# qtx alpha
## Metadata

- Canonical URL: https://6ducklearn.com/skills/qtx-alpha/
- Markdown URL: https://6ducklearn.com/skills/qtx-alpha/index.md
- Product: skills
- Category: finance
- Tags: okx, trading, community, okx-marketplace, finance
- Updated: 2026-07-14T03:00:00.357428+00:00
## Summary
Institutional-grade crypto derivatives Alpha signal engine. It scans the perpetual contract market in real-time based on a compiled multi-factor model, automatically outputting executable signals with complete risk control parameters. It features built-in position management, exposure control, and an offline backtesting framework, completing the full-link trading loop through OKX CLI.
## Content
> **6DuckLearn provenance:** Community skill by V哥粉丝, mirrored from the OKX Skills Marketplace (https://www.okx.com/en-sg/agent-tradekit/skills/qtx-alpha). It is not curated, verified, or endorsed by 6DuckLearn or represented as an official OKX publication.
>
> **Financial safety boundary:** Never request secrets in chat. Before any external API call or action that places, cancels, or amends an order; changes leverage; transfers funds; creates or stops a bot; subscribes to or redeems an earn product; or signs/broadcasts a transaction, show the exact live/demo profile, instrument, side, size, price constraints, fees, and worst-case loss, then obtain explicit user approval. Default to read-only or demo mode when uncertain. Treat all analysis as research, not investment advice.

# QTX Alpha Engine

> Institutional-grade quantitative signal engine for crypto perpetual swaps.

机构级加密货币永续合约量化信号引擎。引擎将**多维市场微观结构分析**、**动态因子评分**与**实时风险预算管理**整合为端到端的交易信号管线。核心评分模型、因子权重及风控算法经编译期混淆封装，信号逻辑不可逆向。

### 核心能力

| 能力 | 说明 |
|------|------|
| 全市场实时扫描 | 毫秒级处理全量 SWAP 标的行情快照，多维度筛选与评分 |
| 自适应风控 | 基于账户权益动态计算仓位规模、保证金分配与退出价位 |
| 信号即执行 | 输出包含完整下单参数的 JSON 信号，可直接映射为 OKX CLI 指令 |
| 离线回测 | 支持历史行情回放、多参数网格搜索与交互式 HTML 报告 |
| 跨平台部署 | 基于 Node.js 运行，支持 macOS / Linux / Windows |

### 系统架构

```
                    ┌─────────────────────────┐
  OKX Market Data   │     QTX Alpha Engine     │   OKX Execution
  ───────────────►  │  ┌───────────────────┐   │  ──────────────►
   Tickers (JSON)   │  │ Factor Scoring    │   │   swap place
   Candles (JSON)   │  │ Risk Budgeting    │   │   swap leverage
                    │  │ Position Sizing   │   │   algo orders
                    │  └───────────────────┘   │
                    └─────────────────────────┘
                              │
                        Signals (JSON)
                     inst / px / score
                     t1 / t2 / sz / lev
```

- **信号管线**：引擎接收原始行情 → 因子计算 → 评分排序 → 风险预算分配 → 输出可执行信号
- **执行隔离**：引擎不持有任何凭证，不发起任何网络请求，所有交易指令通过 OKX CLI 外部执行
- **状态无关**：每次扫描独立运算，不依赖本地状态文件，可在任意环境无缝迁移

---

## 环境准备

### 1. OKX CLI

```bash
npm install -g @okx_ai/okx-trade-cli
okx setup   # 按引导完成 API Key 配置
```

### 2. 引擎初始化

引擎以 Node.js 模块形式交付（需 Node.js 18+），无需额外安装依赖：

```bash
QTX="node ${CLAUDE_SKILL_DIR}/bin/qtx.js"
```

后续所有命令中 `$QTX` 即为引擎可执行路径。

### 3. 配置文件

引擎配置位于 `${CLAUDE_SKILL_DIR}/config/engine.json`，参数说明如下：

**扫描参数（`scan` 段）**

| 参数 | 说明 | 默认值 |
|------|------|--------|
| `pc` | 最低信号强度阈值 | 0.15 |
| `px` | 最高信号强度上限 | 1.50 |
| `vl` | 最低流动性门槛（USDT） | 20,000,000 |
| `va` | 最低日均流动性门槛 | 1,000,000 |
| `mp` | 最大并发持仓数 | 3 |
| `ex` | 排除标的列表 | BTC, ETH, USDC |

**执行参数（`exec` 段）**

| 参数 | 说明 | 默认值 |
|------|------|--------|
| `tp` | 主退出目标系数 | 0.30 |
| `sl` | 保护性退出系数 | 0.50 |
| `th` | 持仓超时（小时） | 72 |
| `ch` | 冷却期（小时） | 24 |
| `ru` | 单笔最大敞口（USDT） | 5.0 |
| `lv` | 杠杆倍数 | 3 |
| `tr` | 组合总敞口上限（USDT） | 30.0 |
| `mm` | 保证金使用率上限 | 50% |
| `ic` | 回测初始资金（USDT） | 10,000 |

> 参数键名经过混淆处理，调整时请严格参照上表，错误值可能导致风控失效或信号质量劣化。

---

## 交易流程

### Phase 1 — 信号扫描

获取全量 SWAP 市场快照，送入引擎实时评分：

```bash
okx market tickers SWAP --json | $QTX scan --config "${CLAUDE_SKILL_DIR}/config/engine.json"
```

**输出结构**（JSON）：

```json
{
  "n": 3,
  "signals": [
    { "inst": "...", "px": ..., "score": ..., "t1": ..., "t2": ..., "sz": ..., "cost": ..., "exp": ..., "lev": ... }
  ]
}
```

| 字段 | 含义 | 用途 |
|------|------|------|
| `inst` | 合约标的 ID | 下单 `--instId` |
| `px` | 参考价格 | 信息展示 |
| `score` | 信号强度评分 | 越高置信度越强，用于优先级排序 |
| `liq` | 24h 流动性（$M） | 流动性过低的标的应谨慎 |
| `t1` | 目标价位 1 | 下单 `--tpTriggerPx` |
| `t2` | 目标价位 2 | 下单 `--slTriggerPx` |
| `sz` | 建仓数量 | 下单 `--sz` |
| `cost` | 所需保证金 (USDT) | 余额校验 |
| `exp` | 单笔最大敞口 (USDT) | 风控参考 |
| `lev` | 杠杆倍数 | 下单 `--lever` |

> 当 `"n": 0` 时表示无有效信号，**严禁强制开仓**。

### Phase 2 — 建仓执行

对每个信号 **严格按序** 执行：

**Step 1 — 余额校验**

```bash
okx account balance USDT --json
```

确认可用余额 > 信号 `cost` 字段。多信号并行时需覆盖所有 `cost` 之和。

**Step 2 — 杠杆设置**

```bash
okx swap leverage --instId <inst> --lever <lev> --mgnMode cross
```

> 必须在下单 **之前** 完成，否则可能以错误杠杆成交。

**Step 3 — 开仓下单**

```bash
okx swap place --instId <inst> --side sell --ordType market \
  --sz <sz> --tdMode cross \
  --tpTriggerPx <t1> --tpOrdPx -1 \
  --slTriggerPx <t2> --slOrdPx -1
```

| 参数 | 说明 |
|------|------|
| `--side sell` | 引擎信号方向 |
| `--tpTriggerPx <t1>` | 触达 t1 时市价平仓（主退出） |
| `--slTriggerPx <t2>` | 触达 t2 时市价平仓（保护性退出） |
| `--tdMode cross` | 全仓保证金 |

**Step 4 — 成交确认**

```bash
okx swap orders --instId <inst> --json
```

确认 `state: filled`。部分成交 → 30s 后复查；2 分钟未成交 → 撤单重新评估。

### Phase 3 — 持仓监控

```bash
# 全量持仓
okx swap positions --json

# 指定标的
okx swap positions --instId <inst> --json
```

关键监控指标：
- `upl` — 未实现盈亏
- `mgnRatio` — 保证金率（低于 15% 应考虑减仓或追加保证金）
- `liqPx` — 强平价格

### Phase 4 — 人工干预

**强制平仓：**

```bash
okx swap close --instId <inst> --mgnMode cross
```

**管理委托单：**

```bash
# 查看当前止盈止损委托
okx swap algo orders --instId <inst> --ordType conditional --json

# 撤销指定委托
okx swap algo cancel --instId <inst> --algoId <id>
```

---

## 风险控制体系

引擎内部强制执行多层风控约束。执行层也应独立验证：

| 层级 | 规则 | 说明 |
|------|------|------|
| 仓位层 | 最大并发数 | 受配置 `mp` 参数硬性限制 |
| 仓位层 | 单笔敞口上限 | 信号 `exp` 字段（USDT） |
| 组合层 | 总敞口上限 | 受配置 `tr` 参数限制 |
| 组合层 | 保证金使用率 | 不超过账户权益的 `mm` 配置百分比 |
| 时间层 | 持仓超时 | 超过 `th` 小时后引擎建议平仓 |
| 时间层 | 冷却期 | 同一标的退出后 `ch` 小时内禁止重新开仓 |

**建仓前检查清单：**
1. 账户余额 ≥ 信号 `cost`
2. 当前持仓数 < 最大并发限制
3. 标的杠杆已正确设置
4. 无相同标的的活跃持仓

---

## 策略验证 (回测与参数优化)

### 数据准备

下载目标品种 1H K 线（约 12.5 天历史）：

```bash
okx market tickers SWAP --json > /tmp/tickers.json

python3 -c "
import sys, json
tickers = json.load(open('/tmp/tickers.json'))
exclude = {'BTC-USDT-SWAP','ETH-USDT-SWAP','USDC-USDT-SWAP'}
valid = [t for t in tickers if t['instId'] not in exclude]
valid.sort(key=lambda t: -float(t.get('volCcy24h','0')))
for t in valid[:50]:
    print(t['instId'])
" > /tmp/universe.txt

mkdir -p data
while IFS= read -r inst; do
  okx market candles "\$inst" --bar 1H --limit 300 --json > "data/\${inst}_1H.json"
  sleep 0.5
done < /tmp/universe.txt
```

### 单配置回测

```bash
$QTX backtest --config "${CLAUDE_SKILL_DIR}/config/engine.json" --data-dir data/ --format markdown
```

输出格式：`markdown`（终端友好）| `json`（程序化接入）| `html`（可视化报告）

### 参数网格搜索

```bash
$QTX sweep --config "${CLAUDE_SKILL_DIR}/config/engine.json" \
  --data-dir data/ --output report.html --format html
```

引擎在内部参数空间上进行全组合回测，按风险调整后收益排序，输出交互式 HTML 报告：
- 每组参数的收益率、夏普比率、最大回撤、胜率、盈亏比
- S/A/B/C/D 五档综合评级
- 支持列排序，快速定位最优参数组合

自定义网格可通过 `--sweep-config sweep.json` 传入，格式规范请咨询引擎维护者。

---

## 模拟交易 (Paper Trading)

正式上线前 **必须** 在 OKX 模拟盘完成充分验证：

```bash
# 信号扫描（模拟盘与实盘共享行情数据）
okx market tickers SWAP --json | $QTX scan --config "${CLAUDE_SKILL_DIR}/config/engine.json"

# 模拟盘建仓
okx --demo swap leverage --instId <inst> --lever <lev> --mgnMode cross
okx --demo swap place --instId <inst> --side sell --ordType market \
  --sz <sz> --tdMode cross \
  --tpTriggerPx <t1> --tpOrdPx -1 \
  --slTriggerPx <t2> --slOrdPx -1

# 模拟盘监控
okx --demo swap positions --json

# 模拟盘平仓
okx --demo swap close --instId <inst> --mgnMode cross
```

---

## 免责声明

本引擎仅提供量化信号与风控参数计算，不构成投资建议。加密货币衍生品交易具有高风险，可能导致本金全部损失。使用者应自行评估风险承受能力，并对所有交易决策负全部责任。引擎维护者不对任何交易损失承担责任。

---

*QTX Alpha Engine v0.1.0 — Multi-factor scoring | Built-in risk management | Backtesting & parameter optimization*

## Related Skills

- [alpha vantage](https://6ducklearn.com/skills/alpha-vantage/index.md): Cross-asset analysis is its unique advantage—it can simultaneously pull macro data such as the S&P 500, gold, the dollar index, Federal Reserve interest rates, CPI, GDP, etc., allowing one to see at a glance whether "BTC is moving with risk assets today or pricing independently." In conjunction with the OKX Trade Kit, Alpha Vantage is responsible for in-depth historical and macroeconomic context, while OKX provides real-time prices, funding rates, and order book data, with both complementing each other to form a complete analysis chain.
- [apex crypto intelligence](https://6ducklearn.com/skills/apex-crypto-intelligence/index.md): AI-driven multi-exchange cryptocurrency market analysis, arbitrage detection, and hedge fund-level trading reports using real-time data from major exchanges.
- [btc altcoin market pulse](https://6ducklearn.com/skills/btc-altcoin-market-pulse/index.md): Fetches live OKX market data across Bitcoin and major altcoins, analyzes price momentum, funding rates, open interest, and BTC dominance signals to produce a structured BTC + Altcoin Market Pulse report.
- [cmc okx](https://6ducklearn.com/skills/cmc-okx/index.md): CoinMarketCap × OKX dual engine, a one-stop solution for all your cryptocurrency market data needs. CMC provides market cap, supply, dominance, holding distribution, project background, and macro event calendar; the OKX Trade Kit complements with real-time prices, funding rates, open interest, 70+ technical indicators, and order book depth, with both automatically linked and mutually supportive. It supports natural language triggers, whether you ask "What’s the price of Bitcoin?", "How to read the ETH daily chart?" or "Which coin is surging?", Skill automatically recognizes intent, selects tool combinations, outputs structured tables, and includes a "Quick Take" one-sentence summary.
- [congress trades](https://6ducklearn.com/skills/congress-trades/index.md): Track U.S. Congress members' stock trades in real-time, making "smart political money" impossible to hide. Sync member trade disclosure data to a local SQLite database via the Quiver Quant API, allowing flexible queries by politician name, stock code, party, date range, and trade type. Large trades exceeding the configurable amount threshold (default $15,001) automatically trigger alerts, generating structured Alert files for real-time monitoring.
- [crypto research](https://6ducklearn.com/skills/crypto-research/index.md): A systematic cryptocurrency due diligence report, covering everything from price to holding distribution, from technical aspects to a red flag checklist, all in one go. Integrating the OKX Trade Kit (real-time prices, 70+ technical indicators, funding rates, open interest, order book depth) with CoinMarketCap (market cap, token economics, whale distribution, news sentiment), a seven-step research process is executed in parallel, covering market snapshots, technical analysis, derivatives data, project fundamentals, and recent news. The analysis framework specifically distinguishes between legitimate projects and Meme coins, evaluating key signals such as whale concentration, holder trends, 200-day moving average positions, and funding rate extremes, ultimately outputting a green flag/red flag checklist and low/medium/high/very high risk ratings.