Sr. August II (@SrAugustII) header
Sr. August II (@SrAugustII) profile picture

I check what protocols claim against what the chain returns. Payment rails, tokenized equities, AML. Full cycles documented as case files. Positions disclosed.

Spain
Joined February 2025
Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)Photo by Sr. August II (@SrAugustII)
From 18 to 21 September, Pendle's API gave a Robinhood Chain market, sNUKE, an implied rate of exactly 350%. I said publicly it was pinned to an AMM bound. Then I read the market contract. It never recorded 350%. New article on what I got wrong.
1
23
Aave prices tBTC at BTC, so only a BTC drop liquidates it. Three days ago a 30% drop handed liquidators about as much tBTC as the on chain exit could take. Today the same drop hands them 245 tBTC and the exit takes 135, with redemption still paused.
7
The Binance exploit position on Aave was liquidated yesterday, half the debt in one call, as the close factor dictates. The liquidator's bonus was 948k. It kept 559k after selling 24.4M USDT for ETH inside the same block. Not sure which number says more about the exit.
6
On Aave's Monad market, 50% of the 512M supplied is collateral that only works in e-mode loops. USDe 125M, syrupUSDC 86M and PT-AUSD 43M sit there to borrow stablecoins at 90 to 93% LTV. PT-AUSD collateral grew by 6M since yesterday. USDC is now 93% utilised and its deposit rate is 5.8%. The PT-AUSD behind 43M of those loops has a 2.4M Pendle pool next to it and matures on 17 December. When the loops unwind, where does the exit liquidity come from?
10
The Binance exploiter's Aave position is 3.3% from liquidation. It won't close at once. At ETH 2,366 a liquidator repays half the debt and takes 24.5M USDT. Health factor jumps to 1.18. The other 35.5M stays until ETH 2,800. Do liquidator bots check whose position they close?
2
7
The Binance exploiter still has an open position on Aave. 60M USDT supplied, 19,762 WETH borrowed, health factor 1.054. Tether has blacklisted the wallet, but the USDT sits in Aave's pool, outside the blacklist. With markets frozen, USDT supply pays 8.8%, about 14.5k USDT a day to that position. At ETH 2,366 it becomes liquidatable. Would a regulated liquidator take that USDT?
12
On 22 October about 4% of Gauntlet USD Alpha, the vault behind CoolWallet's USDC yield, was lent two layers down against reUSD or PT-reUSD. Today it is about 10%. Since Sunday the Prime sleeve on Base went from 63% to 54% of the vault and a USDT Prime sleeve appeared at 11%. Frontier still lends 29% against reUSD and PT-reUSD at 91.5% LLTV, and the USDC RWA sleeve 60%. Nothing wrong with any of it. But a wallet user sees one line, USDC earning about 5%. Who tells them when the mix underneath moves?
16
When 33 PT-reUSD positions were liquidated on Morpho on 25 August, where did the seized collateral go? I traced every transaction. 38.6M PT seized in 14 minutes, no bad debt. Only 1.6% was sold into the Pendle pool, about 1.1% below the oracle. 96% was never sold. One liquidator flash-borrowed, repaid, took the PT, put it straight back as collateral in the same markets and borrowed 31M against it. The positions changed hands at roughly 88% LTV. The exposure never left the market. So in a liquidation like this, who actually takes the risk, the pool or the next borrower? I hold PT-reUSDe, not PT-reUSD.
18
Is the Aave App making DeFi easy, or just convenient? Convenient, clearly. Bank transfer in, no wallet, no gas, a balance that grows every second. That will bring a lot of people on chain. Easy would mean the saver can see what sits under the number. Three things I couldn't see today. Where the 6.00% base comes from. Across Aave V3 today the big USDC markets pay 3.85% on Ethereum, 4.20% on Base and 4.95% on Monad, and none of the stablecoin markets above 100M reaches 6%. Boosts for referrals and recurring deposits take it to 7.25%, and above 100k the rate drops to 4.50%. What Balance Protection covers. The page says it is not insurance and that it addresses certain loss events such as security breaches or technology failures. A first time saver will likely read it as protection for the money. The risks of a lending market, a liquidation that comes too late or a stablecoin losing its peg, are a different thing. What happens if many savers withdraw at once. The page says withdraw anytime. On Aave that depends on how much of the market is lent out at that moment. None of this makes the app bad. It makes the gap visible. Who explains to a first time saver what sits under the rate, and tells them when it changes?
1
1
37
Two months after the PT-reUSD liquidations on Morpho, I went back to see what had changed. Less than I expected. On 25 August a PT price drop of about 3% liquidated 36.1M in fourteen minutes. Today the same two markets carry 134.9M of debt against 170.1M of PT, still at 91.5% LLTV. The oracle is still Pendle's 15 minute TWAP. I read the feed and the TWAP at the same block and they match to six decimals, so a pool of about 13M sets the price for 134.9M of loans. Using each position's distance to liquidation from Morpho's API, a 3% move would put 37.3M past the line, almost three times what the pool holds. Who is on the other side? Three curated vaults lend into these markets. In one of them, 93% of deposits come from a single wrapper that a retail wallet uses for its stablecoin earn product. Savers who see a steady rate are funding leveraged PT loops, mostly without knowing it. The PT matures on 10 December, and 134.9M has to be repaid or rolled before then. What I can't see on chain is the plan. How do curators size debt that becomes liquidatable at small moves against the depth they can actually sell into, after August?
46
About 42.7M of the October PT-USDG sits as collateral on Aave X Layer, most of it in E-Mode loops, ten days before maturity. The Pendle pool next to it holds about $0.11M. The February PT isn't an Aave reserve yet, so I can't see where these positions go on maturity day.
1
12
1/ Pendle's USDai and sUSDai October pools on Arbitrum lost about $54.7M the day after maturity. I followed the money on-chain. 2/ One address, 0x4a74...3bf3, took out $50.05M in a single transaction at 07.40 UTC on 15 October. That's 91.5% of everything that left. 52 addresses exited in total, and the top 5 were 97%. 3/ About four hours later it redeemed the 50.07M USDai to PYUSD one to one, then bridged it to Ethereum in six packets in under an hour. 4/ On Ethereum it split in two. 22.0M repaid a PYUSD loan on Morpho Blue backed by kBTC and released 390 kBTC. The other 28.07M went through three hops and landed at an address Etherscan labels Coinbase 10. About seven hours from pool to exchange. 5/ Only about $1.9M of the 54.7M rolled into the February pools. 6/ While the money moved, 27 zero value and fake PYUSD transfers hit lookalike addresses of the two receivers. Address poisoning in real time. None of the real funds went to a fake address. 7/ What I can't see from the chain is why this position exited instead of rolling. The same two Ethereum addresses appear to have funded it in April, with PYUSD borrowed on Morpho. If anyone knows the strategy, I'd like to understand it.
1
42
Pendle's USDai and sUSDai pools on Arbitrum matured at midnight UTC. I read the four market contracts yesterday morning and again today. The October pools held about $61M. Seven hours after maturity they still hold about $60M. What changed is the mix. Yesterday it was a bit over 40% PT. Today it is about 95% PT, because holders sold PT into the pools before maturity and took the SY out. The February pools on the same assets went from $3.7M to $4.1M. So far only about $0.3M looks like a roll. Most of the October liquidity hasn't moved yet. LPs can withdraw any time, and after maturity the PT redeems at par. I'll read the pools again tomorrow to see whether the money goes to February or leaves Pendle. I own some PT-sUSDai, both October and February.
32
Pendle ORBIO on Robinhood Chain expired at midnight UTC. In its last two hours the contract's implied rate climbed to 54%, on three PT sales. Half an hour out, that rate was a 0.003% price gap. Near expiry, annualised rates shout about nothing.
34
ORBIO on Pendle, Robinhood Chain, matures Oct 8. I checked the contract. The implied rate went from 4.34% to 13.65% in two swaps and then sat there five days with no swap at all. One swap yesterday took it back to 4.32%. In a thin pool the rate is just the last trade.
16
People did trade it. While the API said 350%, the contract's rate moved between 424% and 520%. On the same chain, the same morning, SHROOM and microduck matched their contracts to about 14 digits. Of the three, only sNUKE split.
1
1
To check a rate yourself, pull the market's UpdateImpliedRate logs, take e to the logged rate minus one, and compare it with the API at the same hour. If you read a pool's PT and SY, convert SY to underlying first. I still do not know why sNUKE split.
1
Three yield markets on the same chain expired last night. The deep one, the thin one and the dust one. We logged all three daily for a week and none of them died the same way.
1
25
sNET closed at midnight with 1.5M traded on its last day. What stayed in the pool was PT heavy, about 4 PT for every SY.
1
Correction on my own count. By value, not units, the sNET pool closed about even, since each SY was worth about 4 PT at expiry.