Today in DeFi

Today in DeFi

Guides

Metric’s PropAMM Earns >40% APR and Beat Arb Bots

By Danger, Today in DeFi Founder

Nov 20, 2026
∙ Paid
Subscribe for daily free DeFi news covering launches, tradable catalysts, and actionable farming opportunities.

Today’s News Digest:

  • Lookonchain tracked the Binance hacker moving another 137,602 ETH ($248M). It was the largest of three daily batches.

  • Liquity shared K3 Capital’s sBOLD post-mortem. The vault lost about 3.43M BOLD, or 99.9% of its assets, and about 131k BOLD of the attacker’s loan fees landed in Yearn’s yBOLD.

  • Ledger’s CTO is leaving at the end of 2026.

  • Sui launched Hashi mainnet, which mints hBTC 1:1 against BTC. Fluid and Suilend opened hBTC markets, and Suilend took $5.8M of hBTC in under 8 hours.

  • Aave brought V4 close to $2B in deposits. V4 on Base hit an all-time high in stock-backed loans, and the Android app opened more early-access spots.

  • Lido opened its November DAO votes. Onchain Vote #207 concerns Lido Lend, which is a governance framework rather than a lending launch.

  • 1inch added perps through Hyperliquid in its dApp and wallet. Users fund positions with tokens they already hold, in non-restricted regions only.

  • Revert launched Revert Lend on Aero. Users can borrow against LP positions, including tokenized stocks.

  • Lista launched lending on Robinhood Chain. Its fixed-rate loans also launch on Binance Wallet.

  • Arbitrum reported $7.4M in Q3 revenue to the ARB-controlled treasury, up 200% quarter-on-quarter.

  • Pendle added Ethena USDe markets with quarterly maturities through November 2027. It also added Axis USDx and sUSDx markets.

  • Yearn said yvUSDS-1 is earning more than 20% APY. TVL in the USDS staking pool fell about 85%.

  • Hyperbeat introduced USD+, its flagship stablecoin product on Hyperliquid. It spreads funds across Morpho, HyperLend, Hyperliquid Earn USDC, Liquid Bank credit markets and delta-neutral strategies.

  • Arch Prime opened pre-deposits in USDC (Ethereum or Base) or BTC, with a $10 minimum. Depositors earn yield from day one and stack points for early access.

  • Derpetual introduced its Abstract Market Launcher.


Find and execute the 15%+ stablecoin yields in minutes
Save time without checking every protocol manually. DeFi Saver’s Discover page surfaces rates across Aave, Morpho, Spark, and more, lets you simulate leverage before committing, and handles the full loop in one click. Stop leaving yield on the table.

<Discover Yields on DeFi Saver> | Today in DeFi is supported by DeFi Saver


TL;DR

  • If you LP on Uniswap, arbitrage bots trade against you every time your pool’s price goes stale. On many volatile pairs, that costs you more than you earn in fees.

  • A prop AMM takes its price from an outside feed and quotes around it. Arbs can’t trade against a stale price because the price is never stale. Until now, every prop AMM ran on its own team’s capital.

  • Metric is the first to open it up. A price provider supplies the feed, a curator sets the pool, and anyone can LP. LPs earn the spread instead of a swap fee.

  • Early evidence is good. Castle Labs measured Metric WETH/USDC LPs at +0.44 bps per dollar traded, against −1.01 bps for Uniswap v3 (Oct 15–26).

  • The risks are real: inventory risk in a trending market, trust in the curator and the feed, fee shares that can change with no time lock, young contracts, and a mostly anonymous team.


<Explore LPs on Metric>

Why LPs Should Care

If you’ve LP’d on Uniswap, you’ve been paying arbitrage bots. Metric is built to stop that, and it’s already handling serious volume.

Metric did more than $8 billion of volume in its first month on more than $25 million of TVL, including one day above $600 million (team figures, Nov 15). DefiLlama comes close: $21.51M of TVL on Nov 12 and $7.3B of volume over 30 days. Either way, every dollar in the pools traded about 300 times in a month.

Most of that is stablecoins: the two USDC/USDT pools (Ethereum and BNB) did $3.2B of the $5.0B that went through the 15 largest pools in the last 30 days. But the WETH/USDC pool on Ethereum also turned over $346M on $3.2M of TVL. An AMM can’t usually do that without its LPs paying for it.

Why Uniswap LPs Lose to Arbs

On Uniswap, the pool sets its own price, and it only moves when someone trades. When ETH moves on Binance, the pool is briefly wrong. An arbitrageur buys the cheap side until it matches, and the arb’s profit is the LPs’ loss. Researchers call this LVR (loss-versus-rebalancing).

Concentrated liquidity (Uniswap v3) makes it sharper. A narrow range earns more fees per dollar, but it also puts more of your capital at the exact price arbs trade against. And once the price leaves the range, you stop earning.

So the question for LPs has always been whether swap fees beat what the arbs take. On a lot of volatile pairs, they don’t.

The Fix LPs Couldn’t Use Until Now

A proprietary (prop) AMM flips the design. The price comes from an oracle or a market maker’s own feed, and the pool quotes a bid and ask around it. When ETH moves, the quotes move with it, with no trade needed, so there’s no stale price to pick off.

Prop AMMs already dominate where they run. BisonFi (about 40%) and Tessera (about 20%) lead Solana’s prop AMMs, and FermiSwap handles about 60% of Ethereum’s prop AMM volume. The catch is timing: the price update has to land before the swaps in each block, or the LP gets picked off anyway. That’s cheap on Solana and harder on Ethereum, where gas makes frequent updates expensive.

Until Metric, every one of them ran on its own team’s money, for three reasons:

  • The edge is secret. A fast private feed and per-trade pricing make the model work, and publishing them gives away the edge.

  • Open oracle pools have a known weak spot. If the oracle drifts from the real price, arbs trade against LPs at the wrong price. DODO tried an open oracle pool in 2020 and ran into exactly this.

  • Outside LPs create conflicts. An operator can keep the good flow and leave the toxic flow to LPs. Taking outside money also starts to look like running a fund.

How LPing on Metric Works

Metric splits a prop AMM into three roles:

  • Price providers supply the reference price: Chainlink, Pyth, Metric’s own feed, or a curator’s custom feed.

  • Curators set up the pool: the spread, the price provider and the safeguards. Think of Morpho’s vault curators, but for market making.

  • LPs fund the pool.

Bins that move by themselves. You deposit into bins, which are percentage ranges around the reference price. When the price moves, the bins move with it. A Uniswap v3 range is fixed in price and can go out of range; a Metric bin stays near the action.

LPs earn the spread, not a swap fee. Traders buy at the ask and sell at the bid, and LPs earn the gap around the reference price. Most large pools split it 80% to LPs, 5% to the curator and 15% to Metric. The Moebius-curated pools split it 85/0/15.

Public and private pools. Public pools take outside LPs. Private pools run on the curator’s own capital.

Does it work? In Castle Labs’ 11-day comparison of the WETH/USDC pool (Oct 15–26), measured one minute after each trade, Metric LPs made +0.44 bps per dollar traded, while Uniswap v3 LPs lost 1.01 bps. It’s one pair and nobody has reproduced it yet, but it’s the right sign.

Is It Safe?

  • Audits: five, by Zellic and Sherlock (May–Sep 2026), including a public Sherlock contest in July with up to $150K in rewards. The top payout was $11,936, which suggests no high-severity findings.

  • Bug bounty: up to $500K on Immunefi, live since Nov 10. Critical bugs pay $50K–$500K.

  • Code: pool, factory, swap-math, and oracle contracts are public on GitHub. The license is source-available (BUSL until 2029), not open source.

  • Open issues: Metric publishes a known-issues list, with October reports still pending fixes and November notes on stop-loss limits.

  • Team: mostly anonymous. The CTO is reportedly a former Director of Engineering at Uniswap, and the project is bootstrapped, with no outside funding (Castle Labs).

What Can Go Wrong

  • Inventory risk. The oracle stops you from being picked off at stale prices. It doesn’t stop you ending up holding the wrong asset. If ETH falls for a week, traders keep selling ETH into the pool, and a few bps of spread won’t cover the drop.

  • You’re trusting the feed and the curator. If the reference price is wrong or slow, you trade at the wrong price. A curator using a custom feed can widen spreads with no hard limit.

  • Fees can change without notice. Curator and protocol shares can change with no time lock, even on pools with locked pricing.

  • Safeguards have gaps. The per-bin stop-loss doesn’t protect the first swap into a newly funded bin, and it only checks bins the price crosses. It isn’t a cap on your total loss.

  • New code, anonymous team. The contracts have been live for about a month, with fixes pending.

Here’s what the TVL rankings don’t show. Two of Metric’s largest pools trade ETH with the same curator, the same Pyth feed and the same 80/5/15 split. One has paid LPs at every time horizon, last week and last month. The other lost LPs 36.72% (annualized) at the 30-minute mark last week.

<Explore LPs on Metric>


The Playbook

This isn’t financial advice. These are research notes on a young protocol. Size positions so a total loss wouldn’t hurt you.

Step 1: Read the markout properly

User's avatar

Continue reading this post for free, courtesy of 𝕯𝖆𝖓𝖌𝖊𝖗.

Or purchase a paid subscription.
© 2026 𝕯𝖆𝖓𝖌𝖊𝖗 · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for writing, podcasting, and media founders.