ビットコインの値段が大きく動いた時、反対の動きで利益をねらう作戦

#!/usr/bin/env python3
"""
CCI（Commodity Channel Index）逆張り戦略
CCIの極値で逆張りエントリー
"""
import pandas as pd
import numpy as np


def calculate_cci_signals(df: pd.DataFrame, cci_period: int = 20, 
                         oversold: float = -100, overbought: float = 100) -> pd.DataFrame:
    """
    CCIシグナルを生成
    
    Parameters:
    -----------
    df : pd.DataFrame
        OHLCVデータ
    cci_period : int
        CCI計算期間（デフォルト: 20）
    oversold : float
        売られすぎ水準（デフォルト: -100）
    overbought : float
        買われすぎ水準（デフォルト: 100）
    
    Returns:
    --------
    pd.DataFrame
        シグナル列が追加されたDataFrame
    """
    df = df.copy()
    
    # Typical Price
    df['tp'] = (df['high'] + df['low'] + df['close']) / 3
    
    # Simple Moving Average of TP
    df['tp_sma'] = df['tp'].rolling(window=cci_period).mean()
    
    # Mean Deviation
    df['md'] = df['tp'].rolling(window=cci_period).apply(
        lambda x: np.abs(x - x.mean()).mean()
    )
    
    # CCI計算
    df['cci'] = (df['tp'] - df['tp_sma']) / (0.015 * df['md'])
    
    # シグナル生成
    df['is_buy'] = (
        (df['cci'] < oversold) & 
        (df['cci'].shift(1) >= oversold)
    )
    
    df['is_sell'] = (
        (df['cci'] > overbought) & 
        (df['cci'].shift(1) <= overbought)
    )
    
    # NaN値をFalseに置換
    df['is_buy'] = df['is_buy'].fillna(False)
    df['is_sell'] = df['is_sell'].fillna(False)
    
    print(f"CCI逆張り: 期間={cci_period}, 売られすぎ={oversold}, 買われすぎ={overbought}")
    print(f"買いシグナル数: {df['is_buy'].sum()}")
    print(f"売りシグナル数: {df['is_sell'].sum()}")
    
    return df