Openai/6973721c-28b0-8007-a991-09211709d274
This conversation was exported from ChatGPT[1]
gpt-5-2-thinking[edit | edit source]
User: What’s the economic risks of pendle PT looping on morpho or Euler[edit | edit source]
What’s the economic risks of pendle PT looping on morpho or Euler
Also let’s do optimal LTV settings calculation
Assistant: Below is how I’d think about economic not just smart-contract risk when you loop Pendle PT as collateral and borrow on …[edit | edit source]
Below is how I’d think about economic (not just smart-contract) risk when you loop Pendle PT as collateral and borrow on Morpho (Blue) or Euler, and then how to set an “optimal” LTV (really: optimal vs liquidation risk + execution risk + funding risk).
1) Economic risks of Pendle PT looping (Morpho / Euler)[edit | edit source]
A) You’re short funding + long a “zero-coupon bond”[edit | edit source]
A Pendle PT is like a zero-coupon bond: you buy it at a discount and it converges toward redeemable value at maturity. docs.pendle.finance[2] Looping = levering that convergence trade, but you’re also short the borrow rate (variable funding).
Failure mode: borrow APR spikes (utilization shock) while PT implied yield compresses → carry turns negative, you de-lever into bad liquidity.
B) Liquidation is driven by oracle + LLTV/LT, not “fundamentals”[edit | edit source]
Even if PT “should” converge, your position lives/dies by the money market’s collateral valuation rule.
- Morpho Blue: liquidation eligibility is when LTV > LLTV, and health factor is computed directly from collateral value, debt, and LLTV. Morpho Docs[3]
- Euler: liquidation is triggered when risk-adjusted collateral value < debt, and can happen due to price moves or debt growth. Euler Docs[4]
Key economic risk: oracle design (spot vs smoothed vs PT-specific risk oracle). If the oracle can be pulled by secondary-market dislocations, you can be liquidated even though “maturity value” is fine.
C) “Liquidity gap” risk (the silent killer)[edit | edit source]
Liquidators must sell seized PT somewhere (or hedge it). If the PT market is thin / out of range / stressed, liquidation execution can fail or clear at huge discounts → bad debt risk for the market and forced liquidation losses for you. This is explicitly called out in PT risk discussions for money markets. Aave[5]
Looping amplifies this because everyone is crowded on the same unwind path.
D) Maturity + roll risk (cliff events)[edit | edit source]
As maturity approaches, PT volatility typically declines and convergence strengthens, but your loop introduces a “roll decision”:
- close before maturity (market liquidity)
- hold through maturity (redemption mechanics, settlement timing, post-maturity oracle handling, etc.)
If many players roll at once, PT liquidity + borrow liquidity can both gap.
E) Underlying asset + wrapper risk[edit | edit source]
PT inherits risk from:
- Pendle contracts (tokenization + redemption flow)
- the underlying yield-bearing token / protocol (LST/LRT, stable yield token, etc.) A failure in either can reprice PT sharply and instantly. docs.pendle.finance[6]
F) Reflexivity + “loop convexity”[edit | edit source]
Looping makes your liquidation distance nonlinear:
- small adverse move → higher LTV
- higher LTV → less buffer
- any further move → liquidation cascade
This is why “2–3x is fine, beyond that gets hairy” is a common heuristic in PT looping guides. Medium[7]
2) “Optimal LTV” calculation (practical framework)[edit | edit source]
There isn’t one universal optimum because it depends on:
- oracle type + update frequency
- liquidation bonus / close factor / auction mechanics
- PT liquidity + slippage under stress
- your rebalance frequency (do you monitor hourly, daily, weekly?)
- borrow rate volatility (util shock probability)
So the right way is: choose an acceptable liquidation probability over a horizon and back out a safe LTV.
Step 1 — Define variables (per market)[edit | edit source]
- LLTV (Morpho) or LT (Euler equivalent trigger)
- h = risk horizon (e.g., 1 day if you monitor actively; 7 days if you want “sleep well”)
- σ = volatility of collateral value in loan terms (PT priced in borrowed asset)
- r_b = borrow APR (use a stressed value, not today’s)
- LB = liquidation bonus / penalty (how much collateral you lose when liquidated)
- s = stress slippage haircut for unwinds (PT liquidity + DEX depth)
- z = quantile (e.g., 2.33 for 99%, 3.09 for 99.9%)
Step 2 — Compute a conservative “survival buffer”[edit | edit source]
Two things push LTV up over horizon h:
- collateral value can drop
- debt grows via interest
A simple (usable) bound:
- Collateral shock factor:
q=exp(−z⋅σh)q = \exp(-z \cdot \sigma \sqrt{h})q=exp(−z⋅σh)
- Debt growth factor:
g=exp(rb⋅h)g = \exp(r_b \cdot h)g=exp(rb⋅h)
- Execution haircut for liquidation mechanics + slippage (rule of thumb):
η=1−(LB+s)\eta = 1 - (LB + s)η=1−(LB+s) (If LB=7% and you assume 5% slippage in stress, η≈0.88\eta \approx 0.88η≈0.88)
Step 3 — Safe initial LTV[edit | edit source]
For Morpho Blue, you want to stay below LLTV even after shocks:
LTV0,max≈LLTV⋅qg⋅η\textbf{LTV}_{0,\max} \approx LLTV \cdot \frac{q}{g} \cdot \etaLTV0,max≈LLTV⋅gq⋅η This gives you an LTV that has roughly “(1 − tail probability)” chance of staying solvent over horizon h, under your σ/r assumptions, and acknowledges that liquidation is lossy.
3) Translating LTV into “how many loops” (quick)[edit | edit source]
If you loop by borrowing and buying more PT repeatedly, leverage roughly scales like:
Leverage≈11−LTV\text{Leverage} \approx \frac{1}{1-\text{LTV}}Leverage≈1−LTV1 So:
- LTV 0.70 → ~3.33x gross exposure
- LTV 0.80 → 5x
- LTV 0.85 → 6.67x This is why tiny LTV changes massively change liquidation fragility.
4) What I need from you to compute your optimal LTV numbers (no fluff)[edit | edit source]
Send:
- Which PT (e.g., PT-sUSDe-YYYYMMDD) and chain
- Borrow asset (USDC? DAI? ETH?)
- Market + parameters (Morpho: LLTV + oracle type; Euler: LT / risk params if shown)
- Your desired monitoring horizon: 1d / 3d / 7d
- Your stress assumptions (or I’ll apply defaults): - σ (daily vol of PT/borrow) - stressed borrow APR (e.g., 2× current or p95)
If you don’t want to pull vol data, I can still give you a parameterized LTV curve (LTV as a function of σ and horizon) so you can plug numbers in instantly.
Default “founder-safe” settings (if you want an immediate starting point)[edit | edit source]
Without your specifics, a conservative institutional heuristic is:
- target liquidation probability ≤ 0.1% over 7d
- assume stressed borrow APR = 2× current
- assume stress slippage 5–10% on PT
- then set LTV 10–20 points below LLTV/LT for looping (bigger buffer if oracle is spot-like)
This aligns with how PT-as-collateral risk is discussed: liquidation execution + oracle behavior dominate outcomes, not the “it converges at maturity” story. Aave[8]
If you reply with the exact PT + market, I’ll compute:
- recommended LTV band (aggressive/base/conservative)
- max loops / target leverage
- break-even borrow APR vs PT implied yield
- liquidation distance under 1d/7d shocks
gpt-5-2[edit | edit source]
User: Thanks[edit | edit source]
Thanks
Assistant: Anytime 🙂[edit | edit source]
Anytime 🙂 If/when you want, we can turn this into a one-pager risk memo (Morpho vs Euler side-by-side, LTV bands, stress table) or run exact numbers for a specific PT. Just say the word.
References[edit | edit source]
- ↑ "Economic Risks". ChatGPT. Retrieved 2026-01-23.
- ↑ "docs.pendle.finance". docs.pendle.finance. Retrieved 2026-01-23.
- ↑ "Morpho Docs". Morpho Docs. Retrieved 2026-01-23.
- ↑ "Euler Docs". Euler Docs. Retrieved 2026-01-23.
- ↑ "Aave". governance.aave.com. Retrieved 2026-01-23.
- ↑ "docs.pendle.finance". docs.pendle.finance. Retrieved 2026-01-23.
- ↑ "Medium". medium.com. Retrieved 2026-01-23.
- ↑ "Aave". governance.aave.com. Retrieved 2026-01-23.