Pendle AMM: Time-Decaying Asset Pricing

Standard AMMs were built for volatile token pairs where both assets have uncertain future values. But yield-bearing assets like stETH are fundamentally different - the Principal Token (PT) is mathematically guaranteed to be worth exactly 1.0 at maturity. No AMM built for volatile pairs can account for this structural convergence. Pendle built a custom AMM where the pricing curve automatically shifts as time passes, concentrating liquidity around the implied yield rate, and guaranteeing PT converges to par at expiry.

The Problem with Standard AMMs for Yield Assets

Uniswap's xy=k invariant was designed for pairs like ETH/USDC - two assets where future prices are unknown and independent. But yield assets are different: one asset (PT) has a mathematically deterministic future value (par = 1.0 at maturity). The other asset (YT) has a mathematically deterministic future value (0.0 at maturity). A standard AMM can't encode these constraints.

Standard AMM (xy=k)
  • IL computed from price divergence alone
  • No concept of time-to-maturity
  • PT could theoretically trade anywhere
  • No guarantee of PT->1.0 at maturity
  • Liquidity spread evenly across all prices
  • No oracle integration for implied yield
OK Pendle AMM (Yield-Space)
  • IL naturally approaches zero at maturity
  • Curve parameter t encodes time-to-maturity
  • PT mathematically bounded 0.9-1.0 range
  • Guaranteed convergence to par at expiry
  • Concentrates liquidity around implied yield
  • Oracle-fed yield rate for fair pricing
Concrete Example: 180-Day stETH PT

On Uniswap, a PT-stETH/SY pool could trade PT at 0.85 (implying 30%+ APY) or 0.99 (implying 2% APY) with no mechanism to correct mispricing other than arbitrageurs. On Pendle's AMM, the curve is constructed so the maximum PT price for a 5% yield asset at 180 days is 0.976 - arbitrageurs automatically correct any deviation beyond this bound. The AMM is self-correcting by construction.

Yield-Space Invariant: The Math Behind Pendle AMM

Pendle's AMM is built on a yield-space invariant that replaces the constant-product formula with a time-aware alternative. The key insight: the curve exponent changes based on how close we are to maturity.

x(1?t) + y(1?t) = k
Where t = daysRemaining / totalDuration ? [0, 1]
x = PT Reserve
The amount of Principal Tokens in the liquidity pool. PT behaves like a zero-coupon bond - no yield, just par at maturity.
y = SY Reserve
SY (Standardized Yield) is the underlying yield-bearing asset (e.g., stETH). Its value accrues into YT over time.
t = Time Factor
t = 1 at deployment (fresh pool), t -> 0 as maturity approaches. The curve literally changes shape over time.
k = Invariant
Constant across trades. When you swap, x and y change but k stays fixed (similar to how xy=k in Uniswap).
Connection to Existing AMMs
t = 1 (fresh)
Standard AMM
x + y ? k when exponent->0
? constant-product behavior
t = 0.5 (mid)
Hybrid Curve
Exponent = 0.5
Moderately curved
t -> 0 (near expiry)
StableSwap
x + y = k
Flat, near-zero slippage

Key insight: The same formula with different t values spans the entire spectrum from Uniswap-style constant-product to Curve-style stableswap. Pendle smoothly interpolates between these regimes as time passes - no migration needed, no new pool required.

Curve Behavior at Different Maturities

The same PT/YT pair looks completely different at 365 days vs 7 days. Watch how the curve flattens as maturity approaches - this is the AMM adapting to the reduced uncertainty of the payoff.

PT Price (current)
0.9756
YT Price (current)
0.0244
Time to Maturity
180 days
Curve Type
Moderate curve
Curve shape interpretation: When the curve is steep (far from maturity), PT prices are very sensitive to trades - like a volatile asset. When the curve is flat (near maturity), PT trades like cash - near-zero slippage, minimal price impact. The shaded region shows all four maturities simultaneously, with the current selection highlighted in teal.

Liquidity Concentration Around Implied Yield

Unlike Uniswap V3 where LPs choose their own price ranges, Pendle AMM concentrates liquidity around the oracle-derived implied yield rate. The AMM "knows" what the fair PT/YT price should be and positions LP depth accordingly - automatically.

How Oracle-Fed Yield Works
Step 1: Oracle Reports Yield

Chainlink or similar oracle reports the current staking/APY rate of the underlying asset. If stETH is earning 4.5% and there are 180 days left, the implied PT price is 1/(1+0.045180/365) ? 0.978.

Step 2: AMM Calibrates Curve

The AMM sets its internal parameter such that the PT price on the yield-space curve matches the oracle-derived price. Any deviation creates arbitrage opportunity.

Step 3: Liquidity Auto-Concentrates

LPs provide liquidity to the pool. The AMM routes more capital to the price range around the oracle-derived fair value. Most trading activity happens here.

Step 4: Rebalance on Yield Change

If the oracle yield changes (e.g., staking rate drops from 4.5% to 3.5%), the AMM updates its implied yield and liquidity shifts. LPs don't need to manually rebalance.

LP Position Anatomy
Concentration
Oracle-derived
vs Uniswap V3's manual ranges
Time Sensitivity
Auto-decaying
Curve flattens as t->0
IL Profile
Near-zero at expiry
No equivalent in standard AMMs
*
Fee Efficiency
Higher near expiry
More trades, less IL risk
Uniswap V3 vs Pendle AMM Liquidity

Uniswap V3 LPs must actively manage their range orders - if ETH price moves outside your range, you earn nothing and your capital sits idle. In Pendle AMM, the implied yield oracle means liquidity is always positioned correctly relative to the fair PT/YT price. The AMM handles rebalancing automatically as yield rates change, though LPs can still override with manual parameters.

PT and YT Price Discovery

PT and YT get their prices through AMM trading, not through centralized order books or market makers. The price mechanism is elegant: PT price = the discount to par that the market assigns, and YT price = the present value of expected future yield.

PT Price Discovery

PT trades at a discount to par. The size of that discount IS the fixed yield you're locking in.

PT = 1 / (1 + impliedYield t)

Example: 5% implied yield, 180 days -> PT = 1/(1+0.05180/365) = 0.976. This means you're buying at 0.976 to lock in ~5% fixed yield.

YT Price Discovery

YT price reflects the market's expectation of future yield. Expensive YT = high expected yield.

YT = 1 ? PT = impliedYield t

Same example: YT = 1 ? 0.976 = 0.024. This means the market is pricing in ~5% APY for 180 days. If actual yield rises to 7%, YT price would increase.

The SY Connection

SY (Standardized Yield) is Pendle's wrapper for the underlying asset. When you deposit stETH, it becomes SY-stETH. The AMM actually trades PT/SY and YT/SY pairs, not PT/YT directly. The relationship is always preserved: PT + YT = SY (at any moment in time).

1 PT-stETH + 1 YT-stETH = 1 SY-stETH = underlying value

This invariant is enforced mathematically, not by an oracle. If PT + YT ever diverges from SY value, arbitrageurs immediately profit by splitting/merging until the relationship is restored. This creates a perpetual price floor and ceiling for PT + YT.

? AMM vs Order Book: When Does Pendle AMM Win?

For PT and YT specifically, the AMM model has structural advantages over traditional order books. However, there are specific scenarios where order books could compete.

Criteria Pendle AMM Order Book
Capital Efficiency OK Better - auto-concentrates around fair price X Spread thin across all prices
Time-to-maturity handling OK Built-in - curve auto-adjusts X Manual adjustment required
Large block trades ~ OK - slippage increases for $10M+ OK Better for institutional size
Price discovery depth ~ Good for retail, thin for large OK Superior for complex structures
Impermanent loss risk OK Near-zero near maturity N/A - not an LP model
Gas costs (per swap) OK Low - single contract call X High - matching engine overhead
Best Use Cases for Each
Use Pendle AMM when:
  • Standard PT or YT trades (retail size)
  • You want automatic LP rebalancing
  • You prefer known IL bounds near maturity
  • You want to earn fees without active management
  • You need fast, cheap small-to-medium trades
Consider Order Book when:
  • Institutional-size block trades ($10M+)
  • Complex yield curve structures (spreads, straddles)
  • Need for anonymity in large positions
  • Custom swap guarantees needed
  • Trading less-liquid early-stage markets

Swap Fees and Gas Efficiency

Pendle's fee structure is maturity-tiered - shorter-duration positions pay higher fees to compensate LPs for the increased time-sensitivity of those positions. This is the opposite of what you'd expect from standard AMMs, but it makes mathematical sense.

0.02%
Long-duration PT (>180d)
Low volatility, stable pricing
0.05%
Medium PT (60-180d)
Moderate curvature
0.10%
Short PT (30-60d)
Increasing time sensitivity
0.30%
Very short (<30d)
High gamma, options-like
* Gas Comparison: Pendle vs Uniswap
~150K gas
Pendle PT swap
~180K gas
Uniswap V3 swap
~210K gas
Uniswap V2 swap

Pendle's single-hop AMM is more gas-efficient than Uniswap V3 for comparable swap sizes because V3 requires more complex tick management. For very large trades requiring multi-hop routes, gas savings compound.

Why YT Trades Cost More Than PT

YT has higher effective volatility than PT. A 1% change in yield rate causes a much larger percentage change in YT price than in PT price (especially for long-duration positions). To compensate LPs for this gamma exposure, YT fee tiers are 2-3 higher than PT fee tiers at the same maturity. This is analogous to how options AMMs charge higher fees for higher-volatility assets.

Impermanent Loss: A Different Kind

Impermanent loss in Pendle AMM is fundamentally different from standard AMM IL. The primary driver isn't price divergence - it's time. As maturity approaches, the PT/YT ratio converges to a known endpoint, which paradoxically makes LP positions safer over time.

Entry PT Price
0.9756
Current PT Price
0.9756
Days Elapsed
0
Cumulative IL
~0% (near expiry)
? Early Period IL (Days 365-180)

Highest IL risk. YT is expensive (lots of yield to capture), so if yields fall, YT price drops significantly. LP's main risk is yield rate decline, not price divergence. This period resembles high-volatility options positions.

OK Safe Zone (Days <60)

IL approaches zero. PT is very close to 1.0, YT has little time value left. The curve is almost flat - price sensitivity is minimal. This is the safest period to add LP positions with minimal IL risk.

IL vs Standard AMM: Key Differences
Standard AMM IL
  • Always negative (LP underperforms HODL)
  • Driven by price ratio change
  • Worst at extremes (depegs, crashes)
  • Never recovers (unless price reverts)
Pendle AMM IL
  • Can be positive in yield-rising environments
  • Driven by yield rate changes + time
  • Worst early, recovers toward expiry
  • Guaranteed convergence to zero at maturity

? Integration with Uniswap V4 Hooks

Uniswap V4 introduced hook contracts - modular plugins that execute at specific points in the swap lifecycle. A Pendle-style AMM is one of the most natural applications for V4 hooks, because the time-decay logic can be implemented as a hook without modifying the core AMM invariant.

? How a Pendle Hook Would Work
Before Swap (Hook Point)
  • Read current block timestamp
  • Calculate elapsed time since pool creation
  • Derive t = (totalDuration ? elapsed) / totalDuration
  • Adjust swap output using Pendle formula
newPrice = f(oldPrice, t, dx)
After Swap (Hook Point)
  • Verify PT->par convergence isn't violated
  • Update oracle-derived implied yield
  • Emit events for off-chain tracking
  • Settle any fee adjustments
enforce(PT_final ? 1.0 + epsilon)
Benefits of V4 Hook Integration
  • Inherits Uniswap's massive liquidity network
  • No need to build separate AMM infrastructure
  • Time-decay is a plugin, not a core modification
  • Access to V4's live liquidity across all chains
  • Hook can concentrate liquidity via oracle feed
? Challenges to Address
  • V4 hooks must be deployed as separate contracts
  • Gas overhead of hook calls vs native AMM
  • Oracle dependency (vs pure on-chain pricing)
  • Security surface increases with hook complexity
  • yt/liquidation mechanics need separate handling
The Vision: Self-Decaying LP Positions

With a Pendle hook in Uniswap V4, LPs could deposit into any standard V3 pool and the hook would automatically apply time-decay pricing based on the pool's deployment time. A USDC/USDCe pool deployed 180 days ago would automatically have its liquidity rebalanced toward the "maturity price" - effectively behaving like a self-managed Pendle position within a standard V3 pool. This could revolutionize how yield-bearing assets are LP'd across DeFi, bringing Pendle's time-aware pricing to every Uniswap pool.

PT/YT Investment Simulator

Configure your investment parameters and see how PT and YT prices evolve, what your total value at maturity will be, and how time decay impacts your position.

Position Breakdown
PT You'll Receive
976.32 PT
YT You'll Receive
23.68 YT
PT Price (at buy)
0.9763
YT Price (at buy)
0.0237
Total at Maturity
$1,000.00
Fixed Yield Earned
$24.28 (2.43%)
Annualized APY
4.93%
Time Decay Impact
Moderate
How it works: You invest $1,000 buying PT at a discount. At maturity, PT redeems for $1,000 (par). The difference is your fixed yield. YT gives you yield-bearing exposure that decays to zero at maturity - its price reflects the expected future yield. Adjust the sliders to see how maturity length and yield rate change your position.
Why does standard AMM impermanent loss get worse for time-decaying assets?
In a standard AMM like Uniswap V2, impermanent loss (IL) comes from price divergence between two assets. But for PT/YT, the 'price' of each token is structurally moving toward 1.0 and 0.0 respectively as maturity approaches - purely due to time, not market prices. If you LP PT-stETH/SY on a standard AMM, you'd face IL not just from market price movements but also from the automatic convergence of PT toward par. Standard AMMs have no mechanism to account for this time-driven convergence, making them fundamentally wrong for yield assets. Pendle's AMM mathematically guarantees PT->1.0 at maturity, so IL near expiry approaches zero - something a standard AMM can never achieve.
What happens to LP positions as maturity approaches in Pendle AMM?
LP positions in Pendle AMM become increasingly 'safe' as maturity approaches, in a way that has no equivalent in standard AMMs. Near maturity (say <30 days), the curve is almost flat - similar to Curve's stableswap near peg. This means: (1) IL approaches zero because PT is very close to 1.0 and there's little room for price to diverge, (2) slippage for trades becomes extremely low, which attracts more trading volume, (3) the AMM behaves like you're holding a 1:1 redeemable position. LPs who entered early (when YT was expensive and PT was cheap) earn the time decay of YT in the form of rebalancing. LPs near maturity earn near-zero IL but collect higher trading fees from the concentrated activity.
How does Pendle AMM handle the YT (Yield Token) price discovery?
YT's price is discovered entirely through AMM trading, unlike traditional finance where yield forwards are priced by market makers. The YT price reflects the market's implied future yield rate. If the underlying asset (e.g., stETH) is earning 5% APY and YT is trading at a price implying 6% expected yield, the market expects yields to rise. Conversely, if YT implies 4%, markets expect yields to fall. This makes YT a pure sentiment instrument on future yield rates. The AMM automatically updates YT's effective 'cost of carry' as time passes - early in the period, YT price is high (much yield to capture); late in the period, YT price is low (little yield left). At expiry, YT is worth exactly the yield generated that period.
What is the actual?? formula for Pendle's yield-space invariant?
The core invariant is: x^(1?t) + y^(1?t) = k, where x = PT reserve, y = SY (underlying) reserve, t = time-to-maturity factor (0 at expiry, 1 at launch). More precisely, t = daysRemaining / totalDuration. The exponent (1?t) is the critical parameter: when t=1 (just deployed), exponent=0 and the curve is constant-product (x + y = k, effectively). When t->0 (approaching expiry), exponent->1 and the curve flattens to constant-sum (x + y = k, near-zero slippage). The invariant k is recalculated after every swap. For standard assets with very long duration (e.g., 1 year), the early curve is closer to standard AMM behavior. For short duration (e.g., 7 days), the curve flattens almost immediately.
How does Pendle's oracle-based implied yield work for liquidity concentration?
Pendle uses a Chainlink or similar oracle to determine the current yield rate of the underlying asset (e.g., stETH staking rate). This oracle price feeds into the AMM's liquidity concentration mechanism. Rather than concentrating liquidity around a spot price (like Uniswap V3), Pendle concentrates around the oracle-derived implied yield. This means the AMM 'knows' where the fair PT/YT price should be based on current market conditions. If the oracle reports 5% APY and there are 180 days to maturity, the AMM calibrates the curve so PT ? 1/(1+0.05180/365) ? 0.976. Liquidity providers automatically position around this implied rate. If the oracle yield changes, the AMM rebalances. This is fundamentally different from Uniswap V3 where LPs must manually update their range orders.
Can you explain the connection between Pendle AMM and Uniswap V4 hooks?
Uniswap V4 allows custom 'hooks' - smart contracts that execute at specific points in a swap lifecycle (before/after swap, before/after LP, etc.). A Pendle-style hook would implement time-decaying curve logic: at each block, the hook recalculates the exponent (1?t) based on actual time elapsed, shifting the curve automatically. This is powerful because V4 hooks can: (1) Automatically decay the AMM curve as time passes without requiring LP rebalancing, (2) Enforce PT->par convergence at maturity by hard-coding the expiration logic, (3) Concentrate liquidity around oracle-derived yields via the hook's initialization parameters. The hook pattern is ideal because it keeps the core AMM invariant simple while adding Pendle's time dimension as a plugin. Pendle has publicly discussed implementing its AMM as a V4 hook.
Why is gas cheaper for PT trades compared to YT trades on Pendle AMM?
The fee structure in Pendle AMM is maturity-tiered: shorter maturity = higher fees (because the time-decay curve is more volatile and requires more computation to maintain). Specifically: (1) PT trades near maturity have near-zero price sensitivity - a given trade size moves the price less, so the AMM can charge lower fees while still covering gas, (2) YT trades early in the period are more price-sensitive - the YT component has high gamma (options-like sensitivity to yield changes), so fees are higher to compensate LPs for this volatility, (3) Gas for PT settlement is also lower because PT->SY redemption at maturity is a simple 1:1 transfer, while YT settlement requires tracking and distributing yield that accrued over the period. In practice, PT trades at 0.05% fee tier for long-duration positions but 0.3%+ for very short (<30 day) positions.
How does impermanent loss differ between early LP entry and late LP entry in Pendle?
Entry timing matters enormously in Pendle AMM because the IL profile is inverted compared to standard AMMs. Early entry (e.g., 300+ days before maturity): IL can be substantial in early periods because YT is expensive (high yield to capture), and if yields drop, YT price falls significantly. However, early LPs earn more from YT price appreciation if yields stay high or rise. Near-maturity entry (<60 days): IL is minimal because PT is already close to 1.0 and YT has little time value left. The curve is almost flat, so LP value doesn't fluctuate much. This makes Pendle AMM uniquely accessible for cautious LPs who want to earn fees near zero IL. The break-even point typically occurs around the halfway point of the duration - before that, IL risk is elevated; after that, you're in 'safe zone' territory similar to a stableswap LP.