aszer/factorhack
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

添加talib算子

Browse Source
This commit is contained in:
aszerW
2025-11-09 20:19:08 +08:00
parent dc3d41d6e5
commit e5beada25e
5 changed files with 512 additions and 404 deletions
Download Patch File Download Diff File Expand all files Collapse all files

71
factor_mining/gp_miner.py
View File

@@ -1,6 +1,7 @@
"""
DEAP遗传编程挖掘器实现
"""
import random
import operator
from typing import List, Tuple, Optional
@@ -17,6 +18,7 @@ from data import compute_forward_returns
@dataclass
class GPConfig(MiningConfig):
"""GP挖掘配置"""
population_size: int = 200
generations: int = 30
tournament_size: int = 5
@@ -44,7 +46,9 @@ class GPMiner(FactorMiner):
def _build_pset(self, feature_names: List[str]) -> gp.PrimitiveSetTyped:
"""构建GP原始集合"""
registry = get_registry()
pset = gp.PrimitiveSetTyped("MAIN", [np.ndarray for _ in feature_names], np.ndarray)
pset = gp.PrimitiveSetTyped(
"MAIN", [np.ndarray for _ in feature_names], np.ndarray
)
# 命名参数
for i, name in enumerate(feature_names):
@@ -63,20 +67,18 @@ class GPMiner(FactorMiner):
pset.addPrimitive(op.func, [np.ndarray], np.ndarray, name=op_name)
elif len(params) == 2:
# 二元算子
pset.addPrimitive(op.func, [np.ndarray, np.ndarray], np.ndarray, name=op_name)
pset.addPrimitive(
op.func, [np.ndarray, np.ndarray], np.ndarray, name=op_name
)
# 添加常量
def _const() -> np.ndarray:
return np.array(random.uniform(-2.0, 2.0))
pset.addEphemeralConstant("const", _const, np.ndarray)
# def _const() -> np.ndarray:
# return np.array(random.uniform(-2.0, 2.0))
# pset.addEphemeralConstant("const", _const, np.ndarray)
return pset
def _evaluate_individual(
self,
individual,
target: pd.Series
) -> Tuple[float]:
def _evaluate_individual(self, individual, target: pd.Series) -> Tuple[float]:
"""评估个体适应度"""
func = self.toolbox.compile(expr=individual)
@@ -106,7 +108,10 @@ class GPMiner(FactorMiner):
return (-1e6,)
from validation import compute_rolling_ic
ic_series = compute_rolling_ic(factor, target, window=window, method=self.config.ic_method)
ic_series = compute_rolling_ic(
factor, target, window=window, method=self.config.ic_method
)
mean_ic = ic_series.mean()
if not np.isfinite(mean_ic):
@@ -122,9 +127,7 @@ class GPMiner(FactorMiner):
return (fitness,)
def _individual_to_formula(
self,
individual,
feature_names: List[str]
self, individual, feature_names: List[str]
) -> FactorFormula:
"""将GP个体转换为因子公式"""
# GP表达式是PrimitiveTree，转换为字符串后是函数调用形式
@@ -142,10 +145,7 @@ class GPMiner(FactorMiner):
return FactorFormula(expr_str, feature_names)
def mine(
self,
data: pd.DataFrame,
feature_cols: List[str],
price_col: str = "close"
self, data: pd.DataFrame, feature_cols: List[str], price_col: str = "close"
) -> List[FactorFormula]:
"""执行GP挖掘"""
if self.config.seed is not None:
@@ -175,37 +175,45 @@ class GPMiner(FactorMiner):
gp.genHalfAndHalf,
pset=self.pset,
min_=1,
max_=self.config.max_depth_init
max_=self.config.max_depth_init,
)
self.toolbox.register(
"individual", tools.initIterate, creator.Individual, self.toolbox.expr
)
self.toolbox.register(
"population", tools.initRepeat, list, self.toolbox.individual
)
self.toolbox.register("individual", tools.initIterate, creator.Individual, self.toolbox.expr)
self.toolbox.register("population", tools.initRepeat, list, self.toolbox.individual)
self.toolbox.register("compile", gp.compile, pset=self.pset)
self.toolbox.register(
"evaluate",
self._evaluate_individual,
target=target
)
self.toolbox.register("evaluate", self._evaluate_individual, target=target)
# 遗传算子
self.toolbox.register("select", tools.selTournament, tournsize=self.config.tournament_size)
self.toolbox.register(
"select", tools.selTournament, tournsize=self.config.tournament_size
)
self.toolbox.register("mate", gp.cxOnePoint)
self.toolbox.register("expr_mut", gp.genFull, min_=0, max_=2)
self.toolbox.register("mutate", gp.mutUniform, expr=self.toolbox.expr_mut, pset=self.pset)
self.toolbox.register(
"mutate", gp.mutUniform, expr=self.toolbox.expr_mut, pset=self.pset
)
# 控制树深度
self.toolbox.decorate(
"mate",
gp.staticLimit(key=operator.attrgetter("height"), max_value=self.config.max_depth)
gp.staticLimit(
key=operator.attrgetter("height"), max_value=self.config.max_depth
),
)
self.toolbox.decorate(
"mutate",
gp.staticLimit(key=operator.attrgetter("height"), max_value=self.config.max_depth)
gp.staticLimit(
key=operator.attrgetter("height"), max_value=self.config.max_depth
),
)
# 运行进化
pop = self.toolbox.population(n=self.config.population_size)
hof = tools.HallOfFame(maxsize=max(5, self.config.elitism))
hof = tools.HallOfFame(maxsize=max(5000, self.config.elitism))
stats_fit = tools.Statistics(lambda ind: ind.fitness.values[0])
stats_size = tools.Statistics(len)
@@ -233,4 +241,3 @@ class GPMiner(FactorMiner):
formulas.append(formula)
return formulas

362
factor_mining/operators.py
View File

@@ -9,6 +9,8 @@ from typing import Dict, Callable, List, Optional, Any
from abc import ABC, abstractmethod
import inspect
import talib
class Operator(ABC):
"""算子基类"""
@@ -99,6 +101,9 @@ def get_registry() -> OperatorRegistry:
return _registry
# 定义period参数的值范围
PERIOD_RANGE = range(10, 100) # 10到99
# ==================== 基础数学算子 ====================
@@ -153,8 +158,6 @@ def _pow(x: np.ndarray, y: np.ndarray) -> np.ndarray:
# ==================== 时间序列算子 ====================
def _rolling_mean(x: np.ndarray, window: int) -> np.ndarray:
s = pd.Series(x)
return s.rolling(window, min_periods=max(2, window // 2)).mean().to_numpy()
@@ -184,8 +187,20 @@ def _delay(x: np.ndarray, period: int) -> np.ndarray:
return s.shift(period).to_numpy()
def _pct_change(x: np.ndarray, period: int = 1) -> np.ndarray:
"""百分比变化"""
s = pd.Series(x)
return s.pct_change(periods=period, fill_method=None).to_numpy()
# 注册单参数百分比变化算子
@register_operator("pct", "百分比变化: PCT(x, 1)")
def _pct(x: np.ndarray) -> np.ndarray:
return _pct_change(x, 1)
# 注册时间序列算子（带不同窗口）
for w in (3, 6, 12, 24, 48, 96):
for w in PERIOD_RANGE:
_registry.register_function(
f"sma{w}", lambda x, w=w: _rolling_mean(x, w), f"简单移动平均: SMA(x, {w})"
)
@@ -203,6 +218,347 @@ for w in (3, 6, 12, 24, 48, 96):
)
# ==================== 技术指标算子（含自定义与ta-lib）====================
def _try_float(x):
try:
return float(x)
except Exception:
return x
def _convert_input(v):
# 如果是pd.Series,返回np.ndarray; 如果已经是np.ndarray则原样返回
if isinstance(v, pd.Series):
return v.values
return v
# 注册 ta-lib 技术指标
# 获取 TA-Lib 的所有函数名（常用financial indicators均为大写）
talib_func_list = [f for f in dir(talib) if f.isupper() and callable(getattr(talib, f))]
# 定义需要生成多版本的参数名（period相关参数）
# 按优先级排序，优先匹配主要的period参数
PERIOD_PARAM_NAMES = [
"timeperiod", # 最常见的参数名
"period", # 通用period参数
"optintimeperiod", # TA-Lib内部参数名
]
# 多period参数的函数（需要特殊处理）
# 对于这些函数，明确指定主要period参数，避免自动检测错误
MULTI_PERIOD_FUNCTIONS = {
# 函数名: (主要period参数名, 次要period参数列表，仅用于文档)
"MACD": ("fastperiod", ["slowperiod", "signalperiod"]),
"MACDEXT": ("fastperiod", ["slowperiod", "signalperiod"]),
"MACDFIX": ("signalperiod", []),
"STOCH": ("fastk_period", ["slowk_period", "slowd_period"]),
"STOCHF": ("fastk_period", ["fastd_period"]),
"STOCHRSI": ("timeperiod", ["fastk_period", "fastd_period"]),
"BBANDS": ("timeperiod", ["nbdevup", "nbdevdn"]),
"APO": ("fastperiod", ["slowperiod"]),
"PPO": ("fastperiod", ["slowperiod"]),
"ULTOSC": ("timeperiod1", ["timeperiod2", "timeperiod3"]),
"BOP": ("", []), # 无period参数，注册默认版本
}
def build_talib_wrapper(func, func_name, fixed_params=None):
"""构建talib函数包装器，支持固定某些参数"""
fixed_params = fixed_params or {}
def _talib_wrap(*args, **kwargs):
# 合并固定参数和传入参数
merged_kwargs = {**fixed_params, **kwargs}
# ta-lib 有些函数只支持关键字参数
# 自动转换所有输入类型
args = tuple(_convert_input(arg) for arg in args)
for k in merged_kwargs:
merged_kwargs[k] = _convert_input(merged_kwargs[k])
result = func(*args, **merged_kwargs)
# TA-Lib有些输出是tuple（比如MACD），统一返回ndarray/tuple[ndarray]
if isinstance(result, tuple):
# 保持tuple结构
return tuple(
np.asarray(item) if item is not None else None for item in result
)
return np.asarray(result)
_talib_wrap.__name__ = f"talib_{func_name.lower()}"
return _talib_wrap
for func_name in talib_func_list:
func = getattr(talib, func_name)
sig = inspect.signature(func)
params = sig.parameters
# 检查是否在特殊配置字典中
if func_name in MULTI_PERIOD_FUNCTIONS:
main_period_param, _ = MULTI_PERIOD_FUNCTIONS[func_name]
# 如果配置中指定了主要period参数，使用它
if main_period_param and main_period_param in params:
for period_value in PERIOD_RANGE:
fixed_params = {main_period_param: period_value}
wrapper = build_talib_wrapper(func, func_name, fixed_params)
op_name = f"talib_{func_name.lower()}_{period_value}"
desc = f"ta-lib: {func_name}({main_period_param}={period_value})"
_registry.register_function(op_name, wrapper, desc)
else:
# 配置中指定无period参数，注册默认版本
wrapper = build_talib_wrapper(func, func_name)
op_name = f"talib_{func_name.lower()}"
desc = f"ta-lib: {func_name}"
_registry.register_function(op_name, wrapper, desc)
else:
# 不在特殊配置中，自动检测period参数
period_params = {}
for param_name, param in params.items():
param_lower = param_name.lower()
# 检查是否是period相关参数
if any(
period_keyword in param_lower for period_keyword in PERIOD_PARAM_NAMES
):
period_params[param_name] = param
if period_params:
# 如果有period参数，为每个period值生成一个版本
# 优先选择timeperiod，否则选择第一个
main_period_param = None
for preferred in ["timeperiod", "period", "optintimeperiod"]:
for param_name in period_params.keys():
if preferred in param_name.lower():
main_period_param = param_name
break
if main_period_param:
break
if not main_period_param:
main_period_param = list(period_params.keys())[0]
for period_value in PERIOD_RANGE:
fixed_params = {main_period_param: period_value}
wrapper = build_talib_wrapper(func, func_name, fixed_params)
op_name = f"talib_{func_name.lower()}_{period_value}"
desc = f"ta-lib: {func_name}({main_period_param}={period_value})"
_registry.register_function(op_name, wrapper, desc)
else:
# 如果没有period参数，注册默认版本
wrapper = build_talib_wrapper(func, func_name)
op_name = f"talib_{func_name.lower()}"
desc = f"ta-lib: {func_name}"
_registry.register_function(op_name, wrapper, desc)
# ==================== 自定义常见技术指标 ====================
def _ewm_forward(x: np.ndarray, alpha: float) -> np.ndarray:
"""指数加权移动平均（前向计算）"""
result = np.zeros_like(x)
if len(x) == 0:
return result
result[0] = x[0]
for i in range(1, len(x)):
result[i] = x[i] * alpha + (1 - alpha) * result[i - 1]
return result
def _rsv(x: np.ndarray, window: int) -> np.ndarray:
"""相对强弱值: (当前值 - 最小值) / (最大值 - 最小值)"""
s = pd.Series(x)
rolling = s.rolling(window, min_periods=max(2, window // 2), closed="both")
min_val = rolling.min()
max_val = rolling.max()
diff = max_val - min_val
# 避免除零
diff = np.where(np.abs(diff) < 1e-12, np.nan, diff)
result = (s - min_val) / diff
return result.to_numpy()
def _bband(x: np.ndarray, window: int) -> np.ndarray:
"""布林带指标: (当前值 - 均值) / 标准差"""
s = pd.Series(x)
rolling = s.rolling(window, min_periods=max(2, window // 2), closed="both")
mean_val = rolling.mean()
std_val = rolling.std()
# 避免除零
std_val = np.where(np.abs(std_val) < 1e-12, np.nan, std_val)
result = (s - mean_val) / std_val
return result.to_numpy()
def _rsi(x: np.ndarray, window: int, threshold: float = 0.00001) -> np.ndarray:
"""相对强弱指标: 上涨和下跌的比例"""
s = pd.Series(x)
diff = s.diff()
rolling = diff.rolling(window, min_periods=max(2, window // 2), closed="both")
def _rsi_calc(series):
up_sum = series[series > threshold].sum()
down_sum = abs(series[series < -threshold].sum())
total = up_sum + down_sum
if total < 1e-12:
return np.nan
return up_sum / total
result = rolling.apply(_rsi_calc, raw=False)
return result.to_numpy()
def _rolling_skew(x: np.ndarray, window: int) -> np.ndarray:
"""滚动偏度"""
s = pd.Series(x)
return (
s.rolling(window, min_periods=max(2, window // 2), closed="both")
.skew()
.to_numpy()
)
def _rolling_kurtosis(x: np.ndarray, window: int) -> np.ndarray:
"""滚动峰度"""
s = pd.Series(x)
return (
s.rolling(window, min_periods=max(2, window // 2), closed="both")
.kurt()
.to_numpy()
)
def _rolling_linear(x: np.ndarray, window: int) -> np.ndarray:
"""滚动线性回归斜率"""
s = pd.Series(x)
def _linear_slope(series):
valid = series.dropna()
if len(valid) < 2:
return np.nan
try:
coeffs = np.polyfit(np.arange(len(valid)), valid.values, 1)
return coeffs[0]
except:
return np.nan
result = s.rolling(window, min_periods=max(2, window // 2), closed="both").apply(
_linear_slope, raw=False
)
return result.to_numpy()
def _rolling_autocorr(x: np.ndarray, window: int, lag: int = 1) -> np.ndarray:
"""滚动自相关"""
s = pd.Series(x)
result = s.rolling(window, min_periods=max(2, window // 2), closed="both").apply(
lambda series: (
series.autocorr(lag=lag) if len(series.dropna()) >= 2 else np.nan
),
raw=False,
)
return result.to_numpy()
def _rolling_max(x: np.ndarray, window: int) -> np.ndarray:
"""滚动最大值"""
s = pd.Series(x)
return (
s.rolling(window, min_periods=max(2, window // 2), closed="both")
.max()
.to_numpy()
)
def _rolling_min(x: np.ndarray, window: int) -> np.ndarray:
"""滚动最小值"""
s = pd.Series(x)
return (
s.rolling(window, min_periods=max(2, window // 2), closed="both")
.min()
.to_numpy()
)
def _huanbi(x: np.ndarray, window: int) -> np.ndarray:
"""环比: 当前值 / 窗口起始值"""
s = pd.Series(x)
def _huanbi_calc(series):
if len(series) < 2:
return np.nan
start_val = series.iloc[0]
end_val = series.iloc[-1]
if abs(start_val) < 1e-12:
return np.nan
return end_val / start_val
result = s.rolling(window, min_periods=max(2, window // 2), closed="both").apply(
_huanbi_calc, raw=False
)
return result.to_numpy()
# 注册技术指标算子（带不同窗口）
for w in PERIOD_RANGE:
# EWM算子（使用固定alpha值）
alpha = 2.0 / (w + 1)
_registry.register_function(
f"ewm{w}",
lambda x, w=w, a=alpha: _ewm_forward(x, a),
f"指数加权移动平均: EWM(x, {w})",
)
# 百分比变化
_registry.register_function(
f"pct{w}", lambda x, w=w: _pct_change(x, w), f"百分比变化: PCT(x, {w})"
)
# RSV（相对强弱值）
_registry.register_function(
f"rsv{w}", lambda x, w=w: _rsv(x, w), f"相对强弱值: RSV(x, {w})"
)
# 布林带
_registry.register_function(
f"bband{w}", lambda x, w=w: _bband(x, w), f"布林带指标: BBAND(x, {w})"
)
# RSI
_registry.register_function(
f"rsi{w}", lambda x, w=w: _rsi(x, w), f"相对强弱指标: RSI(x, {w})"
)
# 统计量
_registry.register_function(
f"skew{w}", lambda x, w=w: _rolling_skew(x, w), f"滚动偏度: SKEW(x, {w})"
)
_registry.register_function(
f"kurt{w}", lambda x, w=w: _rolling_kurtosis(x, w), f"滚动峰度: KURT(x, {w})"
)
_registry.register_function(
f"linear{w}",
lambda x, w=w: _rolling_linear(x, w),
f"滚动线性斜率: LINEAR(x, {w})",
)
_registry.register_function(
f"autocorr{w}",
lambda x, w=w: _rolling_autocorr(x, w),
f"滚动自相关: AUTOCORR(x, {w})",
)
_registry.register_function(
f"max{w}", lambda x, w=w: _rolling_max(x, w), f"滚动最大值: MAX(x, {w})"
)
_registry.register_function(
f"min{w}", lambda x, w=w: _rolling_min(x, w), f"滚动最小值: MIN(x, {w})"
)
# 环比
_registry.register_function(
f"huanbi{w}", lambda x, w=w: _huanbi(x, w), f"环比: HUANBI(x, {w})"
)
# ==================== 因子公式解析与计算 ====================

113
factors.py
View File

@@ -1,113 +0,0 @@
"""
因子挖掘模块：支持规则因子和遗传编程因子
"""
import numpy as np
import pandas as pd
from typing import Callable, Dict, List, Optional
from abc import ABC, abstractmethod
class BaseFactor(ABC):
"""因子基类"""
def __init__(self, name: str):
self.name = name
@abstractmethod
def compute(self, data: pd.DataFrame) -> pd.Series:
"""计算因子值"""
pass
class RuleFactor(BaseFactor):
"""规则因子：基于固定规则"""
def __init__(self, name: str, compute_func: Callable[[pd.DataFrame], pd.Series]):
super().__init__(name)
self.compute_func = compute_func
def compute(self, data: pd.DataFrame) -> pd.Series:
return self.compute_func(data)
def create_trend_factor(data: pd.DataFrame) -> pd.Series:
"""趋势因子：价格趋势方向"""
trend = pd.Series(0, index=data.index)
trend[data['close'] > data['ema16']] = 1
trend[data['close'] < data['ema4']] = -1
return trend
def create_volatility_factor(data: pd.DataFrame) -> pd.Series:
"""波动率因子：滚动12期收益率标准差"""
return data['volatility']
def create_volume_price_factor(data: pd.DataFrame) -> pd.Series:
"""量价因子：成交量放大且价格上涨"""
volume_signal = (data['volume'] > data['volume_ma6']).astype(int)
return volume_signal * data['return']
def create_reversal_factor(data: pd.DataFrame) -> pd.Series:
"""反转因子：短期反转效应"""
return -data['return'].shift(1)
def create_momentum_factor(data: pd.DataFrame) -> pd.Series:
"""动量因子：基于MACD"""
return data['macd']
def create_rsi_factor(data: pd.DataFrame) -> pd.Series:
"""RSI因子：相对强弱指数（标准化）"""
return (data['rsi'] - 50) / 50 # 归一化到[-1, 1]
class FactorMiner:
"""因子挖掘器"""
def __init__(self):
self.factors: Dict[str, BaseFactor] = {}
def register_factor(self, factor: BaseFactor):
"""注册因子"""
self.factors[factor.name] = factor
def register_rule_factor(self, name: str, compute_func: Callable):
"""注册规则因子"""
factor = RuleFactor(name, compute_func)
self.register_factor(factor)
def compute_all_factors(self, data: pd.DataFrame) -> pd.DataFrame:
"""计算所有因子"""
factor_df = pd.DataFrame(index=data.index)
for name, factor in self.factors.items():
try:
factor_df[name] = factor.compute(data)
except Exception as e:
print(f"计算因子 {name} 时出错: {e}")
factor_df[name] = np.nan
return factor_df
def get_factor(self, name: str) -> Optional[BaseFactor]:
"""获取指定因子"""
return self.factors.get(name)
def create_default_factors() -> FactorMiner:
"""创建默认因子集合"""
miner = FactorMiner()
# 注册基础因子
miner.register_rule_factor('TREND', create_trend_factor)
miner.register_rule_factor('VOL', create_volatility_factor)
miner.register_rule_factor('VOLP', create_volume_price_factor)
miner.register_rule_factor('REV', create_reversal_factor)
miner.register_rule_factor('MOM', create_momentum_factor)
miner.register_rule_factor('RSI', create_rsi_factor)
return miner

109
signal.py
View File

@@ -1,109 +0,0 @@
"""
信号生成模块
"""
import numpy as np
import pandas as pd
from typing import Optional, TYPE_CHECKING
if TYPE_CHECKING:
from pandas import Series
def generate_signals(
score: 'pd.Series',
buy_threshold: float = 0.8,
sell_threshold: float = -0.8,
window: int = 30,
use_rolling_std: bool = True
) -> 'pd.Series':
"""
基于因子得分生成买卖信号
Parameters:
-----------
score : Series
因子综合得分
buy_threshold : float
买入阈值（标准差倍数）
sell_threshold : float
卖出阈值（标准差倍数）
window : int
滚动窗口（用于计算标准差）
use_rolling_std : bool
是否使用滚动标准差
Returns:
--------
Series: 交易信号（1=买入，-1=卖出，0=持有）
"""
signals = pd.Series(0, index=score.index)
if use_rolling_std:
# 使用滚动标准差
rolling_std = score.rolling(window).std()
buy_line = buy_threshold * rolling_std
sell_line = sell_threshold * rolling_std
else:
# 使用固定阈值
std = score.std()
buy_line = buy_threshold * std
sell_line = sell_threshold * std
# 生成原始信号
raw_signals = pd.Series(0, index=score.index)
raw_signals[score > buy_line] = 1 # 买入信号
raw_signals[score < sell_line] = -1 # 卖出信号
# 只在信号变化时产生交易信号，其他时候保持持仓状态
signals = pd.Series(0, index=score.index)
position = 0 # 当前持仓状态：0=空仓，1=满仓
for i in range(len(raw_signals)):
current_signal = raw_signals.iloc[i]
# 只在信号变化时产生交易
if current_signal == 1 and position == 0:
signals.iloc[i] = 1 # 买入
position = 1
elif current_signal == -1 and position == 1:
signals.iloc[i] = -1 # 卖出
position = 0
# 其他情况保持当前持仓状态，不产生交易信号
return signals.astype(int)
def generate_signals_with_position(
score: 'pd.Series',
buy_threshold: float = 0.8,
sell_threshold: float = -0.8,
window: int = 30,
current_position: int = 0
) -> 'pd.Series':
"""
生成信号（考虑当前持仓状态）
Parameters:
-----------
current_position : int
当前持仓：0=空仓，1=满仓
"""
raw_signals = generate_signals(score, buy_threshold, sell_threshold, window)
signals = pd.Series(0, index=score.index)
position = current_position
for i in range(len(raw_signals)):
signal = raw_signals.iloc[i]
if signal == 1 and position == 0:
signals.iloc[i] = 1 # 买入
position = 1
elif signal == -1 and position == 1:
signals.iloc[i] = -1 # 卖出
position = 0
else:
signals.iloc[i] = 0 # 持有
return signals

135
validation.py
View File

@@ -1,62 +1,43 @@
"""
因子检验模块：IC检验、分组回测、因子跨度回归
因子检验模块: IC检验、分组回测、因子跨度回归
"""
import numpy as np
import pandas as pd
from typing import Dict, List, Tuple
from statsmodels.regression.linear_model import OLS
def compute_ic(factor: pd.Series, forward_return: pd.Series, method: str = 'spearman') -> pd.Series:
"""
计算IC（信息系数）
Parameters:
-----------
factor : Series
因子值
forward_return : Series
未来收益率
method : str
相关性计算方法：'spearman' 或 'pearson'
"""
aligned = pd.concat([factor, forward_return], axis=1).dropna()
if len(aligned) < 10:
return pd.Series(dtype=float)
if method == 'spearman':
ic = aligned.iloc[:, 0].rank().corr(aligned.iloc[:, 1].rank())
else:
ic = aligned.iloc[:, 0].corr(aligned.iloc[:, 1])
return pd.Series([ic], index=[aligned.index[-1]])
def compute_rolling_ic(
factor: pd.Series,
forward_return: pd.Series,
window: int = 30,
method: str = 'spearman'
method: str = "spearman",
) -> pd.Series:
"""计算滚动IC（向量化优化）"""
"""计算滚动IC (向量化优化)"""
# 对齐数据
aligned = pd.concat([factor, forward_return], axis=1).dropna()
if len(aligned) < window:
return pd.Series(dtype=float, index=factor.index[window:])
aligned.columns = ['factor', 'return']
aligned.columns = ["factor", "return"]
if method == 'spearman':
if method == "spearman":
# 使用rank计算Spearman相关性
factor_rank = aligned['factor'].rank()
return_rank = aligned['return'].rank()
# 使用DataFrame的rolling().corr()方法
df_rank = pd.DataFrame({'factor': factor_rank, 'return': return_rank})
ic_series = df_rank['factor'].rolling(window, min_periods=window).corr(df_rank['return'])
# 这里是全局的 rank，理论上应该是按照 window 滚动排序
factor_rank = aligned["factor"].rank()
return_rank = aligned["return"].rank()
# 使用DataFrame的rolling().corr()方法, 该方法pandas优化过
df_rank = pd.DataFrame({"factor": factor_rank, "return": return_rank})
ic_series = (
df_rank["factor"]
.rolling(window, min_periods=window)
.corr(df_rank["return"])
)
else:
# Pearson相关性
df = pd.DataFrame({'factor': aligned['factor'], 'return': aligned['return']})
ic_series = df['factor'].rolling(window, min_periods=window).corr(df['return'])
df = pd.DataFrame({"factor": aligned["factor"], "return": aligned["return"]})
ic_series = df["factor"].rolling(window, min_periods=window).corr(df["return"])
return ic_series
@@ -65,7 +46,7 @@ def group_backtest(
factor: pd.Series,
forward_return: pd.Series,
n_groups: int = 3,
group_period: int = 180
group_period: int = 180,
) -> Dict:
"""
分组回测：将数据按因子值分组，计算各组收益
@@ -75,14 +56,9 @@ def group_backtest(
dict: 包含各组收益、H-L收益差、t统计量等
"""
aligned = pd.concat([factor, forward_return], axis=1).dropna()
aligned.columns = ['factor', 'return']
aligned.columns = ["factor", "return"]
results = {
'group_returns': [],
'h_l_return': [],
'h_l_tstat': [],
'periods': []
}
results = {"group_returns": [], "h_l_return": [], "h_l_tstat": [], "periods": []}
# 按月分组（每180个4h周期）- 使用更高效的步长
step = max(group_period // 2, 90) # 减少重叠计算
@@ -96,16 +72,13 @@ def group_backtest(
# 按因子值分组（向量化）
try:
period_data = period_data.copy()
period_data['group'] = pd.qcut(
period_data['factor'],
q=n_groups,
labels=False,
duplicates='drop'
period_data["group"] = pd.qcut(
period_data["factor"], q=n_groups, labels=False, duplicates="drop"
)
# 计算各组收益（向量化）
group_returns = period_data.groupby('group')['return'].mean()
results['group_returns'].append(group_returns)
group_returns = period_data.groupby("group")["return"].mean()
results["group_returns"].append(group_returns)
# H-L收益差
if len(group_returns) >= 2:
@@ -113,33 +86,31 @@ def group_backtest(
l_return = group_returns.iloc[0] # 低因子组
h_l_diff = h_return - l_return
results['h_l_return'].append(h_l_diff)
results['periods'].append(period_data.index[-1])
results["h_l_return"].append(h_l_diff)
results["periods"].append(period_data.index[-1])
except (ValueError, KeyError):
# qcut失败时跳过
continue
# 计算平均H-L收益和t统计量
if results['h_l_return']:
h_l_series = pd.Series(results['h_l_return'], index=results['periods'])
if results["h_l_return"]:
h_l_series = pd.Series(results["h_l_return"], index=results["periods"])
mean_h_l = h_l_series.mean()
std_h_l = h_l_series.std()
t_stat = mean_h_l / (std_h_l / np.sqrt(len(h_l_series)) + 1e-8)
results['mean_h_l_return'] = mean_h_l
results['mean_h_l_tstat'] = t_stat
results['h_l_series'] = h_l_series
results["mean_h_l_return"] = mean_h_l
results["mean_h_l_tstat"] = t_stat
results["h_l_series"] = h_l_series
else:
results['mean_h_l_return'] = 0
results['mean_h_l_tstat'] = 0
results["mean_h_l_return"] = 0
results["mean_h_l_tstat"] = 0
return results
def factor_span_regression(
factors: pd.DataFrame,
forward_return: pd.Series,
target_factor: str
factors: pd.DataFrame, forward_return: pd.Series, target_factor: str
) -> Dict:
"""
因子跨度回归：检验因子的边际解释力
@@ -160,14 +131,14 @@ def factor_span_regression(
# 对齐数据
data = pd.concat([factors, forward_return], axis=1).dropna()
if len(data) < 30:
return {'beta': 0, 'tstat': 0, 'r2': 0, 'r2_change': 0}
return {"beta": 0, "tstat": 0, "r2": 0, "r2_change": 0}
y = data.iloc[:, -1].values
X_all = data.iloc[:, :-1].values
# 全模型（包含目标因子）
try:
model_all = OLS(y, X_all).fit(cov_type='HAC', cov_kwds={'maxlags': 6})
model_all = OLS(y, X_all).fit(cov_type="HAC", cov_kwds={"maxlags": 6})
r2_all = model_all.rsquared
# 目标因子的系数和t统计量
@@ -177,28 +148,25 @@ def factor_span_regression(
# 不含目标因子的模型
X_without = np.delete(X_all, target_idx, axis=1)
model_without = OLS(y, X_without).fit(cov_type='HAC', cov_kwds={'maxlags': 6})
model_without = OLS(y, X_without).fit(cov_type="HAC", cov_kwds={"maxlags": 6})
r2_without = model_without.rsquared
r2_change = r2_all - r2_without
return {
'beta': beta,
'tstat': tstat,
'r2': r2_all,
'r2_change': r2_change,
'pvalue': model_all.pvalues[target_idx]
"beta": beta,
"tstat": tstat,
"r2": r2_all,
"r2_change": r2_change,
"pvalue": model_all.pvalues[target_idx],
}
except Exception as e:
print(f"回归分析出错: {e}")
return {'beta': 0, 'tstat': 0, 'r2': 0, 'r2_change': 0}
return {"beta": 0, "tstat": 0, "r2": 0, "r2_change": 0}
def validate_factor(
factor: pd.Series,
forward_return: pd.Series,
ic_window: int = 30,
n_groups: int = 3
factor: pd.Series, forward_return: pd.Series, ic_window: int = 30, n_groups: int = 3
) -> Dict:
"""
综合因子检验
@@ -216,11 +184,10 @@ def validate_factor(
group_result = group_backtest(factor, forward_return, n_groups=n_groups)
return {
'mean_ic': mean_ic,
'ic_ir': ic_ir,
'ic_series': rolling_ic,
'mean_h_l_return': group_result['mean_h_l_return'],
'mean_h_l_tstat': group_result['mean_h_l_tstat'],
'group_returns': group_result['group_returns']
"mean_ic": mean_ic,
"ic_ir": ic_ir,
"ic_series": rolling_ic,
"mean_h_l_return": group_result["mean_h_l_return"],
"mean_h_l_tstat": group_result["mean_h_l_tstat"],
"group_returns": group_result["group_returns"],
}