""" 资金流因子 - 使用Polars实现 包含主力资金流、散户资金流等相关因子计算 """ import polars as pl import numpy as np from typing import Dict, List, Optional, Any from operator_framework import StockWiseOperator, OperatorConfig class MoneyFlowIntensityOperator(StockWiseOperator): """主力资金流强度算子""" def __init__(self): config = OperatorConfig( name="money_flow_intensity", description="主力资金流强度", required_columns=['buy_lg_vol', 'buy_elg_vol', 'sell_lg_vol', 'sell_elg_vol', 'vol'], output_columns=['flow_lg_elg_intensity'], parameters={} ) super().__init__(config) def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算主力资金流强度""" epsilon = 1e-8 # 计算大单+超大单净买入量 lg_elg_net_buy_vol = ( pl.col('buy_lg_vol') + pl.col('buy_elg_vol') - pl.col('sell_lg_vol') - pl.col('sell_elg_vol') ) # 计算资金流强度 flow_intensity = lg_elg_net_buy_vol / (pl.col('vol') + epsilon) return stock_df.with_columns(flow_intensity.alias('flow_lg_elg_intensity')) class FlowDivergenceRatioOperator(StockWiseOperator): """散户与主力背离度算子""" def __init__(self): config = OperatorConfig( name="flow_divergence_ratio", description="散户与主力背离度比率", required_columns=['buy_sm_vol', 'sell_sm_vol', 'buy_lg_vol', 'buy_elg_vol', 'sell_lg_vol', 'sell_elg_vol'], output_columns=['flow_divergence_ratio'], parameters={} ) super().__init__(config) def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算散户与主力背离度比率""" epsilon = 1e-8 # 计算小单净买入量 sm_net_buy_vol = pl.col('buy_sm_vol') - pl.col('sell_sm_vol') # 计算大单+超大单净买入量 lg_elg_net_buy_vol = ( pl.col('buy_lg_vol') + pl.col('buy_elg_vol') - pl.col('sell_lg_vol') - pl.col('sell_elg_vol') ) # 计算背离度比率，处理分母为0的情况 divergence_ratio = sm_net_buy_vol / ( lg_elg_net_buy_vol + pl.when(lg_elg_net_buy_vol == 0).then(epsilon).otherwise(0) + epsilon ) return stock_df.with_columns(divergence_ratio.alias('flow_divergence_ratio')) class FlowStructureChangeOperator(StockWiseOperator): """资金流结构变动算子""" def __init__(self): config = OperatorConfig( name="flow_structure_change", description="资金流结构变动", required_columns=['buy_sm_vol', 'buy_lg_vol', 'buy_elg_vol'], output_columns=['flow_struct_buy_change'], parameters={} ) super().__init__(config) def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算资金流结构变动""" epsilon = 1e-8 # 计算总买入量 total_buy_vol = pl.col('buy_sm_vol') + pl.col('buy_lg_vol') + pl.col('buy_elg_vol') # 计算大单+超大单买入占比 lg_elg_buy_prop = (pl.col('buy_lg_vol') + pl.col('buy_elg_vol')) / (total_buy_vol + epsilon) # 计算1日变化 struct_change = lg_elg_buy_prop.diff() return stock_df.with_columns(struct_change.alias('flow_struct_buy_change')) class FlowAccelerationOperator(StockWiseOperator): """资金流加速度算子""" def __init__(self): config = OperatorConfig( name="flow_acceleration", description="资金流加速度", required_columns=['buy_lg_vol', 'buy_elg_vol', 'sell_lg_vol', 'sell_elg_vol'], output_columns=['flow_lg_elg_accel'], parameters={} ) super().__init__(config) def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算资金流加速度""" # 计算大单+超大单净买入量 lg_elg_net_buy_vol = ( pl.col('buy_lg_vol') + pl.col('buy_elg_vol') - pl.col('sell_lg_vol') - pl.col('sell_elg_vol') ) # 计算一阶变化 first_diff = lg_elg_net_buy_vol.diff() # 计算二阶变化（加速度） acceleration = first_diff.diff() return stock_df.with_columns(acceleration.alias('flow_lg_elg_accel')) class LargeFlowMomentumCorrelationOperator(StockWiseOperator): """大单资金流与价格动量相关性算子""" def __init__(self, n: int = 20, m: int = 60): config = OperatorConfig( name=f"lg_flow_mom_corr_{n}_{m}", description=f"{n}日大单资金流与{m}日价格动量相关性", required_columns=['buy_lg_vol', 'buy_elg_vol', 'sell_lg_vol', 'sell_elg_vol', 'close', 'vol'], output_columns=[f'lg_flow_mom_corr_{n}_{m}'], parameters={'n': n, 'm': m} ) super().__init__(config) self.n = n self.m = m def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算大单资金流与价格动量相关性""" # 计算大单净额 net_lg_flow_val = ( (pl.col('buy_lg_vol') + pl.col('buy_elg_vol') - pl.col('sell_lg_vol') - pl.col('sell_elg_vol')) * pl.col('close') ) # 计算滚动净大单流 rolling_net_lg_flow = net_lg_flow_val.rolling_sum(window=self.n) # 计算价格动量 price_mom = pl.col('close').pct_change(self.n) # 计算相关性 # Polars的rolling_corr需要两个表达式 correlation = rolling_net_lg_flow.rolling_corr(price_mom, window=self.m) return stock_df.with_columns( correlation.alias(f'lg_flow_mom_corr_{self.n}_{self.m}') ) class LargeBuyConsolidationOperator(StockWiseOperator): """大单买入盘整期算子""" def __init__(self, n: int = 20, vol_quantile: float = 0.2): config = OperatorConfig( name=f"lg_buy_consolidation_{n}", description=f"{n}日大单买入盘整期", required_columns=['close', 'buy_lg_vol', 'buy_elg_vol', 'sell_lg_vol', 'sell_elg_vol', 'vol'], output_columns=[f'lg_buy_consolidation_{n}'], parameters={'n': n, 'vol_quantile': vol_quantile} ) super().__init__(config) self.n = n self.vol_quantile = vol_quantile def apply_stock(self, stock_df: pl.DataFrame, **kwargs) -> pl.DataFrame: """计算大单买入盘整期""" epsilon = 1e-8 # 计算收盘价滚动标准差 rolling_std = pl.col('close').rolling_std(window=self.n) # 计算大单净流比率 net_lg_flow_ratio = ( (pl.col('buy_lg_vol') + pl.col('buy_elg_vol') - pl.col('sell_lg_vol') - pl.col('sell_elg_vol')) / (pl.col('vol') + epsilon) ) # 计算滚动均值 rolling_mean_ratio = net_lg_flow_ratio.rolling_mean(window=self.n) # 计算低波动阈值 # 这里需要按日期分组计算分位数，比较复杂，简化处理 # 在实际使用时，可能需要DateWiseOperator来处理横截面分位数 return stock_df.with_columns( rolling_mean_ratio.alias(f'lg_buy_consolidation_{self.n}') ) # 资金流因子集合 MONEY_FLOW_OPERATORS = [ MoneyFlowIntensityOperator(), FlowDivergenceRatioOperator(), FlowStructureChangeOperator(), FlowAccelerationOperator(), LargeFlowMomentumCorrelationOperator(), LargeBuyConsolidationOperator(), ] def apply_money_flow_factors(df: pl.DataFrame, operators: List = None) -> pl.DataFrame: """ 应用所有资金流因子 Args: df: 输入的Polars DataFrame operators: 要应用的算子列表，如果为None则使用默认列表 Returns: 添加了资金流因子的DataFrame """ if operators is None: operators = MONEY_FLOW_OPERATORS result_df = df for operator in operators: result_df = operator(result_df) return result_df