Open a market on Pendle and two numbers sit side by side. On 31 October 2026 the sUSDe market expiring on 28 January 2027 showed a fixed yield of 4.51 percent next to an underlying yield of 5.08 percent. Both are true, and they answer different questions. The first is a price you can lock today. The second is what the token has paid over the past seven days, and nobody knows whether it will still pay that in January.
Pendle Finance exists to put a market between those two numbers. It takes a token that earns a floating yield and cuts it into a claim on the money and a claim on the income until a set date, then lets people trade each piece. A cautious holder sells the unknown. A risk-taker buys it for a few cents on the dollar. The price they agree on is the most useful number in the protocol, and most of this lesson is about reading it.
What is Pendle Finance, in one minute
Pendle is an on-chain market for trading yield, and anyone can create a market on it. You deposit a yield-bearing token such as staking means depositing a proof of stake coin with a protocol that stakes it for you and issues a transferable token (such as Lido stETH or rETH) representing the deposit plus rewards, which you can use or sell while the coins stay staked.">wstETH or sUSDe. Pendle wraps it, then splits it into a Principal Token (PT), redeemable at a fixed maturity date, and a Yield Token (YT), which collects all the yield and points until that date. Buying PT at a discount gives a fixed yield. Buying YT is a bet that the yield will beat the price paid.
The docs compare the split to bond stripping in traditional finance: PT is the zero-coupon bond, YT the detached coupons. That one comparison explains most of what follows.
Two reminders before the mechanics. A yield-bearing token either grows in number or grows in value against a base asset, as our guide to rebasing and interest-bearing tokens explains; liquid staking tokens and yield-bearing stablecoins are the raw material of nearly every Pendle market. And a fixed rate can be reached other ways: fixed-rate lending designs match a lender with a borrower for a term. Pendle does something else. It sells the yield itself, and it needs no borrower.
This lesson continues the previous one on on-chain options and vaults. There, a premium paid for volatility was the price of an unknown. Here the unknown is a rate.
Splitting a yield-bearing token: SY, PT and YT
SY, the wrapper you rarely see
Yield tokens behave differently from one another, so Pendle first puts them behind one interface. That wrapper is SY, for Standardized Yield. The interface was proposed as ERC-5115, which is still a draft on the Ethereum standards site. In most cases 1 SY stands for 1 unit of the underlying, though the docs name exceptions such as aUSDC and mPENDLE. SY has no expiry, and the SY contract is what receives the yield and points of the deposited asset. The app handles the wrapping, so most users never touch it.
Minting, recombining and the one equation that matters
The chain for staked ether runs ETH, stETH, SY-stETH, then PT-stETH plus YT-stETH. Before maturity, one PT and one YT can be put back together to get the SY out. That link holds the two prices in place:
Price of PT + price of YT = price of the underlying. The AMM documentation states it in those terms, and every other formula on Pendle follows from it. When YT gets dearer, PT gets cheaper, and the reverse.
| Token | What it is | At maturity | Who wants it |
|---|---|---|---|
| SY | Standard wrapper around the yield-bearing token | Nothing changes, it has no expiry | Nobody directly, it is plumbing |
| PT | Claim on the principal, sold at a discount | Redeems 1 for 1 against the accounting asset | Holders who want a known result on a known date |
| YT | Claim on all yield and points until maturity | Worth zero, unclaimed yield can still be collected | Traders betting on higher yield or farming points |
| LP | Share of a PT/SY pool | Holds the value of the underlying, rewards claimable | Holders who accept pool risk for fees and incentives |
The detail that catches people: the accounting asset
PT does not always redeem for one unit of the token you see in its name. It redeems 1 for 1 against the accounting asset, the one shown in brackets in the market name. In a market written PT-ezETH (ETH), 1 PT pays 1 ETH worth of ezETH, which is less than 1 ezETH, because ezETH has grown against ETH. So a “fixed yield” is fixed in that asset, and only in that asset. A PT on a staked ether token fixes a yield in ether. It says nothing about the dollar price of ether.
What the YT holder really gets
One YT earns what one unit of the underlying earns, until maturity, and the yield can be claimed at any time before or after that date. It arrives as SY, which can be unwrapped. The same goes for points: 1 YT earns the points of 1 unit of the underlying. Pendle keeps a cut. Its fees page says it collects 5 percent of all the yield accrued by YT, points included, as of 31 October 2026. Because points are tracked off-chain, the partner protocol deducts that share when it allocates them. Fee settings can change, so check the page before trading.
PT: a fixed yield you can price yourself
A zero-coupon bond with a pull to par
The V2 whitepaper by Vu Nguyen, dated 14 October 2022 calls PTs analogous to zero-coupon bonds. A zero-coupon bond pays no interest along the way. You buy it below face value, and the gap is your return. PT works the same. Since its yield has been sold off, it trades at a discount to its accounting asset, then climbs toward parity as the date nears, because less and less yield remains to be separated from it.
Pendle’s glossary gives the formula: implied APY = (1 + YT price / PT price)^(365 / days to expiry) – 1. Because PT and YT add up to 1 in the accounting asset, it can be turned around into something easier to use: PT price = (1 + implied APY)^(-days / 365).
A worked example at 8 percent
Example with invented numbers: a market with one year to run, priced at an implied 8 percent. PT costs 1 / 1.08, or 0.9259. YT costs the rest, 0.0741. Spend 925.93 on 1,000 PT, hold them for the year and redeem them for 1,000 of the accounting asset. The gain is 74.07, or 8 percent on the money spent. Gas and the swap fee are left out here. If the implied rate stays at 8 percent, the path to parity looks like this:
| Days to maturity | PT price | YT price | Sum |
|---|---|---|---|
| 365 | 0.9259 | 0.0741 | 1.0000 |
| 180 | 0.9628 | 0.0372 | 1.0000 |
| 90 | 0.9812 | 0.0188 | 1.0000 |
| 30 | 0.9937 | 0.0063 | 1.0000 |
| 0 | 1.0000 | 0.0000 | 1.0000 |
That table is the whole idea of a fixed yield. The return does not come from an interest payment, since a PT holder receives none. It comes from buying at 0.9259 something that will be worth 1 on a known day.
Fixed until the underlying breaks
The glossary uses the word “guaranteed” for the fixed APY. Read that with care, because the same docs explain when it fails. Pendle records the highest exchange rate ever reached between the token and its accounting asset, called the Watermark Rate. If the rate falls below that mark and stays there, PT redeems for less than its full value at maturity, and YT stops earning until the rate recovers. The docs use PT-RLP as their example: normally 1 USDC at maturity, less than 1 USDC if the watermark is broken. A watermark can also be inflated by an oracle error or a brief spike, which leaves a market looking like it has negative yield after the true rate has returned to normal.
So the fixed yield is a promise about the split, made by a contract, resting on a token issued by somebody else. The rate is fixed. Solvency of the issuer is not.
YT: a leveraged bet on yield and points
What a YT costs and what it pays
Keep the same example. A YT at 0.0741 pays the yield of one full unit of the underlying for a year. Each unit of money spent therefore controls the yield on about 13.5 units. The docs call this leveraged exposure, and the result swings accordingly. The table measures everything in units of the underlying token, because YT pays in SY, and takes off the 5 percent fee. It assumes 1 SY is 1 token and that on day one a token is worth exactly 1 of its accounting asset. If the token then grows by g against that asset, a YT collects g / (1 + g) of a token over the period: at 12 percent, 0.12 / 1.12 = 0.1071, or 0.1018 after the fee. Gas and slippage are ignored.
| Realised yield over the year (example) | YT income per unit after the fee | Result on a 0.0741 YT |
|---|---|---|
| 12% | 0.1018 | +37% |
| 8.46% (breakeven) | 0.0741 | 0% |
| 8% (same as implied) | 0.0704 | -5% |
| 4% | 0.0365 | -51% |
| 0% | 0 | -100% |
Two lines deserve a second look. If the yield comes in exactly at the implied rate, the buyer still loses 5 percent, which is the fee. The real breakeven in this example is about 8.46 percent, slightly above what the screen shows. And a yield that merely halves costs half the stake, with no market crash required.
Why tokens and not the accounting asset? Because the alternative to buying YT is holding the token, and the token grows as well. Counted in the accounting asset, the 12 percent line reads 0.1018 x 1.12 = 0.114 received for 0.0741 paid, or +54 percent, which flatters the trade: a plain holder made +12 percent over the same year without doing anything. Measured against that holder, the gain is the +37 percent in the table.
Why YT decays even when nothing happens
Each day that passes is a day of yield that the YT can no longer earn, so its price falls toward zero. The docs give a rule of thumb: if the implied rate stays constant, a YT with 15 days left is worth roughly half what it was worth with 30 days left. That is not a loss in itself, because the holder has collected the yield of those days. It is a loss only if what was collected is less than the price paid. On maturity day YT is worth zero, and nothing needs doing at that point except claiming what has accrued.
A live check shows the sizes involved. On the sUSDe market expiring on 28 January 2027, with 89 days left and an implied rate of 4.51 percent on 31 October 2026, PT works out at about 0.9893 and YT at about 0.0107. If the 7-day yield of 5.08 percent held for all 89 days, a YT would return about 0.0114 after the fee, a gain near 7 percent over the period. That “if” is the entire trade. This is our arithmetic, not a forecast, and Pendle’s own Long Yield APY figure makes the same assumption of a constant yield.
Points: paying for a share of a future airdrop
Points are the other reason people buy YT. Since 1 YT earns the points of 1 unit of the underlying at a fraction of its price, the docs describe it as a popular instrument for farming points. The market data on 31 October 2026 show how central this has become: of the largest markets by liquidity, those on siUSD, apyUSD, reUSD, sUSDx and strUSD all list a points programme. The sUSDe and wstETH markets in the table below list none.
The risk is specific. Points have no price until a project sets one, if it ever does. A YT bought for points is a bet on the size and date of a future token distribution, and if the distribution disappoints, there is no yield large enough to rescue the position. Also, partner protocols apply the 5 percent fee on points off-chain, so the balance you see depends on their records.
Implied APY, underlying APY and the curve across maturities
Reading a market page
Three numbers do most of the work. Underlying APY is a 7-day moving average of what the token has earned, so it looks backward. Implied APY looks forward: it is the market’s price for the yield until maturity, and it is the same number as the fixed APY of a PT buyer. Long Yield APY is an annualised estimate of what a YT buyer earns if the underlying yield stays where it is. It can be negative.
One point in Pendle’s own course surprises newcomers: the underlying APY does not feed into the implied APY directly. Implied APY moves on supply and demand for PT and YT. Buying YT or selling PT pushes it up. Buying PT or selling YT pushes it down. The course compares it to Uniswap: one is a marketplace of spot prices, the other of implied yields.
Here is what that market looked like on the day of writing:
| Market (chain) | Expiry | Implied APY | Underlying APY | Liquidity |
|---|---|---|---|---|
| sUSDe (Monad) | 28 Jan 2027 | 4.51% | 5.08% | $20.1M |
| sUSDe (Ethereum) | 26 Nov 2026 | 4.74% | 5.07% | $5.1M |
| siUSD (Ethereum) | 7 Jan 2027 | 9.36% | 6.66% | $18.8M |
| apyUSD (Ethereum) | 5 Nov 2026 | 17.34% | 13.92% | $17.6M |
| reUSD (Ethereum) | 10 Dec 2026 | 12.45% | 7.08% | $13.2M |
| sUSDS (Ethereum) | 26 Nov 2026 | 5.72% | 3.80% | $3.7M |
| wstETH (Ethereum) | 30 Dec 2027 | 2.31% | 4.25% | $4.7M |
Source: Pendle markets API, snapshot of 31 October 2026, about 10:00 UTC. The figures move constantly and the app may differ.
When implied sits below underlying, as on sUSDe and wstETH, the market is pricing lower yield ahead, or PT demand has pushed the rate down. A wstETH implied rate of 2.31 percent against 4.25 percent earned, for a term of 14 months, is a large gap on a small 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. The API reported a range of 3 to 7.2 percent for that market, so the rate sat below the band where its liquidity is concentrated, and it should be read with care. When implied sits above underlying, the reading is harder. On reUSD, a fixed 12.45 percent against 7.08 percent earned can mean that PT buyers want paying for the risk of the token, or that YT buyers are bidding up the yield for the points attached. The data cannot tell you which. What it does not mean is free money: a high fixed APY is a price, and someone set it for a reason.
The implied-rate curve: what exists and what does not
Bond readers will want a yield curve, and the docs do say that one asset can have several maturities, each with its own market and implied yield. Plot those and you have a curve. We found no curve page in the docs, though, so what follows is our own reading of its market list.
The data set a limit. On 31 October 2026, 85 markets were active, and most assets had one liquid maturity. The best example is sUSDe, at 4.74 percent for 26 days and 4.51 percent for 89 days. That slopes slightly downward, both points under the 7-day yield of about 5.07 percent, which suggests the market expected sUSDe yield to ease. Even this comparison has a flaw, as the two pools sit on different chains. Two points also make a line, not much of a curve.
Other multi-maturity assets look richer than they are. apyUSD shows 17.34 percent for 5 November in a pool of $17.6 million and 17.62 percent for 11 February in a pool of $0.14 million. sUSDai on Arbitrum shows 10.33 percent for February ($4.3 million) and 9.01 percent for June ($0.02 million). The apxUSD markets run from about 15 to 24 percent across pools of under $1.3 million. A single small trade can move such rates, so they say little about expectations. Before reading any slope, read the liquidity column.
When a curve is real, the logic is the same as for government bonds. Upward means the market expects higher yield later or wants paying for the wait. Downward means it expects yield to fall. The difference between two maturities also gives a rough forward rate, the yield the market implies for the period between them. On the two sUSDe points, that forward rate for the 63 days between 26 November and 28 January works out at about 4.4 percent, our arithmetic. The calculation is only as good as the thinner of the two pools.
The AMM that holds it together
Pendle does not run a generic pool. The docs say its AMM was adapted from Notional Finance’s AMM, and pools hold PT against SY, with no YT inside. YT trades go through the same pool by flash swap. To buy YT, the contract borrows SY from the pool, mints PT and YT, hands the buyer the YT and sells the PT back to repay the pool. Selling YT runs the same steps in reverse.
Three design choices make this a rate market. First, liquidity is concentrated in a pre-set range of implied APY, not of price. Outside that range the pool has little depth, so check the range shown for a market. On 31 October 2026 the Ethereum sUSDe market reported a range of 3 to 11 percent. Second, the curve shifts over time and narrows the price range of PT as maturity nears, which the docs call dynamic curve tightening: less time left means less uncertainty about yield. Third, the fee follows the yield traded, not the principal.
The fee formula is short: trading fee = (fee tier / 365) x days to maturity. With an example tier of 0.3 percent, a swap with a year to run costs 0.3 percent, and one with 30 days left costs about 0.025 percent. The tier is set per market and shown in its specs.
Liquidity providers earn swap fees, PENDLE incentives, part of the PT fixed yield and part of the underlying yield. The docs say that impermanent loss from time decay is zero at maturity, since PT converges on the underlying, and that an LP who stays to maturity ends up equivalent to holding the underlying. Before maturity, the position can still lag if the implied rate moves sharply, and it carries everything PT carries.
Hedging rates: fix, short or go long a yield
Fixing a floating yield
Example with invented numbers. A treasury holds $1,000,000 in a yield-bearing stablecoin earning a floating 6 percent. It fears 3 percent next year. A one-year market prices the implied rate at 5 percent, which puts YT at about 0.0476. The treasury splits its tokens and sells the YT for about $47,600 today, then keeps the PT to redeem 1,000,000 at maturity. Its result is now set, before gas and fees.
| Yield over the year (example) | Income unhedged | Paid for the YT, today |
|---|---|---|
| 3% | $30,000 | about $47,600 |
| 6% | $60,000 | about $47,600 |
| 10% | $100,000 | about $47,600 |
The money from the YT arrives on day one, so it can be put to work. Spent on more PT at about 0.9524, the $47,600 grows to about $50,000 at maturity, which is the 5 percent the market offered. The treasury gave up the upside in exchange for certainty about the rate. A simpler way to reach the same place is to swap directly into PT, as the docs describe. Note the limit of the hedge: if the stablecoin itself fails, the treasury holds PT on a broken token, and the yield hedge does nothing for that.
Shorting yield, and what a PT buyer is actually doing
Pendle’s course treats buying PT as a short-yield position. When the market expects lower yield, implied APY falls and PT rises. A holder who bought PT can sell before maturity and take that gain early. The reverse also applies, and it is what the fixed-income crowd calls duration risk. Take the 8 percent example. Six months later, with about 180 days left, a PT is worth 0.9628 if the implied rate has held. If the implied rate has moved to 12 percent, it is worth about 0.9456. Held to maturity the buyer still gets the promised 8 percent, but selling at that moment would give about 2.1 percent instead of 4 percent.
Going long yield as a hedge
YT can also protect someone who pays a floating rate. A borrower of stablecoins fears that rates will rise. If a market exists on the matching lending token, buying YT pays more as rates climb, which offsets part of the higher interest. Example only, and an imperfect one: borrowing rates and supply rates move together but not one for one, and the YT is worth zero at maturity whichever way things go. Think of it as an insurance premium that is spent over time.
LP positions and PT as collateral
Providing liquidity with the default zap also leans short yield, because part of the deposit buys PT. Pendle offers a Keep YT mode for those who want to stay neutral. Separately, the docs list Morpho, Silo and Euler among money markets that take PT as collateral. That permits loops: buy PT, borrow against it, buy more. Such a position stacks the fixed yield of PT against a floating borrowing cost, and it adds liquidation risk to everything else in this lesson.
Boros: the same idea applied to funding rates
Pendle also runs Boros, which its developer docs describe as an on-chain interest rate swap platform on Arbitrum. Its subject is the funding rate of perpetual futures, explained in our guide to on-chain perpetuals and derivatives. It combines an order book with an AMM and trades on margin, so it is a separate product with separate risks, liquidation among them.
What can go wrong with PT, YT and LP positions
The underlying breaks
Pendle inherits every failure of the assets it lists. November 2025 gave the clearest case. Stream Finance halted withdrawals on 4 November after an external fund manager disclosed a $93 million loss, and its staked xUSD token traded below 30 cents on the dollar, according to The Defiant. Stream had borrowed Elixir’s deUSD to support xUSD, and two days later Elixir shut down minting and redemption and said it would sunset deUSD, The Block reported.
Pendle had a pool on staked deUSD. Elixir named Pendle LPs among those affected when it announced USDC compensation, ChainCatcher reported. Foresight News then reported on 13 November 2025 that Pendle had published a plan for holders of PT, YT, LP and SY in the sdeUSD pool, paying USDC by their dollar value at a snapshot block. That detail rests on this one press report: we could not find Pendle’s own post, or how much holders recovered. PT holders on that market had bought a fixed yield. They ended up with a claims process.
Liquidity before maturity
The fixed yield holds only if you can wait. Sell early and you take the market price, plus whatever slippage the pool imposes. Many pools are small: of the 85 active markets on 31 October 2026, 55 had under $1 million of liquidity and 26 had under $100,000. A position that looks safe to maturity can be expensive to leave, particularly when the underlying is under stress and everybody wants out at once.
YT going to zero
This is the design, not an accident. A YT buyer whose average yield falls short of the price paid loses money, and with yield at zero loses all of it. There is no recovery after maturity, because the token has no further claim.
Smart contract and integrator risk
On 3 September 2024 Penpie, a separate yield protocol built on top of Pendle, lost about $27 million. The attacker created a fake Pendle market, which Penpie’s contracts accepted as real, then used a reentrancy attack on its reward-harvesting function. Pendle said its own funds were secure and paused its contracts, Rekt reports. The DefiLlama hacks list has an entry for Penpie and none under Pendle’s own name as of 31 October 2026.
Permissionless market creation is the structural point in that story. Anyone can create a market on-chain, while the official app lists only reviewed ones, so a third-party contract that trusts any market is exposed. Pendle says its code has been audited, and its FAQ also states that it is not responsible for losses from exploits in third-party contracts. Most newer SY contracts are upgradable proxies as well, which means someone holds the power to change them.
Reading the fixed APY wrongly
The commonest error needs no hack. A fixed 4 percent on a PT in ether terms can sit alongside a 40 percent fall in the dollar price of ether. A fixed rate on a stablecoin assumes that stablecoin keeps its peg. And an APY quoted on a 26-day market is an annualised figure: over 26 days, 4.74 percent a year is about 0.33 percent of your money, before gas. On a small position, gas can take a large share of that.
The market itself has shrunk
Size is a risk measure too, because thin markets are harder to exit. DefiLlama’s Pendle page shows how much total value locked has fallen, about 92 percent from the peak. We do not explain the fall here, since the data show the size of the move and not its cause.
| Date | Pendle TVL (DefiLlama) |
|---|---|
| 19 September 2025 (peak of the series) | $13.39 billion |
| 1 November 2025 | $6.95 billion |
| 29 January 2026 | $3.41 billion |
| 1 May 2026 | $1.50 billion |
| 3 August 2026 | $1.20 billion |
| 2 October 2026 | $1.27 billion |
| 31 October 2026 | $1.02 billion |
Of the 31 October figure, Ethereum holds about $761 million, Monad $148 million and Arbitrum $61 million. Fees on Pendle V2 came to about $0.4 million over 30 days, and about $67.2 million since launch, per DefiLlama’s fee overview on the same date.
Who gets paid: fees, buybacks and the move from vePENDLE to sPENDLE
Pendle Finance has two fee sources: the 5 percent taken on YT yield, and swap fees. It also collects the yield and points of matured PT left unredeemed, so leaving a position idle after maturity costs you that income. Of swap fees, 20 percent goes to liquidity providers. The rest, plus all YT fees, is divided into 80 percent for PENDLE buybacks, 10 percent for the treasury and 10 percent for operations.
The token side changed in 2026. sPENDLE replaced the old model of multi-year locks: staking went live on 20 January 2026 and new vePENDLE locks were paused on 29 January 2026, as The Block reported. PENDLE stakes 1 for 1 into sPENDLE and can be withdrawn after 14 days, or at once for a 5 percent fee. The docs say up to 100 percent of repurchased PENDLE goes to active sPENDLE holders, paid out as sPENDLE every two weeks, and a holder counts as inactive only by failing to vote while a Pendle proposal is open.
Old vePENDLE holders received a non-transferable virtual balance, with a multiplier of up to 4x for two years left on the lock, decaying to 1x at the unlock date. The tokenomics page now describes the vePENDLE contract as winding down. Team and investor tokens were fully vested by September 2024. None of this changes how PT and YT work. It only decides who receives the fees.
Taxes and rules: mostly unanswered questions
No regulator or tax authority we could find has addressed PT and YT specifically, so treat what follows as a list of questions and not as answers. We found no statement on yield tokenization from the SEC, the CFTC or the FCA.
In the United States, the IRS says digital assets are property, and trading one for another is a reportable transaction, whether or not it produces a gain. Buying PT with USDC looks like such a trade. Whether wrapping into SY or splitting into PT and YT is a disposal has no published answer, and neither does the character of the PT discount at redemption or of the YT income. See our guide to US crypto taxes.
In the UK, the HMRC Cryptoassets Manual has a section on DeFi lending and staking, updated on 28 November 2025. Its contents do not name yield tokenization. For gains, the test it applies to loans and staking is whether beneficial ownership of the tokens passes to the borrower or platform, which turns on the terms of the deal. How that test applies to wrapping or splitting a token is open. Our UK crypto tax guide covers the general rules. Keep records of every step, because any later analysis will need the date and value of each one, and take advice on a large position.
A careful first look at a Pendle market
Nothing here is a recommendation to use Pendle Finance. If you only want to understand it, the docs and the market list are the place to start. We found no documented public testnet, so a first real position means real money and real gas: work out the gas cost first, and keep the amount small enough that the fee does not swallow the yield. Before any trade, these checks take a few minutes:
- Read the market name to the end, because the asset in brackets is what PT redeems against.
- Note the maturity date and count the days, then turn the APY into a plain percentage for that period.
- Compare implied and underlying APY, and ask who is on the other side and why.
- Check the liquidity and the implied range, since a thin pool means a poor exit.
- Open the Watermark Rate view and compare it with the current exchange rate.
- Find out who issues the underlying, what backs it and how you would get out of it.
- Set a reminder for the maturity date, because unredeemed positions stop earning and their yield goes to Pendle.
The docs keep a roll-over guide for moving a matured position into a new market. Redeeming PT after maturity carries no protocol fee, only gas, and the closest maturity on the sUSDe table is 26 November 2026.
