pendle-pt-research

# Pendle PT Research

Research Pendle PT markets with a decision-first lens.

Author credit: @Moshu

This skill covers:
- **unlevered PT research** — buy PT, hold to par, or rotate near expiry
- **looped PT research** — deposit PT into a money market, borrow against it, buy more PT, and recurse when the spread and risk budget justify it
- **practical execution comparison** — compare raw PT APY against easier execution paths such as manually confirmed Contango-supported routes

Do not optimize only for the highest displayed APY. The job is to sort PT markets into investable buckets that answer questions like:
- which PTs are attractive to buy and hold until par
- which near-expiry PTs may be good for fast rotations
- which PTs are attractive to **loop for leveraged fixed yield**
- which PTs are best on paper but still manual-only
- which slightly lower-yielding PTs may still be better because execution is easier
- which markets have enough liquidity to enter and exit cleanly
- which underlying assets introduce too much risk even if APY looks good

When Peter asks specifically about **manual Contango-supported PT routes**, read `references/manual-contango-comparison.md` before answering.

## Core framework

Always evaluate Pendle PT markets across these dimensions first:

1. **Time to par**
2. **APY / implied yield**
3. **Underlying risk**
4. **Liquidity / market quality**
5. **Loopability / money-market support** when leverage is relevant

Do not rank purely by APY.

A PT with lower APY but cleaner underlying risk and better liquidity can be a better trade than a high-APY PT with ugly exit conditions or fragile collateral.

For looped PT strategies, a market with slightly lower raw PT APY can still be superior if it has:
- strong money-market support
- a borrowable major stable like USDC
- favorable max LTV
- healthy borrow liquidity
- materially lower borrow cost than PT APY

## Three strategy buckets

### 1. Hold-to-par bucket

Use for markets with roughly **1-4 months to expiry** when the yield is attractive and the underlying risk is acceptable.

Focus on:
- annualized PT discount / implied APY
- confidence in the underlying asset / yield source
- enough liquidity for exit if needed
- whether holding to maturity is simpler than actively managing in and out

This bucket is usually best when:
- maturity is not too far away
- the underlying is understandable and reasonably trusted
- liquidity is healthy enough
- the yield is good without needing heroic timing

### 2. Near-par rotation bucket

Use for PTs that will reach par soon but still show unusually attractive APY.

Focus on:
- very short time to maturity
- whether the annualized yield is still meaningfully elevated
- ability to enter and exit repeatedly without getting chewed up by slippage / fees
- whether market depth is good enough for repeated rotations

This bucket is usually best when:
- time to maturity is short
- APY remains unusually high for that short duration
- liquidity is strong
- the market can be treated as a short-duration yield parking spot

### 3. PT looping bucket

Use for PTs that can be deposited into a money market and borrowed against to create **leveraged fixed-yield exposure**.

Focus on:
- raw PT implied APY
- supported money markets such as Morpho, Euler, Aave, Dolomite, or other Pendle integrations
- borrowable assets, especially major stables like USDC
- borrow APR at realistic size
- max LTV and a practical safety buffer below max LTV
- available borrow liquidity
- Pendle Pencosystem `Max Looping APY` when visible
- whether execution is available manually or through automators such as Contango

This bucket is usually best when:
- the PT already has attractive raw yield
- PT collateral support exists on a credible money market
- borrow cost is comfortably below PT APY
- liquidity is deep enough for both PT entry and borrow execution
- the leverage can be run conservatively with a real liquidation buffer

## Ranking workflow

### Step 1: Build the candidate list

Start from Pendle PT markets page:
- https://app.pendle.finance/trade/markets

For each candidate PT market, capture at least:
- market name
- chain
- expiry date
- days to expiry / time to par
- displayed APY or implied yield
- fixed-yield direction (PT)
- underlying asset / protocol / vault
- market liquidity / TVL / depth if visible
- volume if visible
- any incentive / points context if relevant

If needed, also read Pendle docs or market details pages to understand the underlying source of yield.

### Step 1B: Build the loopability map

For PT looping research, inspect the data sources in this order:
1. protocol-verified sources such as Morpho GraphQL or Euler indexer GraphQL
2. Pendle Pencosystem partner data
3. manually confirmed Contango-supported routes when Peter provides them
4. heuristic inference only as a last resort

For each loopable candidate PT, capture at least:
- supported money market(s)
- borrowable asset(s)
- current borrow APR when known
- max LTV when known
- available liquidity to borrow when known
- Pendle `Max Looping APY` when shown
- whether the route is `Morpho verified`, `Euler verified`, `Contango manual`, or `heuristic only`

Treat this loopability map as a first-class input, not an afterthought.

### Step 2: Classify the underlying risk

Assign an explicit risk tier.

Suggested tiers:
- **Low** — major stable / major LST / highly recognizable underlying with relatively simple structure
- **Medium** — decent protocol quality but more smart-contract, strategy, bridge, or asset-specific risk
- **High** — fragile, obscure, highly reflexive, leveraged, low-trust, or structurally complex underlying

When rating risk, consider:
- what generates the yield
- whether principal ultimately depends on another protocol / vault / bridge
- smart contract complexity
- asset volatility
- depeg risk
- whether the underlying is easy to explain in one sentence

If the underlying is hard to explain clearly, penalize it.

## Liquidity / market-quality lens

Do not treat all PT APY equally.

Check for:
- total liquidity / TVL
- recent trading volume
- likely slippage for realistic position sizes
- ease of entering and exiting without moving the market too much
- whether the market looks alive or stale

Prefer markets where Peter can move in and out without drama.

Penalize:
- thin liquidity
- stale / low-volume markets
- markets that are only attractive on paper for tiny size

For looped PT research, apply the same discipline to the financing leg:
- borrow liquidity must be deep enough for realistic size
- the money market integration must look maintained and usable
- max LTV is not the operating target; leave a real buffer below it
- a high displayed `Max Looping APY` is not enough if available liquidity is tiny or borrow cost is unstable

## Suggested output format

Use bullet lists on Discord. Avoid markdown tables there.

For each PT market, include:
- market
- chain
- expiry
- time to par
- natural PT APY
- underlying asset / source
- risk tier
- liquidity notes
- who it suits: `hold-to-par`, `near-par rotation`, `pt-looping`, `multiple`, or `avoid`
- short thesis
- key risk / invalidation
- final rank or score

For each loopable PT market, also include:
- money market(s)
- borrowable asset(s)
- borrow APR when known
- max LTV when known
- available borrow liquidity when known
- Pendle `Max Looping APY` if visible
- estimated net loop spread or loop attractiveness
- route status such as `Morpho verified`, `Euler verified`, `Contango manual`, or `heuristic only`
- practical leverage note such as `light`, `moderate`, or `aggressive only`

When comparing practical loop candidates, separate the output into:
- **Best natural PT yield**
- **Best practical PT loops**
- **Best Contango-supported manual set**
- **Best paper APY but manual-only**
- **Avoid / low-conviction markets**

## Scoring guidance

Use a weighted judgment rather than pretending the numbers are exact.

For unlevered PT ranking, a good default weighting:
- 30% time to par fit
- 30% APY attractiveness
- 25% underlying risk quality
- 15% liquidity / exit quality

For PT looping ranking, use a different weighting:
- 25% raw PT APY attractiveness
- 25% borrow spread quality and borrow APR
- 20% max LTV / leverage capacity
- 15% PT liquidity + borrow liquidity
- 15% underlying / protocol / integration risk

Adjust judgment when needed. For example:
- near-expiry rotation candidates may deserve more weight on liquidity
- longer hold-to-par candidates may deserve more weight on underlying risk
- loop candidates deserve heavier penalties for thin borrow liquidity, fragile integrations, or tiny practical size

## Decision heuristics

### Prefer for hold-to-par
- 30-120 days to maturity
- attractive fixed yield
- understandable underlying
- healthy liquidity
- clean thesis with limited babysitting required

### Prefer for near-par rotation
- short duration to maturity
- still elevated APY despite near expiry
- strong liquidity / active market
- easy to recycle capital repeatedly

### Prefer for PT looping
- PT APY comfortably above borrow APR
- support on a credible money market
- borrowable major stable like USDC
- healthy max LTV with room to run conservatively below the cap
- enough PT liquidity and borrow liquidity for realistic size
- visible `Max Looping APY` or a clearly positive manually estimated spread
- operational path that is simple enough to repeat manually or via Contango

### Penalize hard
- high APY driven by shaky or obscure underlying risk
- thin liquidity
- market too small for realistic sizing
- hard-to-explain yield source
- poor exit quality
- thin or unstable borrow liquidity
- loop economics that only work at paper max LTV and fall apart with a sensible safety buffer
- integrations that exist in theory but are hard to access or unsupported operationally

## Use the reference file

For detailed ranking dimensions and a reusable research template, read:
- `references/ranking-framework.md`

## Phase 2 scripts

These scripts now exist for the reusable workflow:
- `scripts/scan-markets.py` — pull and normalize Pendle PT markets across supported chains
- `scripts/rank-markets.py` — score and bucket markets into hold-to-par vs near-par rotation, with asset-family and stable-subtype filters
- `scripts/report-markets.py` — print a compact research report
- `scripts/generate-brief.py` — generate a markdown research brief

## PT loop scanner extension

Use these scripts inside this skill:
- `scripts/scan-pencosystem.py` — fetch the live Pendle Pencosystem partner directory
- `scripts/scan-morpho-pt-markets.py` — verify live Morpho PT-collateral markets, loan assets, borrow APY, utilization, and LLTV via GraphQL
- `scripts/scan-euler-pt-markets.py` — verify Euler PT token coverage and Euler vault matches for the target PT set via the Euler indexer GraphQL
- `scripts/scan-contango-pt-markets.py` — verify exact PT route support from Contango's public Pendle external integration config
- `scripts/merge-loop-data.py` — combine ranked PT markets with live Pencosystem partner data and protocol-verified market support when available
- `scripts/rank-looped-markets.py` — score PT looping candidates using PT APY, borrow economics, leverage capacity, liquidity, and execution practicality
- `scripts/report-looped-markets.py` — print a compact loop-focused report with money markets, borrow assets, and leverage notes

Suggested outputs:
- `data/pencosystem.latest.json`
- `data/loop-venues.latest.json`
- `data/loop-ranked.latest.json`

Typical flow:

```bash
cd scripts
python3 scan-markets.py --active-only
python3 rank-markets.py
python3 report-markets.py --top 10
```

PT looping flow:

```bash
cd scripts
python3 scan-markets.py --active-only
python3 rank-markets.py --stable-only --chains ethereum base arbitrum --min-days 7 --max-days 120 --min-liquidity 1000000
python3 scan-pencosystem.py
python3 scan-morpho-pt-markets.py --stable-only --min-apy 0.08
python3 scan-euler-pt-markets.py --stable-only --min-apy 0.08
python3 scan-contango-pt-markets.py --stable-only --min-apy 0.08
python3 merge-loop-data.py --stable-only --min-apy 0.08
python3 rank-looped-markets.py
python3 report-looped-markets.py --top 10
```

Filtered example for a more practical first pass:

```bash
cd scripts
python3 scan-markets.py --active-only
python3 rank-markets.py --chains ethereum base arbitrum --min-days 7 --max-days 120 --min-liquidity 1000000
python3 report-markets.py --top 10
```

Stable-only example (good default for Peter's likely next strategy work):

```bash
cd scripts
python3 scan-markets.py --active-only
python3 rank-markets.py --stable-only --chains ethereum base arbitrum --min-days 7 --max-days 120 --min-liquidity 1000000
python3 report-markets.py --top 10
```

Stable-subtype example:

```bash
cd scripts
python3 scan-markets.py --active-only
python3 rank-markets.py --stable-only --stable-subtype stable-major stable-synthetic stable-rwa --chains ethereum base arbitrum --min-days 7 --max-days 120 --min-liquidity 1000000
python3 report-markets.py --top 10
python3 generate-brief.py --top 8
```

Outputs are written to:
- `data/markets.latest.json`
- `data/ranked.latest.json`
- `data/risk-overrides.json`

## Current implementation notes

- The Phase 2/3 scripts use live market data from Pendle's backend endpoint and currently scan supported chains in paginated batches.
- Risk scoring now supports explicit protocol / underlying overrides via `data/risk-overrides.json`.
- Manual exclusions / score nudges / bucket overrides now live in `data/market-notes.json`.
- Asset-family filtering now supports `stable`, `eth-beta`, `btc-beta`, and `other`, plus a first-class `--stable-only` mode.
- Stable sub-buckets now support `stable-major`, `stable-synthetic`, `stable-rwa`, and `stable-other` through `data/stable-subtype-overrides.json` and `--stable-subtype`.
- Ranking now also emits lightweight allocation suggestions (`tiny`, `small`, `medium`) as a first sizing heuristic.
- The risk model is still intentionally heuristic, not authoritative credit analysis.
- Protocol / underlying risk labels should be treated as a starting point for research review, not a final truth source.
- Use the filters (`--chains`, `--min-days`, `--max-days`, `--min-liquidity`, `--risk`, `--asset-family`, `--stable-only`) to narrow the field into practical buckets before reading the report.
- Use the scripts to narrow the field quickly, then manually inspect the highest-ranked candidates before deploying capital.
- PT looping research must explicitly check Pendle's per-market **Pencosystem** integrations rather than assuming every high-APY PT is loopable.
- Treat `Max Looping APY` and venue metadata from the Pencosystem as decision inputs, but still sanity-check borrow liquidity and route usability.
- Contango support should be surfaced when present because it materially changes operational simplicity, but Contango availability is not a substitute for good loop economics.

## Peter-specific preferences

- Bias toward opportunities that are practical to enter and exit, not just theoretically high-yield
- Separate **hold-to-par** ideas from **near-par rotation** ideas
- Prefer strong underlying quality and usable liquidity over APR-chasing
- Keep the output decision-oriented so it directly informs whether to deploy capital