Everything onchain capital markets. building @credditxyz
- Tweets 2,061
- Following 541
- Followers 2,589
- Likes 8,905
The whole debate between isolation and pooled lending models is only semantics
in practice on the supply side they are all pooled models
- Aave v4 hubs pool liquidity and allocate to spokes
- Meta Morpho vaults pool liquidity and allocate to markets
- Fluid instances pools liquidity and allocate to single collateral vaults
- Aave v3 instances pools liquidity and allocate via cross margin supply and borrow caps
They differ in terms of features such as oracles, rehypothecation, governance, circuit breakers, ability to set risk premia etc.
that is they differ in terms of tools available to a risk manager to manage the "pool"
the only thing isolated on the supply side is directly supplying to a morpho market but that is not happening in practice
Aave V4 myths
“Aave V4 doesn't isolated markets.”
No. Aave V4 hubs and spokes are isolated by default based on their risk profiles. Risk-adjusted markets can share liquidity through hubs, up to defined caps, supporting new use cases without unnecessarily fragmenting liquidity.
Full liquidity isolation is often counterproductive: it fragments capital, reduces utilization, and increases costs for users. These trade-offs become even more visible when incentives used to bootstrap isolated liquidity eventually run out.
“Hub-and-spoke fragments liquidity.”
It’s the opposite.
In V4, spokes represent lending markets, while hubs can share liquidity across those markets. This allows isolated risk profiles to access pooled liquidity, improving utilization and capital efficiency.
“It’s just isolated markets. Aave is catching up with curated vaults.”
A curated vault typically launches with zero liquidity and requires capital or incentives to bootstrap.
A V4 spoke can launch with the entire hub balance sheet behind it from day one.
That’s the difference between an isolated market and an isolated risk profile with access to pooled liquidity.
“V4 is complex.”
The architecture is simpler while remaining flexible enough to support a wide range of use cases. The overall codebase is also significantly smaller than Aave V3.
“V4 is still a new deployment. It’s too early to use.”
V4 is already securing $1.2B in deposits and is deployed across multiple networks, including Ethereum, Avalanche, and Arc.
V4 is already scaling.
“V4 is less open to curators.”
V4 already supports third-party curators such as EtherFi, with more to come.
The key difference is that curators can build and manage an entire market structure, rather than simply manage deposits inside a vault. This gives them the opportunity to participate in the economics of the broader lending market instead of being limited to fees on deposit AUM.
Over time, curators and integrators should be able to own more of their market structure and retain more of the economics they create.
4
281
The launch of yet another L1 with @arc is the least exciting news of the day that I can think of
1
10
203
99% of the work routers do is identical.
It's a waste of capital.
Fynd (router from @PropellerSwap) is open source.
Everyone should get behind it and improve it!
Routing would get better, more robust, and more transparent
edge can come from private orderflow etc.
This should be titled "0x routing made a mistake"
If you're an aggregator, you can't just route to arbitrary hooks! It's why Uniswap's own router has an approval process developers.uniswap.org/hook-…
4
3
19
1,287
letsgetonchain retweeted
This is now my favourite page on Fluid (which we had on Juplend for a while)
It shows every multisig, its authorities, timelocks, oracles, feeds, risk research docs, and much more!
5
8
45
6,290
LPing on AMMss is just really fun
might be a good idea to model that into utility curves before writing off AMMs
good guide on how to LP here:
nitter.net/DefiIgnas/status/20951…
good tools @revertfinance @vfat_io
Ultimately all talk around LVR arguing passive LPs on AMMs have no future because they are loss making in expectation, assume all market participants have identical utility curve
That one utility curve seeks to maximise riskless profit (what remains for a market maker after hedging costs and their cost of capital)
The truth is that not all market participants share this same utility curve
AMMs allow all the utility curves to express themselves
3
18
2,474
claude is down. time to checkout memes on robinhood
whats good?
btw uniswap UI is still feels awful in 2026
lifetime fumble
incomprehensible to me why they cant get a descent product manager to make a usable UI
4
13
981
So every mainstream media outlet is about to write about AMMs cause we’ll see GameStop like squeezes caused by pairing stonks with memes on AMMs
It couldn’t have been written better
8
498
finally got down to study @aave v4 this weekend
Its cool! But I honestly don't see any difference to how Fluid works besides new terminology
The Aave Hub <-> Fluid liquidity layer
- sets interest rates, sets supply and borrow caps that for the downstream aave Spokes / Fluid borrow vaults
The Aave spokes <-> Fluid borrow vaults
- modular building blocks that can set LTVs, collateral risk premiums (surcharge over borrow rate set by the fluid liquidity layer / aave Hub) etc.
Liquidations:
Aave v4 adds the ability to only liquidate the minimum required to restore health, something that has been the default on Fluid since its launch.
A new aave Hub <-> a new Fluid liquidity layer instance like @Bitwise x @ethena deployment
Any nuances I am missing?
7
41
4,255
letsgetonchain retweeted
I’ve been studying different DeFi protocol designs lately, so I decided to fully break down how @0xfluid works from a technical point of view.
40+ pages on liquidity layer, vaults, ticks, branches, and liquidations:
rajkoz.com/posts/fluid-vault…
12
23
177
14,034
dont want to simp too hard but @DCinvestor is probably my favorite follow
good takes, good taste (what a collection🤯), always principled, long term visioned
"bro, do you even own NFTs anymore???"
yes, yes i do
opensea.io/DCinvestor
gallery.so/dcinvestor
1
8
1,405
Narrative violation?
USDC Lenders demand a premium (~0.3% APY) when lending against isolated highly liquid crypto collateral vs a diversified set of collateral on @aave v3
This is the period after the kelp incident
2
6
721
letsgetonchain retweeted
Here’s a list of Morpho PT markets using TWAP prices that may be vulnerable to similar oracle attacks:
• PT-USDat-14JAN2027
• PT-sUSDD-27AUG2026
• PT-AUSD-8OCT2026
• PT-USDat-27AUG2026
• PT-USDG-24SEP2026
• PT-USDAi-15OCT2026
Some of these assets like USDat are great collaterals (in my opinion) as USDat is backed only by US T-bills.
But because they use a TWAP price oracle, the loopers who loop these PTs with very high leverage will easily get liquidated in a similar oracle attack.
If you are already looping any of these PT assets, my recommendation is to maintain a health ratio of at least 1.05-1.06 for your leveraged positions.
I also highly recommend curators to switch the oracles for these markets to fully linear discount oracles if you want to protect borrowers.
This really sucks.
A lot of loopers in the PT-reUSD market have been liquidated a few hours ago.
This was possible due to the design of the PT oracle.
A lot of oracles for fixed yield PT tokens are based on the 15-30 min TWAP price of the PT tokens on Pendle.
The issue with that is that if a whale suddenly buys a lot of YT tokens, then the PT price decreases and loopers can get liquidated if their leveraged PT position is close to the liquidation LTV.
This is exactly what happened with the PT-reUSD market, leading to $30M+ in liquidations, with the PT-reUSD price dipping 2.8% in minutes.
I think @SteakhouseFi made a mistake by using a 15-minute avg. TWAP price oracle for its PT-reUSD market on Morpho and I hope they adjust their oracle designs in the future to avoid such scenarios.
The oracle teoretically worked as intended, but because it was based on the PT-reUSD TWAP price and was easy to manipulate, it led to millions in losses.
The most reliable oracle type for PT markets would be a fully linear discount oracle in my opinion that doesn't take into consideration the PT TWAP price.
A fully linear discount oracle would enable looping with a lower maximum leverage I guess, but it would no longer be affected by the PT price volatility.
12
13
65
13,188
Looping PTs on morpho given the oracle configuration is a bad r/r
The oracle marks PTs at the lower of the discount to par AND secondary PT market price.
@pendle_fi secondary markets are illiquid.
Combine that with high leverage for PTs and its the perfect recipe to get rekt.
In general never run a levered carry trade where your yield leg is marked on secondary.
On @aave this is not the case PT-sUSDe is marked at redemption value minus a controlled discount.
No secondary affects your collateral leg.
on @credditxyz you can view oracle configurations for any carry trade, and study what exact scenarios can lead to liquidations!
The Morpho @re PT-Reusd market was just the victim of an oracle manipulation attack, due to mis configured oracles the wallet below was able to purchase millions of YT’s to liquidate approximately ~37m worth of loops
debank.com/profile/0x854e3f3…
The wallet is freshly funded, curious to see how the Re team handles this, stay safe out there 🫡
4
1
27
2,991
Ultimately all talk around LVR arguing passive LPs on AMMs have no future because they are loss making in expectation, assume all market participants have identical utility curve
That one utility curve seeks to maximise riskless profit (what remains for a market maker after hedging costs and their cost of capital)
The truth is that not all market participants share this same utility curve
AMMs allow all the utility curves to express themselves
4
1
35
7,419
Im curious on market share numbers for the different ways to batch multiple calls into one tx on Ethereum.
I'm thinking market share of the below:
1) EIP 7702 + EOA
2) ERC 4337 UserOps
3) @safe smart account that is configured 1/1
anyone has opinions why one dominates over the other?
for example i was wondering why @DeFiSaver offers only smart account batching instead of EIP 7702 for users with supporting wallets?
6
11
697