QKA — 快量化

Quant Kit for A-shares

简洁易用的 A 股量化回测框架。

安装

包安装

pip install qka

需要 Python 3.10+。

AI 技能安装

为 Claude Code、Cursor 等 AI 编程工具安装 QKA 技能：

npx skills add zsrl/qka

安装后，AI 助手即可自动加载 QKA 框架的 API 文档，生成符合规范的量化策略代码。

快速上手

数据

from qka import Data

data = Data(
    symbols=['sz.000001', 'sh.600000'],
    indicators={
        'sma_5':  ('ta.trend.sma_indicator', 'close', 5),
        'rsi_14': ('ta.momentum.rsi', 'close', 14),
    },
)
df = data.get()  # 返回宽表 DataFrame，列名 {symbol}|{factor}

策略

from qka import Strategy

class MyStrategy(Strategy):
    def __init__(self):
        super().__init__()
        self.lookback = 20  # 自定义参数

    def on_bar(self, date):
        close = self.get('close')            # 当前横截面
        hist = self.history('close', 20)     # 历史窗口
        # 交易逻辑：self.broker.buy / self.broker.sell
        # 仓位计算：self.sizing.percent / self.sizing.fixed_shares

回测

from qka import Backtest

strategy = MyStrategy()
bt = Backtest(data, strategy)
bt.run(cash=200000, start_date='2024-01-01', benchmark='sh.000300')
print(bt.metrics['total_return_pct'])   # 总收益率
print(bt.metrics['sharpe_ratio'])        # 夏普比率

核心能力

多数据源 — baostock（默认）、akshare、QMT
预计算指标 — ta 库全部 60+ 指标，('ta.trend.sma_indicator', 'close', 5) 格式直接透传
事件驱动回测 — 按日推进，self.get() 横截面 + self.history() 窗口序列
仓位管理 — sizing.percent() / sizing.fixed_amount() / sizing.fixed_shares() / sizing.atr_risk()
交易模拟 — 佣金万 2.5、印花税万 5（仅卖出）、滑点 0.1%，最低佣金 5 元
绩效指标 — 总收益率、年化、夏普比率、最大回撤、Calmar、胜率、盈亏比等 13 项
基准对比 — 支持沪深 300（或指定指数）对比

文档

框架 API 完整文档见 skills/qka/SKILL.md——所有类的方法签名、参数、返回值和约束都在里面。

完整示例

from qka import Data, Strategy, Backtest

data = Data(
    symbols=['sz.000001'],
    indicators={
        'sma_5':  ('ta.trend.sma_indicator', 'close', 5),
        'sma_20': ('ta.trend.sma_indicator', 'close', 20),
    },
)

class MaCross(Strategy):
    def __init__(self):
        super().__init__()
        self.pct = 0.2

    def on_bar(self, date):
        close = self.get('close')
        fast = self.get('sma_5')
        slow = self.get('sma_20')
        for sym in close.index:
            price = float(close[sym])
            if price <= 0:
                continue
            if fast[sym] > slow[sym]:
                size = self.sizing.percent(self.pct, price)
                if size > 0:
                    self.broker.buy(sym, price, size)
            else:
                pos = self.broker.positions.get(sym, {}).get('size', 0)
                if pos > 0:
                    self.broker.sell(sym, price, pos)

strategy = MaCross()
bt = Backtest(data, strategy)
bt.run(cash=200000, start_date='2024-01-01')
print(bt.metrics['total_return_pct'])

许可证

MIT

致谢

baostock — 免费 A 股数据
ta — 技术指标库

⚠️ 量化交易存在风险，请充分了解后再使用本框架。

Metadata

Release files for qka 2.2.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for qka 2.2.2
File	Size	Uploaded
qka-2.2.2.tar.gz	52.7 kB	Sep 10, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for qka 2.2.2
File	Interpreter	ABI	Platform	Reset
qka-2.2.2-py3-none-any.whl 57.8 kB Sep 10, 2026	Python 3	none	any	Details

Total release size: 110.5 kB

Release files / qka-2.2.2.tar.gz

Download URL	qka-2.2.2.tar.gz
Size	52.7 kB
Tags	Source
SHA-256 checksum How to use checksums	`07201b6c1d67acf9a011dd606a08344d53b38018bdb1d70ca1a1894da60d92bd`
BLAKE2b-256 checksum How to use checksums	`6818e8234587bf6b355936052e17d3f8b16af48a093ea59e4840e54746d5ab56`
Upload date	Sep 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Release files / qka-2.2.2-py3-none-any.whl

Download URL	qka-2.2.2-py3-none-any.whl
Size	57.8 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`e99bb8e18b48f827ebfab7b0b8f4fce4800af91b9f4c38916dcc839d533dfc86`
BLAKE2b-256 checksum How to use checksums	`e07370a553b9f8013ad3972beb94c145f8abdd179faf89db7412ec49d928eb77`
Upload date	Sep 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`