lighter-perp-dex-trading-bot

KO | EN

gitlite — search

#python #ai #typescript #go #golang #rust #linux #nodejs #cli #kubernetes #awesome-list #pytorch #security #ios

lighter-perp-dex-trading-bot

(★ 10)

lighter-perp-dex-trading-bot — Lighter Perp Dex Trading Bot. lighter-perp-dex-trading-bot is an open-source lighter perp dex trading bot. Find lighter-perp-dex-trading-bot, lighter perp dex trading bot, lighter perp dex trading bot. lighter-perp-dex-trading-bot: Lighter Perp DEX Trading Bot — paper + live CLOB rival to Hyperliquid trading bot

lighter-perp-dex-trading-bot 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Lighter Perp DEX Trading Bot

Emerging Hyperliquid rival — points-aware volume with hard PnL brakes 
 lighter · Emerging L2 / appchain · paper + live · risk-gated · MIT

Languages: **English** · [中文](README.zh.md) · [Deutsch](README.de.md) · [Español](README.es.md)

> **Search keywords:** lighter dex bot · lighter perpetual trading · hyperliquid rival bot · emerging perp DEX

---

## Project workflow

Clone → configure → paper → credentials → live. Risk guardian is always on before any order intent.

```mermaid
flowchart LR
  A[Clone repo] --> B[npm install]
  B --> C[Edit settings.json]
  C --> D[typecheck + test]
  D --> E[npm run paper]
  E --> F{Paper OK?}
  F -->|Yes| G[Fill .env]
  F -->|Tune| C
  G --> H[npm run live --confirm-live]
  H --> I[Monitor / risk]
  I -->|Limit| J[Halt]
```

---

## Platform fit

---

## Trading strategy

Emerging Hyperliquid rivals (Lighter, Aster, and peers) compete on **points, rebates, and thin microstructure**. This bot only trades when **expected edge in bps** clears a hard floor after fees/rebates, optionally boosts score with a **points weight**, prefers maker when enabled, and enforces a **hard PnL stop** — so incentive farming never outruns risk capital.

### How it works
- **Edge filter** — Require expected edge ≥ `minExpectedEdgeBps` after modeled fees/slippage/rebates.
- **Points overlay** — Blend `pointsWeight` into the score when incentive campaigns matter (does not override risk).
- **Maker bias** — If `makerBias`, prefer resting liquidity to harvest rebates / reduce taker bleed.
- **Trade budget** — Cap churn with `maxTradesPerDay` and `riskPerTradePct` sizing.
- **Hard PnL stop** — Flatten / halt new risk when session PnL ≤ `-hardPnlStopUsd`.
- **Guard module** — Emerging venues get extra operational guardrails before live sends.

### When it works
**Best regime:** incentive seasons with real two-sided flow where maker rebates or points improve net bps, without chronic illiquidity. Treat volume as a means, not the goal.

### When it fails
**Fails when:** points change rules overnight, books are too thin for advertised edge, maker bias leaves you adverse-selected, or hard stop is set so tight that noise stops the desk daily.

### Key parameters
- `strategy.minExpectedEdgeBps` — minimum net edge after costs
- `strategy.pointsWeight` — incentive score blend (0–1 scale in settings)
- `strategy.makerBias`, `maxTradesPerDay`, `riskPerTradePct`
- `strategy.hardPnlStopUsd` — session hard stop
- `risk.*` — portfolio brakes

### Strategy risk notes
- Incentive APYs are marketing; adverse selection on thin books is the real cost.
- API / chain maturity risk is higher than on incumbent venues.
- Never size for points; size for liquidation distance and hard stops.

---

## Strategy diagram

Signal → filters → sizing → risk → paper/live → exit:

```mermaid
flowchart TD
  S[Microstructure + incentives] --> E{Edge ≥ min bps?}
  E -->|No| H[Skip]
  E -->|Yes| P[+ points weight]
  P --> M{Maker bias?}
  M -->|Yes| Mk[Resting order]
  M -->|No| Tk[Taker path]
  Mk --> G[Hard PnL stop + risk]
  Tk --> G
  G -->|OK| X[Paper / emerging live]
  G -->|Stop| Halt[Halt]
```

---

## Strategy mathematics

Emerging CLOB desk (Lighter / Aster): score **microstructure edge + points weight**, prefer makers, and enforce a **hard strategy PnL stop** under the portfolio daily loss brake.

Let microstructure edge $e_t$ (bps) and points score $p_t \in [0,1]$ with weight $w=$ `pointsWeight`:

$$
\mathrm{score}_t = (1-w)\,e_t + w\cdot \kappa\, p_t
$$

where $\kappa$ maps points to bps-equivalent. **Trade iff**:

$$
\mathrm{score}_t \ge \texttt{minExpectedEdgeBps}
\;\land\;
\text{daily fills} < \texttt{maxTradesPerDay}
$$

**Maker bias**: if `makerBias`, rest liquidity; else take. **Hard stop**:

$$
\text{halt strategy} \iff \mathrm{PnL}_{\text{day}} \le -\texttt{hardPnlStopUsd}
$$

(defaults −$200, still under portfolio `maxDailyLossUsd: 300`).

### Edge profile chart

```mermaid
xychart-beta
    title "Net PnL vs pointsWeight (conceptual)"
    x-axis ["0.00", "0.10", "0.15", "0.30", "0.50"]
    y-axis "Indexed net PnL" -20 --> 120
    bar [85, 96, 100, 55, -10]
    line [80, 94, 100, 50, -15]
```

*Points as tie-breaker ($w=0.15$) help; overweighting points for farming flips expectancy negative.*

### Implications

- `hardPnlStopUsd` is a venue-level brake — do not disable it to chase seasons.
- Maker bias matters while rebates / points reward resting liquidity.

---

## Parameter explanations

Every knob below maps 1:1 to `settings.json`. Strategy params define the edge; risk params are hard brakes.

| Parameter | Location | Default | Meaning | Why it matters | Typical safe range |
|---|---|---|---|---|---|
| `pointsWeight` | strategy | `0.15` | Weight of points/incentive in edge score | Points farming without ignoring PnL | 0.05 – 0.25 |
| `minExpectedEdgeBps` | strategy | `12` | Min expected edge (bps) to trade | Skip is default on thin emerging books | 8 – 20 |
| `riskPerTradePct` | strategy | `0.35` | Equity % risked per trade | Tighter than mature CLOB desks | 0.2 – 0.45 |
| `maxTradesPerDay` | strategy | `10` | Daily entry cap | Points FOMO throttle | 5 – 14 |
| `makerBias` | strategy | `true` | Prefer maker / rebate path | Emerging venues pay makers; takers bleed | true while rebates exist |
| `hardPnlStopUsd` | strategy | `200` | Strategy-level hard PnL stop ($) | Venue-specific brake under portfolio daily loss | 100 – 250 |
| `maxDailyLossUsd` | risk | `300` | Hard daily PnL halt ($) | Stops revenge trading after a bad session | 150 – 400 on $10k |
| `maxDrawdownPct` | risk | `10` | Peak-to-trough halt (%) | Liquidation-aware book brake for leveraged perps | 6 – 12 |
| `maxNotionalUsd` | risk | `8000` | Gross notional cap | Limits leverage surface across open risk | ≤ 80% equity |
| `maxPositionUsd` | risk | `3000` | Single position / clip cap | Stops one signal from dominating the book | ≤ 30% equity |
| `maxLeverage` | risk | `5` | Hard leverage ceiling | Venue liquidation buffer — not a target | 2 – 5 for desk mode |
| `killSwitch` | risk | `false` | Immediate halt flag | Ops kill without redeploy | flip true on incident |

---

## Tested / recommended parameter set

Paper-desk calibration on the included synthetic market model (same decision path as live). Use as a starting desk, then retune to your RPC and size.

```json
{
  "risk": {
    "maxDailyLossUsd": 300,
    "maxDrawdownPct": 10,
    "maxNotionalUsd": 8000,
    "maxPositionUsd": 3000,
    "maxLeverage": 5,
    "killSwitch": false
  },
  "strategy": {
    "type": "emerging_clob",
    "pointsWeight": 0.15,
    "minExpectedEdgeBps": 12,
    "riskPerTradePct": 0.35,
    "maxTradesPerDay": 10,
    "makerBias": true,
    "hardPnlStopUsd": 200
  },
  "paper": {
    "initialEquityUsd": 10000,
    "feeBps": 6,
    "slippageBps": 4,
    "volatility": 0.014
  }
}
```

---

## Deep analysis — PnL & trading metrics

| Metric | Value |
|--------|------:|
| Net PnL | **$198.5** (1.99%) |
| Win rate | 44.1% |
| Profit factor | 1.36 |
| Expectancy / trade | $5.22 |
| Max drawdown | 5.8% |
| Avg trade R | 0.33 |
| Return / risk (Sharpe-like) | 0.98 |
| Trades in sample | 38 |
| Fee drag | 6.0 bps |
| Slippage drag | 4.0 bps |
| Gas / priority drag | 3.0 bps |

### Equity curve narrative

Emerging CLOB paper (Lighter/Aster-class, $10k, 32 sessions) finished **+$198.5 (+1.99%)** — intentionally modest. Equity flatlines on `edge_low` skips, steps on maker fills when expected edge ≥ 12 bps, and hard-stops strategy PnL at **−$200** before the portfolio daily −$300 brake.

### Fee / slippage / gas impact

6+4 bps; emerging gas/prio ~3 bps. Maker bias + rebate assumption improved expectancy ~+$1.4/trade vs taker-only — still not a license to farm points at any cost.

### Trade count / churn vs edge

38 fills under `maxTradesPerDay: 10`. Raising `pointsWeight` to 0.4 increased volume but net PnL went negative — points are a tie-breaker, not the thesis.

### Regime notes

- Works when microstructure edge clears 12 bps and maker rebates / points are secondary — Lighter points seasons and Aster rebate books alike.
- Fails when incentive farming overrides `minExpectedEdgeBps`, books are too thin for resting size, or hard PnL stops are disabled for “one more farm day”.

---

## Architecture

```
lighter-perp-dex-trading-bot/
├── src/
│   ├── strategy/          # venue-specific engine (emerging_clob)
│   ├── broker/            # paper + live adapters (shared decision path)
│   ├── risk/              # guardian — always before order intent
│   ├── config/            # Zod-validated settings.json
│   ├── points/
│   ├── clob/
│   ├── signals/
│   ├── guard/
│   ├── accounting/
│   ├── analytics/
│   └── ops/
├── settings.json
├── dashboard/
├── tests/
└── docs/banner.jpg
```

---

## Quickstart

```bash
npm install
npm run typecheck && npm test
npm run paper
npm run dashboard
```

### Live

```bash
cp .env.example .env
# fill wallet / RPC credentials (LIGHTER_*)
npm run live -- --confirm-live
```

## Configuration

- **Strategy + risk:** edit `settings.json` (Zod-validated on boot). Strategy block is venue-specific (`emerging_clob`).
- **Secrets only in `.env`:** see `.env.example` (`LIGHTER_*` + chain RPC). Never commit keys.
- Paper and live share `src/strategy` + `src/risk`; only `src/broker` switches.

## Risk management

Concrete values from the shipped `settings.json` — not slogans.

- `risk.maxDailyLossUsd: 300` — halt if daily PnL ≤ −$300
- `risk.maxDrawdownPct: 10` — halt at 10% peak drawdown
- `risk.maxNotionalUsd: 8000` / `maxPositionUsd: 3000` / `maxLeverage: 5`
- `risk.killSwitch: false` — set `true` to freeze all intents
- `live.confirmRequired: true` + `--confirm-live`; prefer `dryRunOrders: true`
- `strategy.hardPnlStopUsd: 200` — venue stop under the $300 daily portfolio halt
- `strategy.minExpectedEdgeBps: 12` + `pointsWeight: 0.15`
- `strategy.makerBias: true` / `maxTradesPerDay: 10` / `riskPerTradePct: 0.35`

- Live refuses to start without `--confirm-live` and wallet/RPC credentials in `.env`
- Prefer dedicated hot wallets; never commit `.env`
- Paper and live share the decision path — only the broker adapter changes

## License

MIT — see [LICENSE](LICENSE).

lighter-perp-dex-trading-bot

lighter-perp-dex-trading-bot 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

lighter-perp-dex-trading-bot

lighter-perp-dex-trading-bot 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README