The past few months have been trying for crypto markets. Poor risk management at centralized lenders like Celsius, Voyager, and BlockFi has led to bankruptcies and FTX bailouts. Fortunately DeFi giants like Compound and AAVE are weathering the storm, but Rari's Fuse protocol is dead in the water and we all know what happened to Anchor. On the AMM side of things, Bancor paused its IL protection program, and late last year we learned that 50% of Uniswap V3 LPs are losing money. TVL in LP management services has stagnated, and even Paradigm's LP research falls flat in real-world market conditions.
So, with this in mind... wat do? The incumbents seem distracted (Compound is pivoting to a strange only-borrow-USDC model, AAVE is building a stablecoin, and Uniswap is dealing with the SEC), so I think it's up to new-comers to innovate. We're starting to see this in protocols like Euler Finance and Silo Finance, which are innovating on oracle reliability and risk isolation. But there's an opportunity to do more. If we give borrowing a new purpose, we can reinvigorate lending markets and solve AMM IL at the same time.
On the supply side, █████ will combine the simplicity of Compound with the innovation of Euler/Silo/Idle. You'll be able to deposit any asset, customize risk exposure (think personal Fuse Pool config) and watch yield accumulate. On the borrow side, █████ enables something new: permissionless market-making on margin with up to 40x leverage for Uniswap V3 positions. Institutional and JIT/MEV market-making activity will drive real borrow demand, over and above the speculative usage we see in other lending markets.
As these sophisticated, high-frequency actors gobble up more of the swap-fee pie, I expect passive LPs to migrate to █████. This allows them to earn a portion of market-making revenue (via MM interest payments) without worrying about IL and other AMM details.
Permissionless asset listing via Uniswap V3 TWAP oracles (like Euler)
Isolated, two-asset money markets for every Uniswap pair -- contains risk and allows for higher collateral factors (like Silo)
Built-in aggregator to optimize yield across whatever set of lending pairs you choose (like Fuse + Idle)
Market-making on margin with up to 40x leverage on Uniswap V3 positions (NEW)
Dynamic interest rate models to minimize governance overhead and maximize capital efficiency (NEW)
A new liquidation engine to keep markets healthy despite massive leverage (NEW)
Like structured products, automated LP strategies only make money in certain market conditions. As you try to expand the set of market conditions in which they're profitable, you realize you need more data or faster execution than is reasonable for trustless on-chain vaults. Worse yet, smart contracts reveal intentions ahead of time, making them easy to frontrun. █████ fixes this by delegating responsibilities: passive users deposit liquidity while sophisticated borrowers manage it. This allows borrowers to run fancy off-chain algorithms, change their strategies in realtime, and avoid being frontrun.
That said, automated LP strategies still have their place. TWAP oracles need a certain amount of liquidity to function properly, and DAOs often act as liquidity-providers-of-last-resort (Olympus, Redacted, etc.). For those use-cases, you care more about LP utility than profitability, i.e. "Can my users make trades?" vs "Am I always making money?". Full-range/ambient allocations like Aloe Blend are perfect for these scenarios.
I ran some simulations, and even just for the USDC-ETH pair, millions of dollars are being left on the table. Right now a lot of value is captured by arb-bots, but informed MMs should be able to take over. This system also allows MMs to operate in an ETH-XXX pair without holding XXX on their balance sheet.
For a long time, it's been tough to imagine mass-market appeal of CFMM yields. xy=k looks simple, but behavior is often counterintuitive and everyone is (rightfully) scared of IL. The interplay of the two assets and how that impacts your portfolio is just not gonna make sense to the average person. █████ simplifies all that -- it's just interest now. Once it's built, I'd like to see integrations across L1s/L2s, privacy layers like Aztec, and mobile apps/wallets. Maybe there's even a path to debit card functionality (like the failed centralized lenders) while preserving self-custody of interest-bearing tokens. Opportunities are endless once AMM yields fit the interest rate mold.