aeroxw/easy_tdx_max
1
0
Fork 0
mirror of https://ghfast.top/https://github.com/aeroxw/easy_tdx_max.git synced 2026-09-12 13:24:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

release: v1.8.0 - backtest engine with batch strategy comparison

Browse Source 
- Add backtest section to README with CLI usage and run_all_strategies.py demo
- Update all version numbers to 1.8.0 (pyproject.toml, __init__.py, cli/__init__.py, docs/conf.py)
- Fix turtle_breakout strategy: TAQ returns 3 values (UP, MID, DOWN)
- Add run_all_strategies.py batch comparison script
- Update README intro to highlight backtest feature
- Add backtest to CLI command table and architecture tree

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
GitHub
2026-06-09 20:35:38 +08:00
co-authored by Claude Opus 4.8
parent 70c69c8a66
commit b5b5d0dc5b
10 changed files with 401 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.claude/settings.local.json
+3 -1
View File
@@ -76,7 +76,9 @@
"Bash(mypy src/easy_tdx/backtest/types.py)",
"Bash(mypy tests/unit/test_backtest_types.py)",
"Bash(mypy src/easy_tdx/backtest/types.py tests/unit/test_backtest_types.py)",
"Bash(mypy *)"
"Bash(mypy *)",
"Bash(git diff *)",
"Bash(echo \"exit: $?\")"
]
}
}
README.md
+133 -3
View File
@@ -14,13 +14,15 @@ easy-tdx 要做的事很简单：**把机构的数据锁砸开，扔到每个普
它是一个完全免费、无需注册、无需 API Key、纯开源的行**情核武器**。
一行命令，A股、港股、美股、期货——K线、报价、资金流向、板块轮动、分时明细、逐笔成交，**毫秒级拉满**。
**32个技术指标**（MACD、KDJ、RSI、BOLL……连“捉妖大师”和“30日乖离率信号”都给你算好）开箱即用。
**32个技术指标**（MACD、KDJ、RSI、BOLL……连”捉妖大师”和”30日乖离率信号”都给你算好）开箱即用。
**缠论分析**（笔、中枢、买卖点、背驰）一键出结果——你不再需要手画分型、猜线段。
**内置回测引擎**——写个策略文件，一行命令跑回测，9 个经典策略自带，批量对比哪个最赚钱一目了然。
装上就能跑。**Python API + CLI 双通道**，输出 JSON 天然喂给 AI Agent：Claude Code、OpenClaw、Hermes 直接吃。
**你不懂 TCP 协议？不用。**
**你不会写量化框架？不用。**
**你不懂 TCP 协议？不用。**
**你不会写量化框架？不用。**
**你想回测验证策略？自带引擎，不用。**
**你不想给任何平台付一分钱？完全不用。**
`pip install easy-tdx`，30秒后——你屏幕上的数据，和机构看到的**是同一份**。
@@ -265,6 +267,120 @@ easy-tdx chanlun SZ 000001 --period 30MIN
`[✓]` 表示确认背驰。趋势背驰(上)代表上涨趋势可能结束，趋势背驰(下)代表下跌趋势可能结束。
### 回测引擎
内置向量回测引擎，加载 Python 策略文件即可跑回测。策略继承 `Strategy` 基类，在 `init()` 注册指标，在 `next()` 逐 bar 生成买卖信号，引擎完成订单模拟、持仓跟踪和绩效分析。
**单策略回测：**
```bash
easy-tdx backtest SZ 300308 --strategy-file strategies/expma_cross.py --count 2000 --cash 1000000 --adjust QFQ --table
```
输出示例：
```
=== 回测绩效概要 ===
总收益率: 1413.51%
年化收益: 40.85%
最大回撤: 76.75%
夏普比率: 0.88
胜率: 20.8%
交易次数: 24
```
**全策略批量对比：**
项目自带 `run_all_strategies.py`，一次跑完 `strategies/` 下所有策略并排名：
```bash
python -X utf8 run_all_strategies.py SZ 300308 --count 2000 --cash 1000000 --adjust QFQ
```
输出示例（以 SZ 300308 为例）：
```
发现 9 个策略文件
标的: SZ 300308 | K线: 2000 | 资金: 1,000,000 | 复权: QFQ
================================================================================
>> 运行策略: bias_reversal ... 完成 (2.4s)
>> 运行策略: bollinger_breakout ... 完成 (0.6s)
>> 运行策略: expma_cross ... 完成 (0.6s)
>> 运行策略: kdj_golden ... 完成 (0.1s)
>> 运行策略: ma_cross ... 完成 (1.4s)
>> 运行策略: macd_cross ... 完成 (2.1s)
>> 运行策略: rsi_reversal ... 完成 (0.2s)
>> 运行策略: turtle_breakout ... 完成 (0.1s)
>> 运行策略: volume_price ... 完成 (6.3s)
================================================================================
[*] 策略绩效排名 (按总收益率降序)
================================================================================
排名 策略 总收益率 年化收益 最大回撤 夏普 胜率 交易次数 盈亏比
----------------------------------------------------------------------------------------------------
*1* 1 expma_cross 1413.51% 40.85% 76.75% 0.88 20.8% 24 6.45
*2* 2 ma_cross 1258.07% 38.94% 58.01% 0.87 38.2% 55 2.21
*3* 3 turtle_breakout 905.07% 33.76% 48.30% 0.83 75.0% 4 10.14
4 bias_reversal 504.94% 25.47% 42.25% 0.70 66.3% 95 2.08
5 macd_cross 387.67% 22.11% 61.08% 0.60 40.0% 85 2.20
6 volume_price 247.72% 17.01% 65.73% 0.50 43.3% 254 1.40
7 bollinger_breakout 169.65% 13.32% 49.71% 0.44 66.7% 24 1.93
8 rsi_reversal 95.89% 8.85% 56.51% 0.33 57.1% 7 2.48
9 kdj_golden 89.10% 8.36% 61.86% 0.32 66.7% 3 10.49
```
综合评分（夏普 × 0.4 + 收益/回撤 × 0.3 + 胜率 × 0.3）：
```
*1* 1 turtle_breakout 23.04 0.83 0.70 75.0%
*2* 2 bias_reversal 20.35 0.70 0.60 66.3%
*3* 3 bollinger_breakout 20.26 0.44 0.27 66.7%
```
换一个标的再跑：
```bash
# 贵州茅台
python -X utf8 run_all_strategies.py SH 600519 --count 2000 --cash 1000000 --adjust QFQ
```
#### 自带策略示例
`strategies/` 目录下有 9 个开箱即用的策略文件，可直接用于 `--strategy-file`：
| 文件 | 策略 | 类型 | 适合行情 |
|------|------|------|----------|
| `ma_cross.py` | 双均线交叉（MA5/MA20） | 趋势跟踪 | 单边趋势 |
| `expma_cross.py` | EMA12/EMA50 交叉 | 趋势跟踪 | 单边趋势（比 MA 更灵敏） |
| `macd_cross.py` | MACD 金叉死叉 | 趋势跟踪 | 中长线趋势 |
| `bollinger_breakout.py` | 布林带突破 | 震荡反转 | 横盘震荡 |
| `rsi_reversal.py` | RSI 超买超卖 | 反转 | 震荡市 |
| `kdj_golden.py` | KDJ 低位金叉/高位死叉 | 反转 | 短线震荡 |
| `turtle_breakout.py` | 海龟交易法（唐安奇通道） | 趋势突破 | 牛市启动 |
| `bias_reversal.py` | 乖离率反转 | 反转 | 震荡回归 |
| `volume_price.py` | 量价配合 | 综合判断 | 放量突破 |
编写自定义策略只需继承 `Strategy` 基类：
```python
from easy_tdx.backtest import Strategy
from easy_tdx import MyTT
class MyStrategy(Strategy):
def init(self):
self.ma = self.I(MyTT.MA, self.data.close, 10)
def next(self):
if self.data.close[0] > self.ma[self._bar_index]:
self.buy(size=0) # size=0 表示全仓
elif self.position["size"] > 0:
self.sell(size=0) # size=0 表示清仓
```
完整 API 参考：[docs/backtest_usage.md](docs/backtest_usage.md)
### 捉妖大师（重点）
捉妖大师是多周期涨幅共振指标，通过 20/60/120 日涨幅及指数平滑判断短中长线趋势是否同向，用于筛选趋势刚启动的强势股。
@@ -416,6 +532,7 @@ easy-tdx offline sync-all
| `symbol-info` | 个股特征快照 |
| `indicator` | 技术指标计算（32 个：MACD/KDJ/RSI/BOLL/DMI/ATR...） |
| `indicator-list` | 列出可用技术指标 |
| `backtest` | 回测引擎（加载策略文件，输出绩效报告） |
| `f10` | F10 公司信息 |
| `fund-flow` | 历史资金流向 |
| `ex kline` | 扩展市场 K 线 |
@@ -935,6 +1052,7 @@ src/easy_tdx/
├── commands/ # 标准协议命令（无 IO）
├── codec/ # price / volume / datetime / frame / bitmap 编解码
├── chanlun/ # 缠论技术分析（K线合并/分型/笔/线段/中枢/买卖点/背驰）
├── backtest/ # 回测引擎（Strategy基类/向量化引擎/绩效分析）
├── models/ # 纯 dataclass，无业务逻辑
├── offline/ # 离线数据读写模块（读取 + 写入同步）
└── cli/ # easy-tdx CLI（click）
@@ -963,6 +1081,18 @@ ruff format --check src/ tests/ # format check
## Changelog
### 1.8.0 (2026-06-09)
**回测引擎** — 内置向量回测引擎，支持自定义策略回测和全策略批量对比。
- 新增 `backtest` 子包：Strategy 基类、BacktestEngine、OrderSimulator、PortfolioTracker、PerformanceAnalyzer
- 新增 `easy-tdx backtest` CLI 命令，支持 `--strategy-file`、`--cash`、`--commission`、`--adjust` 等参数
- 绩效报告包含 19 项指标：总收益率、年化收益、最大回撤、夏普比率、索提诺、卡玛、胜率、盈亏比等
- 新增 `strategies/` 目录，包含 9 个开箱即用的策略示例（MA/EMA/MACD/BOLL/RSI/KDJ/BIAS/海龟/量价）
- 新增 `run_all_strategies.py` 批量对比脚本，一键跑完全部策略并按收益率和综合评分排名
- 自带策略在 SZ 300308 上 3 年回测：收益率最高 1413%（expma_cross），综合最优 turtle_breakout
- 30+ 离线单元测试覆盖，零网络依赖
### 1.7.1 (2026-06-08)
**Bug 修复** — 修复缠论笔计算在持续下跌/上涨走势中因"分型陷阱"导致近期笔丢失的问题。
docs/conf.py
+1 -1
View File
@@ -4,7 +4,7 @@ copyright = "2025, Justin Gu"
author = "Justin Gu"
# The full version, including alpha/beta/rc tags
release = "1.7.1"
release = "1.8.0"
# -- Extensions ---------------------------------------------------------------
extensions = [
pyproject.toml
+1 -1
View File
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "easy-tdx"
version = "1.7.1"
version = "1.8.0"
description = "通达信 TCP 协议行情数据客户端，支持在线行情、离线数据读取与写入同步"
readme = "README.md"
requires-python = ">=3.10"
run_all_strategies.py
+231
View File
@@ -0,0 +1,231 @@
"""批量回测脚本：依次运行 strategies/ 目录下所有策略并比较结果。
用法::
python run_all_strategies.py SZ 300308 --count 2000 --cash 1000000 --adjust QFQ
输出每个策略的绩效指标，并按总收益率排名。
"""
from __future__ import annotations
import json
import sys
import time
from pathlib import Path
# 确保 easy_tdx 可导入
sys.path.insert(0, str(Path(__file__).parent / "src"))
import click
@click.command()
@click.argument("market")
@click.argument("code")
@click.option("--count", default=2000, type=int, help="K线数量")
@click.option("--cash", default=1000000.0, type=float, help="初始资金")
@click.option("--commission", default=0.0003, type=float, help="佣金率")
@click.option("--adjust", default="QFQ", help="复权: NONE/QFQ/HFQ")
@click.option("--period", default="DAILY", help="K线周期")
def run_all(
market: str,
code: str,
count: int,
cash: float,
commission: float,
adjust: str,
period: str,
) -> None:
"""批量运行 strategies/ 目录下所有策略并比较结果。"""
from easy_tdx.backtest.engine import BacktestEngine
from easy_tdx.backtest.strategy import Strategy
from easy_tdx.cli.parsers import parse_adjust, parse_market, parse_period
from easy_tdx.mac.client import MacClient
# 1. 发现策略文件
strategies_dir = Path(__file__).parent / "strategies"
strategy_files = sorted(strategies_dir.glob("*.py"))
if not strategy_files:
click.echo("未找到策略文件 (strategies/*.py)", err=True)
raise SystemExit(1)
click.echo(f"发现 {len(strategy_files)} 个策略文件")
click.echo(f"标的: {market} {code} | K线: {count} | 资金: {cash:,.0f} | 复权: {adjust}")
click.echo("=" * 80)
# 2. 获取数据（所有策略共享同一份数据）
mkt = parse_market(market)
click.echo("正在获取行情数据...")
client = MacClient.from_best_host()
client.connect()
try:
df = client.get_stock_kline(
mkt,
code,
period=parse_period(period),
start=0,
count=count,
adjust=parse_adjust(adjust),
)
finally:
client.close()
click.echo(f"获取到 {len(df)} 条K线数据")
click.echo("=" * 80)
# 3. 逐个运行策略
results: list[dict] = []
for sf in strategy_files:
strategy_name = sf.stem
click.echo(f"\n>> 运行策略: {strategy_name} ...", nl=False)
# 加载策略类
import importlib.util
spec = importlib.util.spec_from_file_location("strategy_module", sf)
if spec is None or spec.loader is None:
click.echo(" [加载失败]")
continue
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
# 查找 Strategy 子类
strategy_cls = None
for attr_name in dir(module):
obj = getattr(module, attr_name)
try:
if isinstance(obj, type) and issubclass(obj, Strategy) and obj is not Strategy:
strategy_cls = obj
break
except TypeError:
pass
if strategy_cls is None:
click.echo(" [未找到 Strategy 子类]")
continue
# 运行回测
t0 = time.perf_counter()
try:
engine = BacktestEngine(
strategy=strategy_cls,
cash=cash,
commission=commission,
)
result = engine.run(df)
elapsed = time.perf_counter() - t0
perf = result.performance
click.echo(f" 完成 ({elapsed:.1f}s)")
results.append({
"strategy": strategy_name,
"total_return": perf.get("total_return", 0),
"annual_return": perf.get("annual_return", 0),
"max_drawdown": perf.get("max_drawdown", 0),
"sharpe": perf.get("sharpe", 0),
"sortino": perf.get("sortino", 0),
"calmar": perf.get("calmar", 0),
"win_rate": perf.get("win_rate", 0),
"total_trades": perf.get("total_trades", 0),
"profit_factor": perf.get("profit_factor", 0),
"volatility": perf.get("volatility", 0),
})
except Exception as e:
elapsed = time.perf_counter() - t0
click.echo(f" 错误 ({elapsed:.1f}s): {e}")
results.append({
"strategy": strategy_name,
"error": str(e),
})
# 4. 输出排名
click.echo("\n" + "=" * 80)
click.echo("[*] 策略绩效排名 (按总收益率降序)")
click.echo("=" * 80)
# 过滤掉有错误的策略
valid = [r for r in results if "error" not in r]
errored = [r for r in results if "error" in r]
if not valid:
click.echo("所有策略均运行失败！")
for r in errored:
click.echo(f" {r['strategy']}: {r['error']}")
raise SystemExit(1)
# 按总收益率排序
valid.sort(key=lambda x: x["total_return"], reverse=True)
# 表头
click.echo(
f"{'排名':>4} {'策略':<22} {'总收益率':>10} {'年化收益':>10} "
f"{'最大回撤':>10} {'夏普':>8} {'胜率':>8} {'交易次数':>8} {'盈亏比':>8}"
)
click.echo("-" * 100)
for i, r in enumerate(valid, 1):
medal = " *1*" if i == 1 else " *2*" if i == 2 else " *3*" if i == 3 else " "
click.echo(
f"{medal}{i:>2} {r['strategy']:<22} "
f"{r['total_return']:>9.2%} "
f"{r['annual_return']:>9.2%} "
f"{r['max_drawdown']:>9.2%} "
f"{r['sharpe']:>8.2f} "
f"{r['win_rate']:>7.1%} "
f"{r['total_trades']:>8} "
f"{r['profit_factor']:>8.2f}"
)
# 最佳策略详细报告
best = valid[0]
click.echo("\n" + "=" * 80)
click.echo(f"[BEST] 最佳策略: {best['strategy']}")
click.echo("=" * 80)
click.echo(f" 总收益率: {best['total_return']:.2%}")
click.echo(f" 年化收益: {best['annual_return']:.2%}")
click.echo(f" 最大回撤: {best['max_drawdown']:.2%}")
click.echo(f" 夏普比率: {best['sharpe']:.2f}")
click.echo(f" 索提诺: {best['sortino']:.2f}")
click.echo(f" 卡玛比率: {best['calmar']:.2f}")
click.echo(f" 胜率: {best['win_rate']:.1%}")
click.echo(f" 交易次数: {best['total_trades']}")
click.echo(f" 盈亏比: {best['profit_factor']:.2f}")
click.echo(f" 年化波动: {best['volatility']:.4f}")
# 综合评分（综合夏普、收益率、回撤）
click.echo("\n" + "=" * 80)
click.echo("[*] 综合评分排名 (Sharpe*0.4 + Ret/DD*0.3 + WinRate*0.3)")
click.echo("=" * 80)
scored = []
for r in valid:
# 避免除以零
ret_dd_ratio = r["annual_return"] / r["max_drawdown"] if r["max_drawdown"] > 1e-6 else 999.0
score = r["sharpe"] * 0.4 + ret_dd_ratio * 0.3 + r["win_rate"] * 100 * 0.3
scored.append((r, score))
scored.sort(key=lambda x: x[1], reverse=True)
click.echo(
f"{'排名':>4} {'策略':<22} {'综合评分':>10} {'夏普':>8} {'收益/回撤':>10} {'胜率':>8}"
)
click.echo("-" * 70)
for i, (r, score) in enumerate(scored, 1):
ret_dd_ratio = r["annual_return"] / r["max_drawdown"] if r["max_drawdown"] > 1e-6 else 999.0
medal = " *1*" if i == 1 else " *2*" if i == 2 else " *3*" if i == 3 else " "
click.echo(
f"{medal}{i:>2} {r['strategy']:<22} {score:>10.2f} "
f"{r['sharpe']:>8.2f} {ret_dd_ratio:>10.2f} {r['win_rate']:>7.1%}"
)
# 报告错误
if errored:
click.echo("\n[!] 以下策略运行失败:")
for r in errored:
click.echo(f" {r['strategy']}: {r['error']}")
if __name__ == "__main__":
run_all()
src/easy_tdx/__init__.py
+1 -1
View File
@@ -107,4 +107,4 @@ __all__ = [
"save_best_ex_host",
]
__version__ = "1.4.0"
__version__ = "1.8.0"
src/easy_tdx/backtest/portfolio.py
+26 -32
View File
@@ -46,17 +46,12 @@ class PortfolioTracker:
Args:
trades: 交易列表
"""
# 构建 datetime → Trade 映射
buy_map: dict[int, Trade] = {}
sell_map: dict[int, Trade] = {}
# 构建 datetime → [Trade] 映射（支持同 bar 多笔交易）
trade_map: dict[int, list[Trade]] = {}
for trade in trades:
if trade.rejected:
continue
if trade.direction == "BUY":
buy_map[trade.datetime] = trade
else:
sell_map[trade.datetime] = trade
trade_map.setdefault(trade.datetime, []).append(trade)
# 遍历每个 bar
for i in range(self._n):
@@ -68,33 +63,32 @@ class PortfolioTracker:
self._position[i] = self._position[i - 1]
self._avg_price[i] = self._avg_price[i - 1]
# 处理买入
if dt in buy_map:
trade = buy_map[dt]
cost = trade.size * trade.price + trade.commission + trade.slippage
self._cash[i] -= cost
# 处理该 bar 的所有交易
for trade in trade_map.get(dt, []):
if trade.direction == "BUY":
cost = trade.size * trade.price + trade.commission + trade.slippage
self._cash[i] -= cost
# 更新均价
if self._position[i] > 0:
total_cost = self._position[i] * self._avg_price[i] + trade.size * trade.price
self._position[i] += trade.size
self._avg_price[i] = total_cost / self._position[i]
else:
# 新开仓或从空仓开仓
self._position[i] = trade.size
self._avg_price[i] = trade.price
# 更新均价
if self._position[i] > 0:
prev_cost = self._position[i] * self._avg_price[i]
total_cost = prev_cost + trade.size * trade.price
self._position[i] += trade.size
self._avg_price[i] = total_cost / self._position[i]
else:
# 新开仓或从空仓开仓
self._position[i] = trade.size
self._avg_price[i] = trade.price
# 处理卖出
if dt in sell_map:
trade = sell_map[dt]
proceeds = trade.size * trade.price - trade.commission - trade.slippage
self._cash[i] += proceeds
self._position[i] -= trade.size
elif trade.direction == "SELL":
proceeds = trade.size * trade.price - trade.commission - trade.slippage
self._cash[i] += proceeds
self._position[i] -= trade.size
# 清空持仓时归零
if self._position[i] <= 0:
self._position[i] = 0.0
self._avg_price[i] = 0.0
# 清空持仓时归零
if self._position[i] <= 0:
self._position[i] = 0.0
self._avg_price[i] = 0.0
@property
def equity_curve(self) -> pd.DataFrame:
src/easy_tdx/backtest/types.py
+3 -3
View File
@@ -19,7 +19,7 @@ class Signal:
"""策略产生的交易信号。
Attributes:
datetime: 信号时间（Unix timestamp 毫秒）
datetime: 信号时间（YYYYMMDD 整数格式，如 20240101）
direction: 交易方向
size: 交易数量（0 = 全仓/清仓）
price: 限价（None = 市价单）
@@ -43,7 +43,7 @@ class Trade:
"""已成交记录。
Attributes:
datetime: 成交时间（Unix timestamp 毫秒）
datetime: 成交时间（YYYYMMDD 整数格式，如 20240101）
direction: 交易方向
size: 成交数量
price: 成交价格
@@ -71,7 +71,7 @@ class Position:
"""持仓快照。
Attributes:
datetime: 快照时间（Unix timestamp 毫秒）
datetime: 快照时间（YYYYMMDD 整数格式，如 20240101）
size: 持仓数量（正=多头，负=空头，0=空仓）
avg_price: 平均持仓成本
market_value: 市值
src/easy_tdx/cli/__init__.py
+1 -1
View File
@@ -23,7 +23,7 @@ from ..backtest.cli import backtest
@click.group()
@click.version_option(version="1.7.0", prog_name="easy-tdx")
@click.version_option(version="1.8.0", prog_name="easy-tdx")
def cli() -> None:
"""easy-tdx -- 通达信行情数据 CLI（默认 JSON 输出，适合 Agent 使用）。
strategies/turtle_breakout.py
+1 -1
View File
@@ -16,7 +16,7 @@ class TurtleStrategy(Strategy):
"""海龟交易法（唐安奇通道突破）策略。"""
def init(self) -> None:
self.upper, self.lower = self.I(MyTT.TAQ, self.data.high, self.data.low, 20)
self.upper, _, self.lower = self.I(MyTT.TAQ, self.data.high, self.data.low, 20)
def next(self) -> None:
cur = self.data.close[0]