Liquidity provision & concentrated liquidity.

What liquidity provision actually is

Every decentralised exchange (DEX) needs someone to be on the other side of your trade. There's no traditional order book full of human market-makers — instead there's a pool: a shared pot of two tokens that traders swap against. When you provide liquidity, you deposit both tokens into that pot. In return you earn a share of the trading fees from everyone who swaps — typically a small percentage (say 0.05%–0.30%) of each trade, split among all the liquidity providers in proportion to how much they supplied. You are, in effect, renting out your tokens to the market and collecting toll fees.

A pool is always a pair: cbBTC/USDC, ETH/USDC, and so on. To provide liquidity you generally supply both sides. That's the first thing to internalise — you're not depositing one asset and earning yield on it like a savings account; you're depositing two, and your returns and risks both flow from the relationship between their prices.

The constant-product idea, in plain terms

The classic DEX engine is the automated market maker (AMM). Instead of matching buyers to sellers, it prices trades with a formula. The original is the constant-product rule: multiply the amount of token A in the pool by the amount of token B, and keep that product constant through every trade. If a pool holds 10 cbBTC and 1,000,000 USDC, the product is 10,000,000. Someone who wants to buy cbBTC adds USDC and removes cbBTC — but the product must stay at 10,000,000, so as cbBTC leaves the pool its price rises along a curve. The more you buy at once, the worse your price gets. That price impact is called slippage, and it's the mechanism that lets a pool quote a price without any human deciding it.

You don't need the algebra to be a good liquidity provider. The one intuition that matters: the pool automatically sells whichever token is going up and buys whichever is going down. It always rebalances toward a 50/50 value split. That single fact is the seed of both your fee income and the risk we'll cover below.

Concentrated liquidity — choosing a range

Early AMMs spread your liquidity across every possible price, from zero to infinity. That's safe but wasteful: almost all real trading happens within a narrow band around the current price, so most of your capital sat idle, earning nothing. Concentrated liquidity (the Uniswap V3 design, now standard) fixed this. Instead of full-range, you choose a price range — say "cbBTC between $95,000 and $115,000" — and all your capital is packed into just that band.

The payoff is capital efficiency. By concentrating, the same dollars provide far deeper liquidity inside your chosen band, so you capture a much larger slice of the fees from trades in that band — often many times what a full-range position would earn from the same capital. A tight range around the current price can earn dramatically more in fees than a wide one.

The catch — and it's the whole game — is that you only earn fees while the market price is inside your range. If cbBTC runs to $120,000, your $95k–$115k position is now entirely on one side of the market, fully converted to the asset that's now cheaper to the pool, and earning nothing until price comes back or you move the range. Narrower ranges earn more per dollar but fall out of range faster. Wider ranges earn less per dollar but stay active through bigger swings. That trade-off is the central decision in concentrated LP, and it's exactly the decision BABA's LP Agile pillar automates.

Impermanent loss — the cost nobody quotes you

Here is the part the headline APR never mentions. Because the pool always sells your rising token and buys your falling one, an LP position ends up worth less than if you'd simply held the two tokens in your wallet. That shortfall is called impermanent loss (IL). "Impermanent" because if price returns exactly to where you started, the gap closes; but the moment you withdraw at a different price, the loss becomes permanent.

Walk through a concrete example. Suppose you provide a full-range cbBTC/USDC position: 1 cbBTC at $100,000 plus $100,000 USDC — $200,000 total. Now cbBTC rises 50%, to $150,000.

• If you had just held: 1 cbBTC (now $150,000) + $100,000 USDC = $250,000.
• In the pool: the AMM sold some of your cbBTC into USDC on the way up. The math of the constant-product curve leaves you holding roughly 0.816 cbBTC (≈ $122,500) and about $122,500 USDC — a total near $244,900.

The pool position is worth about $5,100 less than holding — roughly 2% of the position — even though both went up in dollar terms. That gap is the impermanent loss for a 50% move. It is not a fee, not a hack, not a mistake; it's the structural cost of letting the pool rebalance you. The bigger the price move in either direction, the larger the IL: a 2x move costs about 5.7%, a 4x move about 20%.

One more nuance concentrated LP adds: by packing capital into a band, you also concentrate the impermanent loss. You earn more fees while in range, but you feel IL faster as price moves through your band. The right range is the one where the extra fees comfortably beat the extra IL — which depends on how volatile the pair is and how much volume it does.

Reading APR honestly

A pool's advertised APR usually looks generous because it adds two things together: trading fees (real, earned from volume) and token emissions (incentive rewards the protocol prints to attract liquidity). The honest formula for what you actually keep is:

real yield ≈ trading fees + token emissions − impermanent loss − gas

Headline APR almost always quotes only the first two terms and silently drops the last two. Three traps to watch. First, emissions can be paid in a token whose price is falling — a 60% emissions APR is worth half that if the reward token halves while you hold it. Second, APR is annualised from a recent snapshot; a number computed during one frantic high-volume day can be ten times the sustainable rate. Third, IL and gas are real subtractions that the dashboard rarely shows. Always mentally discount a headline APR until you've accounted for what's missing.

The two BABA LP venues

BABA's LP Agile pillar runs on two concentrated-liquidity DEXs:

• Aerodrome on Base — Base is Coinbase's L2, and Aerodrome is its leading liquidity venue. Here you earn trading fees plus AERO emissions on pairs like cbBTC/USDC. The emissions are meaningful, which is why Aerodrome often shows the higher headline APR — and exactly why you must discount that headline for the AERO price and the IL.
• prjx on HyperEVM — the second venue, a concentrated-liquidity DEX on HyperEVM (the EVM layer paired with Hyperliquid). It keeps part of the LP pillar inside the same ecosystem as the MD trading pillar, so capital and attention don't have to scatter across unrelated chains.

You don't have to run both. Aerodrome is the natural first venue because Base gas is cheap and the cbBTC/USDC pair is liquid and easy to reason about. prjx is the option for when you want a second venue closer to the Hyperliquid side of the engine.

When LP beats holding — and when it doesn't

LP is at its best in pairs that trade a lot but don't trend violently — high volume relative to volatility. That's where fees pile up while IL stays small. It also wins when you genuinely want exposure to both tokens anyway; the fees are then a bonus on assets you'd have held regardless.

LP loses to simply holding when you're strongly directional. If you're convinced cbBTC is about to double, don't LP it — the pool will dutifully sell your cbBTC the whole way up and hand you the IL. Hold it outright, or trade it on the perp side. The pool rewards you for being neutral and punishes you for being right about a big one-way move. Know which one you are before you deposit.

Rebalancing when price drifts out of range

Because a concentrated position only earns inside its band, your one ongoing job is keeping price in the band. When the market drifts toward an edge — or punches through it — the position stops earning and sits fully converted to one token. Rebalancing means closing that position and reopening a fresh range centred on the new price so fees start flowing again.

Rebalancing isn't free: it costs gas, and crucially it crystallises the impermanent loss — the moment you withdraw at the new price, that "impermanent" gap becomes permanent. So you don't rebalance on every wiggle. You rebalance when the expected future fees from the new range outweigh the gas plus the locked-in IL. Getting that timing right by hand is fiddly and easy to overtrade, which is precisely the chore the engine takes off your plate.

How the LP Agile pillar runs it for you

This is the slowest pillar in the BABA engine, and deliberately so — it's not chasing a move, it's collecting a stream. The LP Agile pillar does the two tedious jobs for you: it produces auto range suggestions (a width tuned to the pair's recent volatility and fee economics, so you're neither earning nothing out of range nor rebalancing every hour) and it runs a daily fee harvest so your earnings are swept and compounded rather than left sitting in the position. You stay in control of the actual deposit and rebalance; the engine removes the guesswork and the babysitting. Treat LP as the patient income floor underneath the more active pillars — small, steady, and compounding while you sleep.