暗号資産「ADA/USDT」で、「スイングインデックス」を使って値段のパワーを調べてみました！

"""
Swing Index Trading Signal Generator
Wilder's Swing Indexを使用した短期トレンド検出
"""
import pandas as pd
import numpy as np


def calculate_swing_index_signals(df: pd.DataFrame,
                                 limit_move: float = 3.0) -> pd.DataFrame:
    """
    Swing Index戦略のシグナル生成
    
    Parameters:
    -----------
    df : pd.DataFrame
        OHLCVデータ
    limit_move : float
        リミットムーブ値（デフォルト: 3.0）
    
    Returns:
    --------
    pd.DataFrame
        シグナルが追加されたDataFrame
    """
    df = df.copy()
    
    # 前日データ
    df['close_prev'] = df['close'].shift(1)
    df['open_prev'] = df['open'].shift(1)
    df['high_prev'] = df['high'].shift(1)
    df['low_prev'] = df['low'].shift(1)
    
    # 価格差分計算
    df['c_cp'] = df['close'] - df['close_prev']
    df['c_o'] = df['close'] - df['open']
    df['cp_op'] = df['close_prev'] - df['open_prev']
    
    # K値計算
    df['h_cp'] = np.abs(df['high'] - df['close_prev'])
    df['l_cp'] = np.abs(df['low'] - df['close_prev'])
    df['k'] = df[['h_cp', 'l_cp']].max(axis=1)
    
    # R値計算
    df['h_l'] = df['high'] - df['low']
    df['r1'] = df['h_l']
    df['r2'] = np.abs(df['high'] - df['close_prev'])
    df['r3'] = np.abs(df['low'] - df['close_prev'])
    df['r'] = df[['r1', 'r2', 'r3']].max(axis=1)
    
    # Swing Index計算
    df['si'] = 50 * (df['c_cp'] + 0.5 * df['c_o'] + 0.25 * df['cp_op']) * (df['k'] / (limit_move * 100)) / (df['r'] + 0.0001)
    
    # Accumulative Swing Index
    df['asi'] = df['si'].cumsum()
    
    # ASIのトレンド
    df['asi_ma'] = df['asi'].rolling(window=14).mean()
    df['asi_trend'] = df['asi'] - df['asi_ma']
    
    # シグナル生成
    df['asi_trend_prev'] = df['asi_trend'].shift(1)
    df['is_buy'] = (df['asi_trend'] > 0) & (df['asi_trend_prev'] <= 0) & df['asi_trend'].notna()
    df['is_sell'] = (df['asi_trend'] < 0) & (df['asi_trend_prev'] >= 0) & df['asi_trend'].notna()
    
    # 不要カラム削除
    drop_cols = ['close_prev', 'open_prev', 'high_prev', 'low_prev', 'c_cp', 'c_o', 'cp_op',
                 'h_cp', 'l_cp', 'k', 'h_l', 'r1', 'r2', 'r3', 'r', 'asi_trend_prev']
    df.drop(drop_cols, axis=1, inplace=True, errors='ignore')
    
    return df