At about 09:50 UTC on 30 October 2026, DefiLlama showed the Uniswap v3 WETH-USDT 0.3% blockchain that runs exactly as written when it is called, holding and moving funds according to its code, without an intermediary able to change the outcome.">smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool on Ethereum paying 642.9% a year in fees. A day earlier the same pool showed 40.8%. Over the two weeks before that it had moved between 14% and 44%. Its deposits had barely changed: 88.5 million dollars the day before, 87.4 million that morning. Learning how to provide liquidity on Uniswap starts with knowing why that figure means so little.
The number was no glitch, and that is the useful part. We added up the pool’s swap events on Ethereum for the same 24 hours and found about 520 million dollars of trades, close to the 513 million DefiLlama reports. Some 87% of it landed in three hours from about 20:30 UTC on 29 October. For the first hour of that burst the pool’s own price ran far from the wider market, touching 3,482 dollars per ether while DefiLlama’s reference price sat near 2,260 dollars. This lesson does not try to explain why. What matters here is that the trades and the fees were real, and the yearly rate was not one anybody earned.
The screen sells a rate. Your result depends on volume you cannot predict and on what the price does to your tokens. The previous lesson, on how aggregators route trades through pools, looked at the trader’s side of a pool like this. This one looks at the other side. By the end you will have picked a pair with a method, computed a real return by hand, run through a position on a testnet, and know how to open, collect and close a small one on a layer 2.
Nothing here is investment advice, and no pool in this lesson is a recommendation. They are examples, dated, because their figures change by the hour.
Adding liquidity means selling both sides of a market
A liquidity provider, or LP, deposits two tokens into a pool, and traders swap against that deposit for a fee. So far it sounds like a savings account. It works more like a shop that must always quote a price: when ether rises, the pool sells your ether on the way up, and when ether falls, it buys more on the way down. The mechanics of why, and the proof of the loss this creates, are in our lesson on liquidity pools and impermanent loss. In one line, wallet. It becomes permanent when the provider withdraws.">impermanent loss is the gap between your position and what the same tokens would be worth had you left them in your wallet, about -2.0% for a 1.5x price move, -5.7% for 2x and -20% for 4x in a full-range 50/50 pool.
Your other counterparty is the arbitrageur. Every time the market moves on a centralised exchange, the pool quotes a stale price until someone trades it back into line, and that someone pockets the difference. Milionis, Moallemi, Roughgarden and Zhang gave this cost a name in a paper first posted in August 2022: loss-versus-rebalancing, or LVR. They split an LP’s return into market exposure and a second part, fees minus what arbitrageurs take. On the Uniswap v2 ETH-USDC pool, they report that over 99.991% of the variance of LP returns came from market exposure. Our lesson on AMM maths shows where that cost comes from. In practice, your result over a few months is mostly the price of ether, and the fees are the smaller line you are trying to win.
Choosing a pair: start from volume, not from the rate
Fees need volume: the volume to TVL ratio
Every fee APR you see comes from one sum: daily volume times the fee tier, times 365, divided by the total value locked. Take the Uniswap v3 USDC-USDT 0.01% pool on Ethereum on 30 October 2026: 130.1 million dollars of volume at 0.01% is 13,010 dollars of fees that day. Divided by 35.1 million dollars of staking or liquidity, measured at current market prices. It is the main size metric in DeFi.">TVL and multiplied by 365, that gives 13.5%, exactly what DefiLlama printed. DefiLlama’s adapter code confirms the method, which it calls 24-hour fees times 365 divided by TVL. No compounding, no impermanent loss, no gas, and no deduction for the protocol’s share of fees where that is switched on.
The ratio of volume to TVL tells you how hard each deposited dollar works. It is also a quick way to compare pools before reading any APR. Here are eight pools as DefiLlama showed them that morning.
| Pool (as of 30 October 2026, about 09:50 UTC) | TVL | 24h volume | Volume/TVL | Fee APR, last day | 30-day mean |
|---|---|---|---|---|---|
| Uniswap v3 WETH-USDT 0.3%, Ethereum | $87.4M | $513.4M | 5.9 | 642.9% | 36.3% |
| Uniswap v3 USDC-WETH 0.05%, Ethereum | $79.0M | $388.2M | 4.9 | 89.7% | 8.9% |
| Uniswap v3 USDC-WETH 0.3%, Ethereum | $54.6M | $176.4M | 3.2 | 353.8% | 22.8% |
| Uniswap v3 WETH-USDC 0.3%, Base | $149.4M | $180.9M | 1.2 | 132.5% | 54.9% |
| Uniswap v3 WETH-USDC 0.05%, Arbitrum | $36.0M | $130.7M | 3.6 | 66.2% | 22.3% |
| Uniswap v3 USDC-USDT 0.01%, Ethereum | $35.1M | $130.1M | 3.7 | 13.5% | 1.3% |
| Uniswap v4 USDT-USDS 0%, Ethereum | $55.5M | $215.7M | 3.9 | 0.85% | 0.35% |
| Uniswap v4 PYUSD-USDS 0%, Ethereum | $100.0M | $54.5M | 0.5 | 0.12% | 0.09% |
The last-day APR beats the 30-day mean on every line, by 1.4 times on the quietest pool and nearly 18 times on the loudest, because that 24-hour window held an unusual burst of trading. The two Uniswap v4 stablecoin pools show that volume alone pays nothing: they are listed at a 0% tier, so they trade heavily and pay LPs close to nothing. Before you trust a rate, divide it by what the pool normally pays. To see the normal level, open the pool’s chart rather than the headline figure.
Stablecoin pairs, correlated pairs and volatile pairs
A pair behaves according to how far its two tokens can drift apart. A volatile pair such as WETH-USDC earns the highest fees because arbitrage and hedging flows are constant, and it carries the full impermanent loss. A correlated pair, such as ether against a liquid staking token, drifts less, so the loss is smaller, as long as the link holds.
Stablecoin pairs are where beginners get the wrong idea. Screens often mark them as having no impermanent loss, and on a quiet day that is close to true. It holds only while both coins keep their peg, so they do have depeg risk. If one stablecoin falls, arbitrageurs sell it into the pool and take the good one out. A narrow position ends up almost entirely in the weaker coin: 5,000 dollars of a USDC-USDT style position would be worth about 4,500 dollars if that coin traded at 0.90 dollars (example figures). Curve and the specialised AMMs are built for this pair type, and they carry the same exposure.
Their yield is less steady than it looks too. That USDC-USDT pool paid between 0.25% and 3% a year on each of the nine days to 29 October, then 13.5% on the 30th. Its 30-day mean was 1.3%. On 5,000 dollars, that is the gap between about 1.85 dollars and 0.18 dollars a day. The busy reading came with a warning too: during the same evening burst, the pool’s price reached 1.25 USDT per USDC while DefiLlama’s reference prices held both coins near 1 dollar, so a pair of stablecoins can leave its usual range on-chain for a time without either coin losing its peg elsewhere.
Fee tier, range width and what out of range means
On Uniswap v3, the developer documentation lists standard tiers of 0.05%, 0.30% and 1%, and governance can add others, which is how the 0.01% tier exists. Uniswap v4 drops fixed tiers: a pool’s creator can set any fee from 0% to 100% in steps of 0.0001%, or use dynamic fees. In practice, calm pairs settle on low tiers and wilder pairs on high ones, so you pick the pool where the volume already is.
The range is a bigger decision. Our lesson on concentrated liquidity on Uniswap v3 and v4 covers ticks and capital efficiency. The practical trade-off is short: a narrow range earns more per dollar while the price stays inside, takes more impermanent loss, and stops earning entirely once the price leaves. At that point your position has been converted into one token and sits idle until the price returns or you move it. A full range earns less and needs almost no attention, which is why it suits a first attempt.
Red flags in the pool list
A large TVL proves very little. On 30 October 2026 the Uniswap v2 WISE-WETH pool held 126.4 million dollars and did 4,221 dollars of volume in 24 hours, for a fee APR of 0.004%. Such a pool looks safe from a distance and pays nothing. At the other end, the v2 PEPE-WETH pool showed 356.5% on a one-day burst of 102 million dollars, against a 30-day mean of 10.9%, with a token that can lose much of its value in a week.
Other warnings need no data. A token you cannot sell freely, a pair created in the last few days, a TVL held by one or two wallets. On v4, a pool may also run a hook, a contract that adds custom logic to it, and a hook you have not read is code you are trusting with your deposit. A later lesson covers due diligence on protocols. For now, stick to pairs you could sell tomorrow on a major exchange.
The real return: fees minus impermanent loss minus gas, then tax
The lines of the account
Write these down before you add a position, and then again when you close it.
- Fees earned, in dollars, at the rate your position really got, which can differ from the pool average.
- Impermanent loss: your position’s value at exit, minus what the same deposit would be worth if you had held it.
- Gas and entry costs: every transaction, plus the swap and bridge needed to get the two tokens in the first place.
- Tax, including the hours spent working out what you owe.
- The price move of the tokens themselves, which the other lines leave out.
That last line trips people up. Impermanent loss compares you with holding. It says nothing about whether holding was any good. If ether halves, an LP who beat holding by 150 dollars has still lost a four-figure sum.
A worked example with round numbers
These figures are an example and not a forecast. You deposit 5,000 dollars into a full-range 50/50 pool, 1.25 ETH and 2,500 USDC, with ether at an example price of 2,000 dollars. You stay 90 days. The pool pays 30% a year on your starting value throughout, a simplification, so fees come to 5,000 times 30% times 90/365, or 369.86 dollars. Two transactions on a layer 2 cost at most about 0.02 dollars.
| ETH price at exit (example) | Value if held | Value in pool | Impermanent loss | Result against holding | Total at exit | Result in dollars |
|---|---|---|---|---|---|---|
| unchanged at 2,000 | 5,000.00 dollars | 5,000.00 dollars | 0.00% | 369.84 dollars | 5,369.84 dollars | 369.84 dollars |
| up 50% to 3,000 | 6,250.00 dollars | 6,123.72 dollars | -2.02% | 243.57 dollars | 6,493.57 dollars | 1,493.57 dollars |
| doubles to 4,000 | 7,500.00 dollars | 7,071.07 dollars | -5.72% | -59.09 dollars | 7,440.91 dollars | 2,440.91 dollars |
| down 25% to 1,500 | 4,375.00 dollars | 4,330.13 dollars | -1.03% | 324.97 dollars | 4,699.97 dollars | -300.03 dollars |
| halves to 1,000 | 3,750.00 dollars | 3,535.53 dollars | -5.72% | 155.38 dollars | 3,905.38 dollars | -1,094.62 dollars |
Read the last two columns separately. When ether doubles, fees fall short of the loss and you end 59.09 dollars behind someone who did nothing, while still sitting on a healthy gain. When ether halves, you beat the holder by 155.38 dollars and are still down by more than a thousand dollars.
The break-even is simple to compute: divide the dollar impermanent loss by the daily fees. Here the daily fees are 4.11 dollars. A doubling costs 428.93 dollars against holding, so the position needs about 104 days of that rate to catch up. A halving costs 214.47 dollars, or about 52 days. Run the same sum with a 10% APR and each of those periods triples, which is why the 30-day mean matters more than the headline.
On a concentrated position, both the fee line and the loss line grow: same table, bigger numbers on each side. For the stablecoin case, gas stops being a footnote on Ethereum if the rate is near 1%: earning 18 cents a day takes a while to cover even a modest transaction.
What the research found, and its limits
The best-known study is “Impermanent Loss in Uniswap v3”, by Loesch, Hindman, Richardson and Welch, posted on 17 November 2021. Across 17 non-stablecoin pools with more than 10 million dollars of TVL, covering about 43% of Uniswap v3 TVL, it counts 199.3 million dollars of fees against 260.1 million of impermanent loss between 5 May and 20 September 2021. In aggregate those LPs ended about 60.8 million dollars worse off than holders, and 49.5% of them had negative returns. Cointelegraph’s report at the time credited Topaz Blue and the Bancor Protocol, a competing AMM project, so the source has an interest.
Those are four and a half months of 2021, on the first version of concentrated liquidity. Do not read them as today’s odds. We know of no recent study of the same quality that says what share of retail LPs make money in 2026, and any number you see on that question deserves a look at its method.
Four traps in displayed APRs
One day, annualised
The opening example is the clearest case. On 30 October 2026 the WETH-USDT 0.3% pool showed 642.9% for one reason: about 1.5 million dollars of fees in a single day, multiplied by 365. Most of the volume behind it came in a three-hour burst, and the previous 14 readings were between 14.3% and 44.2%. Anyone who deposited after seeing the figure would have missed the fees that created it.
The rate also tells you nothing about how that day went for the LPs in the pool. Fees are one line of the account, and the same trades changed the mix of tokens each position held. We did not compute a profit or loss for those positions, and a screen showing 642.9% has not computed one either.
Reward tokens are not fees
Some pools add token rewards on top of fees, the practice known as liquidity mining. It differs from staking in what you take on: a staker usually holds one token, while a liquidity miner holds a pair, with its impermanent loss, and gets paid in a third asset. On 30 October 2026 DefiLlama showed the Aerodrome USDC-AERO pool on Base, with about 6.1 million dollars of TVL, at a 196.7% reward APR and a 30-day mean of 68.6%. That is paid in AERO, whose price can fall as rewards are sold, and programmes can shrink or end. Always split the two columns, and value rewards at what you could sell them for.
Dilution, and the average that is not yours
A rate attracts money, and money lowers the rate. If a pool doing 10 million dollars of daily volume at 0.3% holds 50 million dollars, it pays about 22% a year. Double the TVL and, with the same volume, it pays half. Read the TVL trend alongside the APR: our guide to measuring TVL properly explains what the figure does and does not count.
The second issue is specific to v3 and v4. The headline fee APR is computed on the pool’s total TVL, including liquidity sitting out of range. Your position may earn well above or well below that average, depending on your range. Be careful with DefiLlama’s 7-day field, apyBase7d, too. For Uniswap v3 pools outside Arbitrum, its adapter code does not average the past week. It estimates a week of fees for a 100,000-dollar position in a range of plus or minus 30% around the price, or 0.1% for stablecoin pools, then multiplies by 52. So it cannot be read beside the one-day figure as the same measure over a longer period. On 30 October it showed 15.9% for the USDC-USDT pool, whose daily readings had been under 3% from 24 to 29 October. That is also why this lesson’s table uses the 30-day mean.
Protocol fees take a further slice. Uniswap’s documentation says they are active on all v2 pools and on select v3 pools, at about one sixth of the swap fee on enabled pools at launch. On those pools an LP keeps less than the gross sum above suggests.
APR and APY are not the same thing
APR is a simple rate. APY assumes you reinvest as you go, which on Uniswap v3 and v4 would mean collecting and adding fees by hand, paying for the transactions each time. A later lesson goes through the arithmetic. When a screen shows APY on a concentrated pool, ask who is doing the compounding.
How to provide liquidity on Uniswap step by step
Here is the full sequence. The two passages after it give the detail for the testnet and for the first real position. Button names in the app change often, so treat the labels below as a guide.
- Decide the amount you can afford to lose entirely, and write it down.
- Use a separate wallet for this, not the one that holds your savings, and keep only what the position needs in it.
- Run the whole cycle once on a testnet: approve, add, collect, remove.
- Pick the pair: filter pools on DefiLlama Yields, then open the pool’s chart and read several weeks of volume and fee APR, not one day.
- Compute the break-even from the worked example, with the 30-day mean as your rate.
- Get both tokens on the chosen chain, and note what the bridge and any swap cost you.
- Connect the wallet at app.uniswap.org, open the Pool section and start a new position. Check the pair, fee tier and chain on screen.
- Set the range. Full range is the simple choice for a first try.
- Enter the amounts, approve each token for that amount only, read every wallet prompt, and confirm.
- Find the position in the app and the transaction on the block explorer, and record the date, amounts and cost.
- Check it weekly: is the price in range, what fees are owed, is the pool’s volume holding up.
- Collect the fees when you choose to, remove the liquidity when you are done, and record both for tax.
Step 3 in detail: the testnet run
Test tokens have no value, so this is the one place where mistakes are free. Uniswap’s open-source interface code still defines Sepolia and Unichain Sepolia as supported testnets as of 30 October 2026, and third-party guides describe a testnet mode in the app’s settings. We could not confirm that adding liquidity works there today, so check in the app: switch on testnet mode, choose Sepolia or Unichain Sepolia, and see whether a new position can be created. You will need test ETH from a faucet.
If the option is missing, Uniswap published a v4 test interface on GitHub for creating pools and adding and removing liquidity, and its README describes it as a playground that is not meant for production. Be aware that the repository is archived and has had no update since June 2025, so it may need technical skill to run and may no longer work.
What the testnet teaches is the order of the clicks: what an approval looks like, how a range is set, where collected fees land, how a position closes. What it cannot teach is anything about returns. Test pools may be empty or set at absurd prices, and their APRs mean nothing.
The first real position on a layer 2, with the gas cost
Gas used to rule out small positions, and in October 2026 it does not. These are three real transactions sent to the Uniswap v3 position manager contract on Base that morning, one of each kind you will need, compared with an estimate for Ethereum. We read their cost from the chain and the action from their call data. Each was somebody else’s position, of a size we did not check.
| Operation (30 October 2026, ETH at about $2,228 at 09:46 UTC) | Gas used | Gas price | Cost |
|---|---|---|---|
| Base, collect fees (transaction 0x8555…) | 180,554 | 0.007 gwei | about $0.003 |
| Base, open a position (transaction 0x4f6e…) | 410,921 | 0.0069 gwei | about $0.006 |
| Base, remove liquidity and collect (transaction 0x3046…) | 173,996 | 0.015 gwei | about $0.006 |
| Ethereum, 400,000 gas estimate at that morning’s price | 400,000 | 0.16 gwei | about $0.14 |
Treat this as a reading from one quiet morning. Gas can rise a long way on a busy day. Even so, the lesson for now is that on a layer 2 the fee to open and close a position is rarely what decides the outcome. The costs that do matter come earlier: the bridge fee, the spread on the swap into your two tokens, and the time you spend on records. Our guide to bridges and cross-chain liquidity covers the first.
On amounts, a first position exists to teach you, so size it to be boring. Tens of dollars are enough to see fees accrue on a layer 2, and they keep a mistake cheap. Approve the exact amount and not an unlimited one: our guide to using a DEX safely explains why, along with how to spot a fake front end. Open the app from a bookmark rather than a search result. And remember the risks that no calculation captures: a bug in a contract, a token that collapses, a bridge that fails.
Collecting fees and withdrawing
Fees behave differently depending on the version. On v2, the fee is added to the pool’s reserves automatically, so it grows your share’s value and comes back when you withdraw. On v3 and v4, per Uniswap’s documentation, fees are held apart and must be redeemed by hand. They do not compound, and they sit uncollected until you act.
In the app, the position page shows the fees owed. Collecting takes those fees and leaves your liquidity in place. Removing liquidity takes out the percentage you choose, and the fees come along with it. A v3 position is an NFT held by your wallet, so whoever controls that wallet controls the position. If you sell or send the NFT, the liquidity goes with it.
When you close, you rarely get back what you put in. Expect a different mix of the two tokens, which is the impermanent loss made visible. If the price is outside your range, you get one token only. Swapping back to your starting asset costs another fee and possibly a taxable event, so include that swap in your account. Set a date to review the position when you open it, and write down the conditions that would make you leave.
Tools to watch a position
You need less than the market suggests. The Uniswap app’s position page gives the state of your range and fees owed. DefiLlama Yields gives the pool’s history, though remember that its 30-day mean is for the pool and not your position. A block explorer gives the record of what happened and at what cost.
For analytics, Revert Finance presents itself as analytics and automation for AMM liquidity providers. We did not test it, and any tool that automates a position asks you to trust another contract. The impermanent loss sums need only a spreadsheet, using the formula in the lesson on pools. A simple sheet with deposit date, amounts, prices, fees collected and gas paid will do more for you than a dashboard, because it is also what your tax return will need.
LP tokens and tax: mostly unsettled
No tax authority we checked has published a complete rule on LP tokens. Whether a deposit is a disposal, what a v3 NFT position is, and how to treat fees each remain open to a degree that varies by country. Our dedicated lesson on DeFi taxes comes later. This is general information rather than tax advice, so take your own records to a professional.
| Country | What the authority says (checked 30 October 2026) | What stays open |
|---|---|---|
| United States | The IRS treats digital assets as property. Under Notice 2024-57, brokers need not file Form 1099-DA for liquidity provider transactions until further guidance; rewards or other compensation are not covered by that relief. | We found no IRS rule on how a deposit, an LP token or a withdrawal is taxed. Reporting relief for brokers does not settle what you owe. |
| United Kingdom | HMRC’s manual has a worked example (CRYPTO61680, updated 28 November 2025) of a liquidity pool on a DeFi lending platform: on deposit, each token paid in is a disposal; on withdrawal, the disposal is of the liquidity tokens handed back, and the tokens received are acquisitions. | It is one example in a section on lending and staking. Whether an AMM position matches it depends on the facts. |
| France | The tax authority’s Q&A (17 July 2026): 31.4% flat tax on individuals’ gains, and deferral on exchanges between digital assets without a cash adjustment. | The page does not mention DeFi or pools. Whether a deposit is a deferred exchange is unanswered. |
| Italy | An Agenzia delle Entrate brochure on the 2025 budget law: substitute tax raised from 26% to 33% on crypto-asset gains realised from 1 January 2026, with no exemption threshold. | We found no specific Agenzia delle Entrate guidance on liquidity pools. |
American and British readers will find the general rules in our guides to crypto taxes in the United States and crypto taxes in the United Kingdom. Whatever the rule turns out to be, you can only apply it with records, and a position that was opened, topped up, collected from and closed is at least four dated events, each with two tokens and two prices.
Is providing liquidity on Uniswap worth it?
Nobody can give an honest yes. It can pay when volume stays high relative to TVL, the price does not run far from your entry, and your costs are low. It tends to disappoint when a pair trends hard in one direction, when you chase a one-day rate, or when the position is small and your time is worth anything.
What you can do is replace the question with a calculation. For the pool in front of you, take the 30-day mean instead of the headline, compute the daily fees on your amount, and see how many days it takes to pay for a doubling and for a halving of the volatile token. Then ask whether you would hold that token anyway, because an LP position is largely a position in it.
Before you press the final button, have these on paper: the amount you can lose, the 30-day fee rate, the break-even in days, the total entry cost, and the date of your first review. The next lesson leaves spot pools behind for on-chain perpetuals, where liquidity providers can end up as the counterparty of leveraged traders.
