296 lines
8.4 KiB
Plaintext
296 lines
8.4 KiB
Plaintext
{
|
||
"cells": [
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"# 价值投资策略 — 构建与回测\n",
|
||
"\n",
|
||
"## 策略思路\n",
|
||
"价值投资（Value Investing）：买入相对于基本面被低估的股票。\n",
|
||
"本策略使用 **市盈率 PE 倒数（EP）** 与 **市净率 PB 倒数（BP）** 合成价值因子，\n",
|
||
"每月调仓，持有估值最低（价值因子最高）的一批股票。\n",
|
||
"\n",
|
||
"## 研究流程\n",
|
||
"1. 加载数据（价格 + PE/PB 估值指标）\n",
|
||
"2. 价值因子计算与 IC 分析\n",
|
||
"3. 分层回测验证因子有效性\n",
|
||
"4. 构建策略并回测\n",
|
||
"5. 绩效评估"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"import sys\n",
|
||
"sys.path.insert(0, '..')\n",
|
||
"\n",
|
||
"import numpy as np\n",
|
||
"import pandas as pd\n",
|
||
"import matplotlib.pyplot as plt\n",
|
||
"\n",
|
||
"from alpha.config import AlphaConfig\n",
|
||
"from alpha.factors import FactorAnalyzer\n",
|
||
"from alpha.strategy import Strategy, QuantileSignal, EqualWeightAllocator\n",
|
||
"from alpha.backtest import BacktestEngine\n",
|
||
"from alpha.evaluation import PerformanceEvaluator\n",
|
||
"from alpha.data_loader import DataLoader\n",
|
||
"\n",
|
||
"%matplotlib inline\n",
|
||
"plt.rcParams['font.sans-serif'] = ['SimHei']\n",
|
||
"plt.rcParams['axes.unicode_minus'] = False\n",
|
||
"\n",
|
||
"print('模块导入成功')"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 1. 加载数据\n",
|
||
"\n",
|
||
"价值因子需要估值数据（PE / PB），通过 `DataLoader.load_factor` 加载。"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"START_DATE = '2020-01-01'\n",
|
||
"END_DATE = '2024-12-31'\n",
|
||
"\n",
|
||
"loader = DataLoader()\n",
|
||
"\n",
|
||
"# 价格数据 + 估值数据\n",
|
||
"price_data = loader.load_prices(START_DATE, END_DATE)\n",
|
||
"pe_data = loader.load_factor('pe_ttm', START_DATE, END_DATE) # 市盈率 TTM\n",
|
||
"pb_data = loader.load_factor('pb', START_DATE, END_DATE) # 市净率\n",
|
||
"\n",
|
||
"print(f'价格数据: {price_data.shape}')\n",
|
||
"print(f'PE(TTM)数据: {pe_data.shape}')\n",
|
||
"print(f'PB数据: {pb_data.shape}')\n",
|
||
"\n",
|
||
"# 展示某天的估值截面\n",
|
||
"sample = pd.DataFrame({\n",
|
||
" 'PE_TTM': pe_data.iloc[-1],\n",
|
||
" 'PB': pb_data.iloc[-1],\n",
|
||
"}).dropna()\n",
|
||
"sample.head(10)"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 2. 价值因子计算与 IC 分析\n",
|
||
"\n",
|
||
"价值因子 = 标准化后的 (1/PE + 1/PB)：\n",
|
||
"$$\n",
|
||
"Value = Z\\left(\\frac{1}{PE}\\right) + Z\\left(\\frac{1}{PB}\\right)\n",
|
||
"$$\n",
|
||
"PE/PB 越低 → 价值因子越高 → 越被低估。"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"# 剔除 PE/PB 非正值（亏损或净资产为负）\n",
|
||
"ep = 1.0 / pe_data.where(pe_data > 0)\n",
|
||
"bp = 1.0 / pb_data.where(pb_data > 0)\n",
|
||
"\n",
|
||
"# 截面标准化后合成\n",
|
||
"def zscore_cross_section(df):\n",
|
||
" return (df - df.mean(axis=1)) / (df.std(axis=1) + 1e-12)\n",
|
||
"\n",
|
||
"value_factor = zscore_cross_section(ep) + zscore_cross_section(bp)\n",
|
||
"\n",
|
||
"# 未来 20 日收益率（价值因子预测周期较长）\n",
|
||
"fwd_20d = price_data.pct_change(20).shift(-20)\n",
|
||
"\n",
|
||
"factor_stacked = value_factor.stack()\n",
|
||
"fwd_stacked = fwd_20d.stack()\n",
|
||
"\n",
|
||
"analyzer = FactorAnalyzer(factor_stacked, fwd_stacked)\n",
|
||
"ic = analyzer.compute_ic(method='rank')\n",
|
||
"\n",
|
||
"print('价值因子 IC 汇总:')\n",
|
||
"for k, v in analyzer.ic_summary().items():\n",
|
||
" print(f' {k}: {v:.4f}')\n",
|
||
"\n",
|
||
"ic.plot(figsize=(12, 3), title='价值因子 Rank IC 时间序列')\n",
|
||
"plt.axhline(y=0, color='r', linestyle='--')\n",
|
||
"plt.show()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 3. 分层回测"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"quantile_ret = analyzer.quantile_returns(n_quantiles=5)\n",
|
||
"print('各分位组平均收益 (Q1=最贵, Q5=最便宜):')\n",
|
||
"print(quantile_ret)\n",
|
||
"\n",
|
||
"quantile_ret['avg_return'].plot(kind='bar', figsize=(8, 4), color='steelblue')\n",
|
||
"plt.title('价值因子分层收益 (未来20日)')\n",
|
||
"plt.ylabel('平均收益率')\n",
|
||
"plt.grid(True, alpha=0.3)\n",
|
||
"plt.show()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 4. 构建策略并回测"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"# 回测配置\n",
|
||
"config = AlphaConfig(\n",
|
||
" initial_cash=1_000_000,\n",
|
||
" commission_rate=0.0003,\n",
|
||
" slippage=0.001,\n",
|
||
" stamp_tax=0.001,\n",
|
||
")\n",
|
||
"\n",
|
||
"# 价值策略：持有价值因子最高的 20% 股票（最被低估）\n",
|
||
"strategy = Strategy(\n",
|
||
" name='价值投资策略-EP+BP',\n",
|
||
" factors=[],\n",
|
||
" signal_generator=QuantileSignal(\n",
|
||
" n_quantiles=5, long_quantile=5, short_quantile=0,\n",
|
||
" ),\n",
|
||
" weight_allocator=EqualWeightAllocator(max_positions=10),\n",
|
||
" description='每月持有EP+BP合成的价值因子最高的20%股票'\n",
|
||
")\n",
|
||
"\n",
|
||
"engine = BacktestEngine(config)\n",
|
||
"equity_curve = engine.run(\n",
|
||
" strategy=strategy,\n",
|
||
" price_data=price_data,\n",
|
||
" factor_data={'value': value_factor},\n",
|
||
" rebalance_freq='M',\n",
|
||
")\n",
|
||
"\n",
|
||
"print(f'回测完成, 共 {len(equity_curve)} 个交易日')\n",
|
||
"print(f'累计收益率: {(equity_curve[\"nav\"].iloc[-1] - 1) * 100:.2f}%')"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 5. 净值曲线与回撤"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"fig, axes = plt.subplots(2, 1, figsize=(14, 8))\n",
|
||
"\n",
|
||
"# 净值曲线\n",
|
||
"axes[0].plot(equity_curve.index, equity_curve['nav'], label='策略净值', color='purple')\n",
|
||
"axes[0].axhline(y=1.0, color='gray', linestyle='--')\n",
|
||
"axes[0].set_title('价值投资策略净值曲线')\n",
|
||
"axes[0].legend()\n",
|
||
"axes[0].grid(True, alpha=0.3)\n",
|
||
"\n",
|
||
"# 回撤曲线\n",
|
||
"nav = equity_curve['nav']\n",
|
||
"running_max = nav.cummax()\n",
|
||
"drawdown = (nav - running_max) / running_max\n",
|
||
"axes[1].fill_between(equity_curve.index, 0, drawdown.values, color='red', alpha=0.3)\n",
|
||
"axes[1].set_title('回撤曲线')\n",
|
||
"axes[1].grid(True, alpha=0.3)\n",
|
||
"\n",
|
||
"plt.tight_layout()\n",
|
||
"plt.show()"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"## 6. 绩效评估"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"# 基准：股票池等权组合\n",
|
||
"bench_ret = price_data.pct_change().mean(axis=1)\n",
|
||
"\n",
|
||
"evaluator = PerformanceEvaluator(equity_curve, benchmark_returns=bench_ret, risk_free_rate=0.03)\n",
|
||
"print(evaluator.summary())\n",
|
||
"\n",
|
||
"report_df = evaluator.full_report()\n",
|
||
"pd.DataFrame(list(report_df.items()), columns=['指标', '数值']).set_index('指标')"
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "markdown",
|
||
"metadata": {},
|
||
"source": [
|
||
"---\n",
|
||
"## 单独估值指标对比\n",
|
||
"\n",
|
||
"对比单独使用 EP、BP 与合成价值因子的效果："
|
||
]
|
||
},
|
||
{
|
||
"cell_type": "code",
|
||
"execution_count": null,
|
||
"metadata": {},
|
||
"outputs": [],
|
||
"source": [
|
||
"# 单独因子的 IC 对比\n",
|
||
"for name, fac in [('EP', ep), ('BP', bp), ('EP+BP', value_factor)]:\n",
|
||
" a = FactorAnalyzer(fac.stack(), fwd_stacked)\n",
|
||
" s = a.ic_summary()\n",
|
||
" print(f'{name}: IC均值={s[\"IC_Mean\"]:.4f}, ICIR={s[\"IR\"]:.4f}, '\n",
|
||
" f'IC>0占比={s[\"IC>0_Ratio\"]:.2%}')"
|
||
]
|
||
}
|
||
],
|
||
"metadata": {
|
||
"kernelspec": {
|
||
"display_name": "Python 3",
|
||
"language": "python",
|
||
"name": "python3"
|
||
},
|
||
"language_info": {
|
||
"name": "python",
|
||
"version": "3.10.0"
|
||
}
|
||
},
|
||
"nbformat": 4,
|
||
"nbformat_minor": 4
|
||
} |