Source context: BullSpot report from 2026-06-02T06:35:14.318Z (Fresh report: generated this cycle).
The Setup Nobody Talks About
Bitcoin is grinding near $70,267 after sliding roughly 40% from its prior high, ETF outflows hit $2.4B last month, and risk-off sentiment has capital rotating fast. Some of that rotation ends up in DeFi pools where LPs are quietly bleeding on positions they think are earning yield. The yield is real. The losses are also real. Most people never reconcile the two because they never learned how the underlying machine actually works.
This is the piece I wish someone had handed me in 2020 before I LP'd my first $50,000 into a "safe" ETH/USDC pool. It won't make you a better yield farmer by itself, but it will let you read a pool's mechanics the way a mechanic reads an engine — by understanding what each part is doing and where it fails.
What a Liquidity Pool Actually Is
A liquidity pool is a pile of two or more tokens locked in a smart contract that anyone can trade against. That's it. There's no order book, no human market maker sitting at a desk quoting two-sided prices. It's just a bucket of tokens and a formula that decides what you pay when you swap.
The reason these exist is simple: bootstrapping liquidity for a new token is brutally hard on a traditional order book. You need buyers and sellers on both sides. AMMs flip the problem. Instead of finding counterparties for every trade, you find depositors willing to put tokens in the bucket, and the formula handles the pricing.
The first version of this is Uniswap v2, and it's still the cleanest mental model. Two tokens, a 50/50 split by value, and a formula that keeps the pool "balanced" as people trade. Most modern AMMs — Sushi, PancakeSwap, Camelot, most of the v2 forks — run the exact same core logic. Uniswap v3 made it more complicated (in a useful way, which we'll get to), but the foundation is the same.
The Constant Product Formula: x * y = k
This is the formula that runs most of DeFi. It looks like a middle school algebra problem, and that should tell you something: the math is simple. The consequences are not.
Imagine a pool with 100 ETH and 200,000 USDC. The "k" constant is 100 × 200,000 = 20,000,000. That number cannot change during a trade. The contract enforces it.
Now someone wants to buy 10 ETH. After their trade, the ETH side has 90 tokens. To keep k constant, the USDC side must be 20,000,000 / 90 = 222,222 USDC. So the buyer pays 22,222 USDC for 10 ETH. The "spot" price of ETH at the start of the trade was 2,000 USDC (200,000 / 100), so they're paying an average of 2,222 per ETH. Slippage: ~11%.
Pool State Before Trade
├── ETH: 100
├── USDC: 200,000
├── k constant: 20,000,000
└── Spot price: $2,000
Trade: Buy 10 ETH
Pool State After Trade
├── ETH: 90
├── USDC: 222,222
├── k constant: 20,000,000 (unchanged)
└── Effective price paid: $2,222/ETH
That's the entire mechanism. Every swap pushes the price along the curve defined by k. Big trades push it more. Small trades barely move it. The pool never runs out of one side because the curve flattens asymptotically as reserves approach zero — at the cost of astronomical slippage. The trade "fails" long before the pool empties.
This is also why thin pools are dangerous. A $1M swap in the example above would cause roughly 50% slippage. A $1,000 swap costs you maybe a few basis points. Liquidity is the only thing that determines whether the AMM is a useful venue or a tax on traders.
Providing Liquidity: How and Why
To LP, you deposit equivalent value of both tokens into the pool. The contract mints you an LP token — a receipt that proves your share of the pool. You earn a pro-rata cut of every swap fee the pool charges (typically 0.3% on Uniswap v2, 0.05% on v3 for stable pairs, etc.).
The "why" is the part that gets people in trouble. People provide liquidity because they see an APY of 15% or 30% and think of it as yield. It is not yield in the way a Treasury bill is yield. It's a combination of:
- Swap fees collected from other traders
- Sometimes incentive tokens (liquidity mining rewards)
- The drift of the pool's composition as price moves
That third item is where the trap lives. As the price of one token rises, the pool automatically sells some of it for the other (to keep the 50/50 value balance). As the price falls, it buys more. The pool is a continuous, mechanical rebalancer. Whether that's good or bad for you depends on which direction price moves.
Impermanent Loss: The Real Cost
Impermanent loss (IL) is the difference between holding the two tokens and LP'ing them. It's "impermanent" only in the sense that it becomes permanent if you withdraw. Let me show you the math on a real scenario.
The Setup
You deposit 10 ETH and 20,000 USDC into a pool when ETH is at $2,000. Total deposit value: $40,000. You get an LP token representing your share.
The Price Doubles: ETH Goes to $4,000
The pool rebalances. To maintain k with the new price:
- New ETH amount: sqrt(20,000,000 / 4,000) = 70.71 ETH
- New USDC amount: sqrt(20,000,000 × 4,000) = 282,843 USDC
- Total pool value: 70.71 × 4,000 + 282,843 = $565,685
Initial Deposit (Hold)
├── ETH: 10 × $4,000 = $40,000
├── USDC: $20,000
└── Total: $60,000
LP Position (Same Price)
├── ETH: 7.07 × $4,000 = $28,284
├── USDC: $28,284
└── Total: $56,569
Impermanent Loss: $3,431 (~5.7% of hold value)
You made $16,569 instead of $20,000. The pool sold some of your ETH for USDC as ETH rose, so you captured less of the upside. If you'd just held, you'd have $60,000.
The Price Halves: ETH Goes to $1,000
Same mechanism, opposite direction. The pool buys ETH as it falls.
- New ETH: sqrt(20,000,000 / 1,000) = 141.42 ETH
- New USDC: sqrt(20,000,000 × 1,000) = 141,421 USDC
- Total value: 141.42 × 1,000 + 141,421 = $282,843
Hold value: 10 × 1,000 + 20,000 = $30,000. You're down either way, but the LP is down ~5.7% more. The loss compounds: if fees don't outpace the IL plus price decline, you lose money relative to a simple hold.
The General Rule
For a 50/50 pool, IL scales with the price ratio:
Price Change Impermanent Loss
───────────── ─────────────────
1.25x ~0.6%
1.50x ~2.0%
2.00x ~5.7%
3.00x ~13.4%
4.00x ~20.0%
5.00x ~25.5%
The fees you earn need to clear these numbers — and they need to clear them net of the price decline, because IL is calculated against a buy-and-hold counterfactual. Most pools, most of the time, do not clear this bar. That's the uncomfortable truth behind the "passive income" marketing.
LP Tokens and Rewards
Your LP token is just a receipt, but it's composable. You can:
- Stake it in a farm for additional incentive rewards
- Use it as collateral in a lending market
- Lock it to get a vote in governance (veToken model)
- Wrap it into a derivative (e.g., Uniswap v2 LP tokens became one of the foundational collateral types of DeFi summer)
This composability is the engine of DeFi, but it's also where risk multiplies. A depeg in one of the underlying tokens, a smart contract bug in the staking contract, an oracle manipulation in the lending market — the LP token is the surface that all those risks stick to. Most LP exploits in the last few years didn't attack the AMM. They attacked something else holding the LP token.
One heuristic I use: if the APY on a pool is more than 3-4x the swap fees, the difference is almost certainly incentive emissions, and the token paying that emission is the thing most likely to go to zero. Run the numbers assuming the rewards disappear. If the strategy still works, fine. If it doesn't, you're subsidizing the protocol's growth with your principal.
Concentrated Liquidity: Uniswap v3 and Beyond
Uniswap v3 changed the game by letting LPs choose a price range. Instead of spreading capital across the entire price curve from zero to infinity, you put your liquidity where you think trading will actually happen. This is more capital efficient — you can earn 10x to 100x more fees per dollar of capital — but it concentrates risk.
If you LP ETH/USDC between $2,000 and $2,200 and ETH trades in that range, you collect fees proportional to the concentration. If ETH moves to $2,500, your position is now 100% USDC. You've stopped earning fees entirely and you've locked in IL versus the hold. If ETH drops to $1,500, you're 100% ETH and in the same boat.
The math looks like this in concept:
Wide Range (Uniswap v2-style)
├── Capital: 100% deployed
├── Fee earning: low per dollar
└── IL: spread across full price history
Narrow Range (v3 with $2,000-$2,200)
├── Capital: ~10% deployed at current price
├── Fee earning: 10-100x higher per dollar
└── IL: 100% of move outside range is captured instantly
The implication: concentrated liquidity is an active strategy disguised as a passive one. You must monitor the position, rebalance as price moves, and accept that "out of range" is a state where you earn nothing and hold the wrong token. This is not for people who think LP'ing is a set-and-forget yield product. It's a market-making strategy with a sophisticated UI.
In bearish regimes like the current one — BTC grinding near $70K, sentiment at -54, ETF outflows dragging — concentrated LP strategies bleed especially fast. The price chops sideways, then dumps, and your narrow range fills with the wrong token. Most v3 positions that opened in early 2026 are underwater right now for exactly this reason.
What This Means for Traders Who Don't LP
Even if you never provide liquidity, AMMs affect every swap you make. Understanding the curve tells you three things most retail traders miss:
Why thin pools are robbery. A $50,000 swap in a $200,000 pool will cost you 20%+ in slippage. Aggregators like CowSwap, 1inch, and Matcha exist specifically to route around this — but they only help if you use them.
Why large wallets get better execution. When a whale swaps $5M, they split the trade across multiple pools and DEXs to minimize curve impact. You can do the same on a smaller scale. Just because the UI shows "one click" doesn't mean the routing is optimal.
Why "cheap" tokens often aren't. A token with $50K of liquidity and a $5M market cap trades on fumes. The price you see is the price for the first $5,000. The price for the next $50,000 is something else entirely, and it can be 30% worse.
The Takeaway
- AMMs are formula-based, not order-book-based. Every swap moves a curve, not a queue. The bigger the trade, the more the curve bites you.
- Impermanent loss is mechanical, not optional. For a 50/50 pool, a 2x price move costs you ~5.7% versus holding. The fees have to beat that plus the price move itself to make LP'ing worth it.
- Most "yield" is incentive emissions, not fees. Run the numbers without the reward token. If the strategy dies, you were paying for the privilege of being exit liquidity.
- Concentrated liquidity is active, not passive. Uniswap v3 lets you choose your range, but choosing wrong means you earn nothing and hold the losing side. Treat it as market making.
- Use aggregators for any non-trivial swap. Curve impact is real. Splitting routes saves real money. The UI is hiding this cost from you by design.
The pools will keep printing APY numbers that look attractive. The math underneath those numbers doesn't care about the marketing. Read the engine before you fund it.