hong, a former venture partner at Standard Crypto, and an anonymous developer known as Blurr. IOSG said the project had no fundraising, no premine, and no team allocation. The only way to obtain PAPER, according to the report, is to lose money on the platform.

IOSG said Papertrade had generated $1.361 trillion in headline trading volume. That figure reflects notional volume amplified by leverage of as much as 1,000x. Over the same period, the entire 24-hour perpetuals volume on Hyperliquid stood at $5.2 billion.

The report said the real-money layer was much smaller. About 23,000 addresses deposited $335 million into the protocol. Cumulative loss-making closes totaled $249 million, and traders’ net losses came to $63.2 million. IOSG framed that $63.2 million as the key figure.

Of that amount, $54.1 million, or 86%, was distributed to PAPER stakers. Another $4.7 million remained in the LP pool, while developers received about $4.5 million, or 7%. IOSG said the wallets collecting the payouts were largely the same wallets taking the losses: the top 1,000 wallets held 88% of PAPER, absorbed 81% of all losses, and received 91% of the distributions.

Staking income came in waves, and PAPER now has a quoted price

IOSG described the system as a rebate loop. Staking income did not arrive evenly. It moved in bursts tied to mining returns. During the first 10 hours after launch, stakers received $24 million. A second wave between Oct. 12 and Oct. 16 produced another $24.9 million. On Oct. 17, distributions fell to $800,000. After a perpetual market for PAPER went live on Oct. 18, payouts rose to $2.1 million.

PAPER also gained a visible market price. On Oct. 18, Paragon launched the HIP-3 perpetual para:PAPER on Hyperliquid. It opened at $0.012, climbed to $0.035 within two hours, and later traded around $0.024. Based on 8.69 billion PAPER already issued, that implied a market capitalization of about $211 million. Using the eventual cap of roughly 12.7 billion tokens, the implied value would be about $308 million. IOSG said the $0.024 level was roughly double the cost of mining PAPER through wash-style paired trading. PAPER itself still cannot be transferred.

How Papertrade works

According to IOSG, Papertrade reads a price when a position is opened and again when it is closed. The contract uses a HyperCore precompile to pull the midpoint between the best bid and best ask on Hyperliquid. The trade itself does not execute on Hyperliquid. There is no order book, no funding rate, and no slippage within the per-position notional cap of $10 million. Traders are not facing another trader. They are facing a pool.

That LP pool is the only counterparty, and there is no outside backstop behind it. IOSG compared it to a tank with an overflow outlet:

• Losers fill it. When a position closes at a loss or is liquidated, 98% of the loss goes into the pool.
• Winners draw from it. When a profitable position closes, the protocol pays out according to its rules, subject to the discounts described later in the report.
• If the pool runs dry, unpaid profit enters a first-in, first-out queue and waits for future losing traders to refill the system.
• Any amount above the $5 million cap is pushed to PAPER stakers.

The team announced on Oct. 17 that the cap would be raised to $10 million. IOSG said on-chain reads still showed a stakerCap of $5 million through Oct. 19, and the report said it found no explanation for the delay.

Every close, whether profitable or not, is also charged a 2% fee on PnL. Stakers receive 1% and developers receive 1%. On losing trades, the fee is taken from the loss, which is why only 98% reaches the pool. On winning trades, it is deducted from profit. If winners are queued, fee sharing is suspended.

IOSG stressed that the money in the pool is not promised to any party. How much ends up with winners and how much goes to stakers depends entirely on how trading unfolds.

The report’s central risk point

IOSG said the most important feature of Papertrade is that it does not guarantee winners immediate payment. On a conventional exchange, winning traders are typically paid by market makers, an insurance fund, or the venue’s own balance sheet. Papertrade has none of those. The pool contains only money left behind by losing traders, and the cap is intentionally limited. Once excess funds are swept to stakers, they cannot be pulled back.

That means any winner whose profit exceeds the pool is effectively waiting to be paid by traders who have not lost yet. The contract has no mechanism that forces those future traders to appear. IOSG said this is not strictly a Ponzi because no yield is promised and the house edge is real. Still, from the perspective of a winner waiting in line, the structure depends on new losses arriving later.

The report said the real danger sits in tail winners. A single position is unlikely to drain the pool because the notional cap is $10 million. Even a 1,000x long would need a move of roughly 57% to create a $5 million liability. The issue is that one wallet can open many positions. IOSG cited address 0x2025, which at one point held about 30 ETH short positions of roughly $9.9 million each, for a combined notional size of $297 million. At that scale, a 2% move in the wrong direction would create liabilities larger than the entire pool.

IOSG said 0x2025 pushed the Hyperliquid midpoint on the first night. Over 35 minutes, winners pulled $2.4 million from the pool. The pool held $4.91 million at 21:45 UTC and only $2.48 million at 22:20 UTC. The report said nothing similar happened afterward.

How PAPER is issued

IOSG said PAPER issuance depends on the size of the LP pool. When the LP is below $2 million, every $1 of loss mints 100 PAPER. Once the LP is above $2 million, the mining rate becomes 100 × (120M / (120M + H))², where H is a non-decreasing high-water mark that tracks cumulative inflows above the $2 million threshold.

At the time of the report, H stood at $243 million. That meant each $1 of loss minted only 10.9 PAPER, down from 100 at launch.

IOSG estimated the final supply would approach about 12.7 billion PAPER. Roughly 12 billion would come from the curve itself, and another 660 million were minted during the first hour before the pool reached $2 million and the 100-per-dollar rate still applied. So far, 8.69 billion PAPER had been issued, or 69% of the cap.

Staking rules

Staking does not mint new tokens. The threshold is 1,000 PAPER, there is no lockup, and deposits and withdrawals can be made at any time. PAPER still cannot be transferred, and 99.3% of supply is staked. Rewards are paid in USDC and distributed pro rata by stake size, with no tiers and no acceleration.

IOSG said the documentation sets out three rules that determine when stakers actually receive cash:

1. Queued winners always come first. If the pool cannot pay a winner in full, the shortfall enters a FIFO queue, the pool’s available cash for everyone else drops to zero, and fee sharing for stakers and developers stops. New losses are tracked in a separate bucket used only to repay the queue. Stakers receive nothing until the queue is cleared. IOSG said this has not happened so far.
2. Overflow is swept in batches, not in real time. A keeper calls pushStakerFees() to move funds above the cap into the staking contract. Once swept, that money leaves the pool and cannot be reclaimed by later winners.
3. Between sweeps, timing matters. Winners are paid inside the closing transaction, while overflow waits for the next sweep.

The house edge is not just the 2% fee

IOSG said Papertrade does not charge by notional volume. Beyond the 2% fee on PnL, the protocol’s edge is embedded in liquidation rules and payout discounts on profitable trades.

• The liquidation threshold is brought forward by 5 basis points. At 1,000x leverage, margin could theoretically withstand a 0.1% move, but liquidation happens at roughly 0.052%, wiping out the full margin.
• Profits face a 0.2 basis point dead zone and then an “asymmetric impact” discount. The smaller the move, the harsher the haircut. Under the launch parameters, BTC winners can receive at most 90% of raw profit. On a 1% move, they receive about 88%. On a 0.1% move, they receive about 74%.
• The remaining profit is then charged the 2% fee.

The value removed by those discounts is not paid to anyone directly. It stays in the pool, and once the pool is full it can also overflow to stakers. IOSG gave an example using a 100 USDC BTC position at 1,000x leverage. A 5.2 basis point move against the trader wipes out the full 100 USDC. A 5 basis point move in the trader’s favor should produce 50 USDC in fair-value profit, but the actual payout is only 30.4 USDC. The report said no parameter changes had been disclosed since launch.

Who controls the switches

IOSG said opening, closing, and staking all pass through Papertrade’s relayer. Withdrawals do not. Users can pull funds directly. A 2-of-3 multisig can pause openings, delist markets, change position caps, minimum margin, and minimum order size, and also adjust the impact curve and fee parameters. Contract upgrades are subject to a seven-day timelock. The relayer uses adaptive ordering, with larger orders prioritized.

What sits behind the $1.361 trillion headline volume

IOSG peeled back the layers of the volume figure. The report said the $1.361 trillion headline number appears to count both opens and closes.

• Open interest was $529 million, backed by $10.7 million in margin, for average leverage of about 50x. On day one, average leverage was about 160x.
• Actual deposits totaled $335 million from about 23,000 addresses, while $271 million had been withdrawn. Locked margin peaked at $58 million on the first day.
• $249 million came from loss-making closes. That is the layer where money actually changed hands.
• $63.2 million was traders’ net loss, and IOSG said that was the only layer available for stakers to share.

On that basis, net losses amounted to about 0.46 basis points of headline volume. Developer fee income was about $4.5 million.

Where the $249 million in losses went

IOSG said losing trades poured about $249 million into the pool on a gross basis. That includes every loss-making close, including the losing leg in paired trades used to mine PAPER. The report broke the destination of those losses into four buckets:

1. $185 million, or 75%, paid profitable positions, most of them the opposite leg of the same paired trades.
2. $54.1 million, or 22%, went to stakers.
3. $4.7 million remained in the pool, just below the cap.
4. About $4.5 million, or 2%, went to developers.

After subtracting what was paid to winners, traders were left with a net loss of $63.2 million, and 86% of that amount went to stakers.

Who holds PAPER

There were 16,198 wallets holding PAPER, according to IOSG, and 13,291 of them met the 1,000-token staking threshold. Ownership was highly concentrated, and holders were also users of the platform. The top 10 wallets held 15% of supply, the top 100 held 52%, and the top 1,000 held 88%. Those same top 1,000 wallets accounted for 81% of all losses and received 91% of all distributions.

IOSG estimated that holders lost an average of $0.0073 per PAPER mined. After distributions, the net loss fell to $0.0016 per token. Using the para:PAPER price of about $0.0243, 97% of supply was in profit after distributions. In aggregate, traders lost $63.2 million, got back $54.1 million in cash, and held PAPER valued at $211 million on the perpetual market. IOSG added that this valuation only becomes real if someone is willing to buy, and PAPER itself still cannot be bought or sold directly.

The treasury is not the LP

IOSG drew a distinction between the Exchange contract’s total USDC balance and the LP pool. The treasury held $63.1 million in USDC, made up of about $47.6 million in trader free balances, $10.7 million in locked margin, $4.7 million in LP, and about $100,000 in unswept developer fees.

That means the only capital that actually belongs to the house is the LP, which accounted for about 7% of the treasury. IOSG said the treasury fell from $138 million to around $60 million over the first two days, then stayed between $58 million and $69 million. Since Oct. 12, the LP has remained between $4.3 million and $5.2 million, while daily deposits and withdrawals were broadly balanced, with each side ranging from $5 million to $26 million.

Losses moved with mining incentives

IOSG broke down losses by day. On the first day, loss-making closes totaled $75 million and net losses were $28.4 million. On Oct. 11 and Oct. 12, gross losses were about $20 million per day and net losses were about $4 million per day. A second wave followed. Gross losses rose to $27 million on Oct. 14 and $41 million on Oct. 15, while net losses from Oct. 13 through Oct. 16 ranged from $4.5 million to $7.4 million per day. On Oct. 17, activity dropped sharply, with $6 million in gross losses and $700,000 in net losses. On Oct. 18, gross losses recovered to $11.9 million and net losses to $2.2 million, most of it after para:PAPER opened.

How staker distributions were produced

IOSG said the LP first hit the $5 million cap about 70 minutes after launch. After an overflow event at 15:45 on Oct. 11, the pool spent much of the next half day below the cap. It moved back above the cap on Oct. 12 and, through Oct. 16, net losses largely kept it pinned there.

Daily distributions to stakers were $3.6 million on Oct. 12, $4.2 million on Oct. 13, $5.4 million on Oct. 14, $6.7 million on Oct. 15, and $5 million on Oct. 16. They then dropped to $800,000 on Oct. 17, rose to $2.1 million on Oct. 18, and reached only $70,000 in the first 10 hours of Oct. 19.

The unresolved cap change remains important. The team said on Oct. 17 that the cap would double to $10 million. IOSG said its tracker reads stakerCap from the contract every five minutes and still saw $5 million, including at 09:30 on Oct. 19. If the cap were actually raised, the pool would need another roughly $5.3 million in net losses before overflow resumed. At the pace seen from Oct. 17 to Oct. 19, that would imply four to five days without overflow.

Now that PAPER has a price, yields can be compared directly

IOSG said it now costs $91 to mine 1,000 PAPER by taking direct losses. At launch, the same amount cost $10. Traders using paired positions lose only about 13.7% of the accounting loss, putting their cost at about $12.5 per 1,000 PAPER. The perpetual market values 1,000 PAPER at about $24.3.

Over the most recent 24 hours, annualized distributions worked out to $0.086 per PAPER. That implied a 352% annualized yield at the perpetual price and 94% using the direct-loss mining cost. Over the most recent seven days, including the tail end of the second wave, annualized distributions came to $0.197 per PAPER, or 812% at the perpetual price. IOSG said the perpetual price therefore represented roughly 100 days of the latest daily distributions, or 45 days of the latest weekly distributions, which it interpreted as the market’s view on how quickly income will decay.

If only cash is counted, no cohort of direct-loss miners had broken even. Day-one users recovered about 0.58x, and every cohort after Oct. 11 was at 0.15x or less. If PAPER is marked at the perpetual price, the picture changes. Day-one users were at 2.2x, and Oct. 11 users were around 1.0x. Direct-loss miners entering after Oct. 12 were still below breakeven, ranging from 0.83x down to 0.27x. Paired traders had a breakeven line of about 0.14x, and every cohort had crossed it.

Activity followed mining returns, then picked up after para:PAPER launched

IOSG said hourly volume averaged $56 billion during the first 10 hours, then fell to $7.5 billion from Oct. 11 to Oct. 12, $3.6 billion from Oct. 13 to Oct. 16, and $2.3 billion on Oct. 17. Each turn matched a change in mining returns. The second wave arrived as the mining rate fell from about 26 to 13 and the cap was expected to rise. The drop on Oct. 17 came on the day the cap was supposed to double. On Oct. 18, once the perpetual price moved above mining cost, volume picked up again.

Net extraction per dollar of volume moved with it. IOSG put the figure at 0.50 basis points on day one, 0.21 to 0.34 basis points on Oct. 11 and Oct. 12, 0.61 to 0.76 basis points from Oct. 13 to Oct. 15, 0.13 basis points on Oct. 17, and 0.37 basis points on Oct. 18. The report said that matched the paired-trading mechanism: one leg of a pair always loses in full, while the winning leg is discounted, so heavier wash-style activity leaves more behind per dollar traded.

What para:PAPER is, and what it is not

IOSG said para:PAPER is a perpetual contract deployed by Paragon on Hyperliquid under HIP-3, with the dex name para. It was not launched by Papertrade and it was not launched by Hyperliquid itself. Paragon spent 570.98 HYPE, worth about $45,000, in the HIP-3 auction to secure the PAPER ticker.

This is a pre-market contract. It offers up to 3x leverage, uses isolated margin only, and is expected to convert into a standard perpetual later. Its oracle is an eight-hour EMA of its own marked price, seeded with Paragon’s listing reference price of $0.012. It has no connection to Papertrade. PAPER cannot be transferred, nobody can deliver it, and there is no spot market for arbitrage.

The market opened at 11:00 UTC on Oct. 18 with about $30,000 in open interest. The price sat at the $0.012 reference level for the first hour, close to the paired-mining cost of $0.0125, then moved higher. It reached $0.023 by 13:00, hit an intrahour high of $0.0352 during the 13:00 hour, fell to a low of $0.0198 at 19:00, and later traded around $0.0243. Twenty-four-hour volume reached $13.9 million, or 70% of all Paragon volume. Open interest stood at 203.5 million PAPER, worth about $5 million. Funding was close to zero.

At $0.0243, the 8.69 billion PAPER already issued implied a market value of $211 million. The 12.7 billion cap implied $308 million.

IOSG’s reading of the feedback loop

IOSG said para:PAPER created the first observable price for PAPER and suggested the market was valuing the token at about twice the paired-mining cost. Once the price rises above mining cost, mining becomes attractive again. Before the perpetual launched, gross losses averaged about $300,000 per hour over 24 hours. In the 11 hours after launch, that rose to about $800,000 per hour. Stakers received $800,000 on Oct. 17 and $2.1 million on Oct. 18. By Oct. 19, activity had cooled back to about $350,000 per hour.

The arbitrage cannot be fully closed. A trader can mine PAPER cheaply and short the perpetual, but the perpetual only carries about $5 million in open interest, allows at most 3x leverage, and can never settle into deliverable tokens. IOSG also said holders were not meaningfully using it as a hedge. Among the top 300 holders, who controlled 71% of supply, only six had opened shorts on para:PAPER, totaling 20 million PAPER, or 0.3% of their holdings. Another five had opened longs.

IOSG said the price signal is thin and self-referential. The oracle is the market’s own EMA, leverage is capped at 3x, and about $5 million in open interest is assigning a $211 million valuation to the token. A higher price makes mining more attractive. More mining creates more losses and more staking distributions. Those distributions can then support the price. The report added that this loop weakens over time because the higher H goes, the fewer PAPER are minted per dollar lost.

The bullish case in the report

IOSG also laid out what it sees as the strongest positive points. First, the mechanism has run as advertised for nine days. The house did not raise capital, yet it survived an early $2.5 million drain. The feared winner queue never appeared in snapshots taken every five minutes, and the fair launch can be verified on-chain.

The strongest bullish argument, IOSG said, is structural. Every rule leans in the same direction: losses are full, gains are discounted, and liquidation comes early. If trading is active enough, traders as a group should lose more than they win, and the pool should refill itself. The data matched that view. Over the first 212 hours, traders were net profitable in only 41 hours, totaling $3.7 million. In the other 171 hours, they lost money.

IOSG described this as liquidity grown out of trading itself. Other perpetual launch models usually need upfront capital. Order books need market makers. Oracle-priced pools such as GMX’s GLP or Jupiter’s JLP need depositors willing to take the other side. Launching a HIP-3 market on Hyperliquid requires staking 500,000 HYPE and then sourcing market makers. Papertrade’s pool started from zero, hit the $5 million cap in about 70 minutes, and since Oct. 12 has stayed within $800,000 of that cap without seed capital or LP incentives.

The report said that, with deployment tools and a separate pool for each market, the design could become a template for permissionless perpetual issuance. The documentation already says any Hyperliquid perpetual can be listed, including HIP-3 markets.

What could go wrong

IOSG listed several weak points.

• The token represents future losses, and those losses arrive in waves tied to mining incentives. Stakers received $24.9 million from Oct. 12 to Oct. 16, then only $800,000 on Oct. 17. IOSG called Oct. 17 the only calm day without a mining reason to trade.
• Late entrants face a much higher mining cost. Each $1 of loss now mints 10.9 PAPER, down from 100 at launch, and 69% of the cap has already been issued. Paired trading increases supply without bringing cash into the LP. Once PAPER becomes transferable, IOSG said buying directly may be cheaper than mining through losses for anyone other than paired traders.
• The oracle relies on a single midpoint, with no TWAP and no deviation circuit breaker. IOSG cited Limxn6 as saying that if a paired trade meets a price gap, the trader effectively gets a free straddle and the LP absorbs the tail risk.
• The relayer can censor. It orders by trade size, cannot be externally audited, and both the relayer and liquidation keeper are run by the team.
• The main price signal has no anchor. para:PAPER can change mining incentives and the market narrative without any change inside Papertrade itself. The team’s announced cap increase also was not executed on-chain, leaving stakers unable to forecast income from announcements alone.

Even so, IOSG said the pool is safer than it was at launch. Open interest stood at $529 million, net longs at $64 million, and the pool at $4.7 million. BTC and ETH would need to rise about 3.7% from the average entry prices of those positions before liabilities to winners exceeded the pool plus losing traders’ margin. On day one, that threshold was only 0.7%. The closest point over the past week was 2.2% on Oct. 15, while the largest one-hour BTC and ETH swing was 4.3%.

IOSG’s conclusion and the next things to watch

IOSG concluded that the mechanism has largely delivered on its design. The house did not raise money, it has remained solvent for nine days, no winner has had to queue, and 86% of net losses were returned to stakers, who were mostly the same wallets taking the losses.

Viewed as an income product, PAPER represents overflow that appears in bursts as long as mining remains attractive, plus a 1% share of fees. It now has a quoted price of about $0.024 on a thin perpetual market, roughly double the paired-mining cost. As long as that spread exists, the perpetual market is effectively paying traders to lose money on Papertrade, and those losses feed stakers. If the spread disappears, income falls back to whatever pure 1,000x gambling leaves behind. IOSG pointed to Oct. 17 as the only calm sample day, when stakers received $800,000.

The report ended with four variables to watch next: when PAPER transferability is enabled, whether the $10 million cap is actually implemented, whether new markets are added, and how open interest and the mining rate evolve. IOSG said the current mining rate is 10.9, and every additional $100 million in gross losses would push it down by more than another third. For mining to stay attractive, PAPER’s price would need to keep rising.