ビットコイン取引のヒント！「人気の価格帯」を見つける作戦

"""
Market Profile Value Area Trading Signal
マーケットプロファイル理論に基づくバリューエリア戦略
"""
import pandas as pd
import numpy as np


def calculate_market_profile_value_signals(df: pd.DataFrame,
                                          profile_period: int = 20,
                                          value_area_pct: float = 0.70,
                                          poc_weight: float = 1.5) -> pd.DataFrame:
    """
    Market Profile Value Area戦略のシグナル生成
    
    Parameters:
    -----------
    df : pd.DataFrame
        OHLCVデータ
    profile_period : int
        プロファイル構築期間（デフォルト: 20）
    value_area_pct : float
        バリューエリアの割合（デフォルト: 0.70）
    poc_weight : float
        Point of Controlの重み（デフォルト: 1.5）
    
    Returns:
    --------
    pd.DataFrame
        シグナルが追加されたDataFrame
    """
    df = df.copy()
    
    # TPO (Time Price Opportunity) の近似
    df['typical_price'] = (df['high'] + df['low'] + df['close']) / 3
    df['price_range'] = df['high'] - df['low']
    
    # 価格レベルごとのボリューム分布を計算
    df['vah'] = pd.Series(index=df.index, dtype=float)  # Value Area High
    df['val'] = pd.Series(index=df.index, dtype=float)  # Value Area Low
    df['poc'] = pd.Series(index=df.index, dtype=float)  # Point of Control
    df['vwap'] = pd.Series(index=df.index, dtype=float) # Volume Weighted Average Price
    
    for i in range(profile_period - 1, len(df)):
        # 期間内のデータ
        period_df = df.iloc[i - profile_period + 1:i + 1]
        
        # VWAP計算
        vwap = (period_df['typical_price'] * period_df['volume']).sum() / period_df['volume'].sum()
        df.loc[df.index[i], 'vwap'] = vwap
        
        # 価格帯を30分割してヒストグラム作成
        price_min = period_df['low'].min()
        price_max = period_df['high'].max()
        price_bins = np.linspace(price_min, price_max, 30)
        
        # 各価格帯のボリューム計算
        volume_profile = np.zeros(len(price_bins) - 1)
        for j in range(len(period_df)):
            row = period_df.iloc[j]
            # 各ローソクが通過する価格帯にボリュームを配分
            for k in range(len(price_bins) - 1):
                if row['low'] <= price_bins[k+1] and row['high'] >= price_bins[k]:
                    # 価格帯を通過する割合でボリュームを配分
                    overlap_low = max(row['low'], price_bins[k])
                    overlap_high = min(row['high'], price_bins[k+1])
                    overlap_ratio = (overlap_high - overlap_low) / (row['high'] - row['low'] + 0.0001)
                    volume_profile[k] += row['volume'] * overlap_ratio
        
        # Point of Control (最大ボリューム価格)
        poc_idx = np.argmax(volume_profile)
        poc_price = (price_bins[poc_idx] + price_bins[poc_idx + 1]) / 2
        df.loc[df.index[i], 'poc'] = poc_price
        
        # Value Area計算
        total_volume = volume_profile.sum()
        target_volume = total_volume * value_area_pct
        
        # POCから上下に拡張してValue Areaを決定
        accumulated_volume = volume_profile[poc_idx]
        va_low_idx = poc_idx
        va_high_idx = poc_idx
        
        while accumulated_volume < target_volume:
            # 上下のボリュームを比較して大きい方を追加
            up_vol = volume_profile[va_high_idx + 1] if va_high_idx < len(volume_profile) - 1 else 0
            down_vol = volume_profile[va_low_idx - 1] if va_low_idx > 0 else 0
            
            if up_vol > down_vol and va_high_idx < len(volume_profile) - 1:
                va_high_idx += 1
                accumulated_volume += up_vol
            elif va_low_idx > 0:
                va_low_idx -= 1
                accumulated_volume += down_vol
            else:
                break
        
        df.loc[df.index[i], 'val'] = (price_bins[va_low_idx] + price_bins[va_low_idx + 1]) / 2
        df.loc[df.index[i], 'vah'] = (price_bins[va_high_idx] + price_bins[va_high_idx + 1]) / 2
    
    # Initial Balance (最初の1時間の高値・安値)
    df['ib_high'] = df['high'].rolling(window=4).max()  # 15分足なら4本で1時間
    df['ib_low'] = df['low'].rolling(window=4).min()
    
        # シグナル初期化
    df['signal'] = 0
    df['is_buy'] = False
    df['is_sell'] = False
    
    # 現在のポジション状態を追跡
    position = 0
    
    for i in range(profile_period, len(df)):
        # 買いシグナル（Value Area Lowからの反発）
        if position <= 0 and df['close'].iloc[i] <= df['val'].iloc[i] and df['close'].iloc[i] > df['val'].iloc[i-1]:
            df.loc[df.index[i], 'is_buy'] = True
            df.loc[df.index[i], 'signal'] = 1
            position = 1
            
        # 売りシグナル（Value Area Highからの反落）  
        elif position >= 0 and df['close'].iloc[i] >= df['vah'].iloc[i] and df['close'].iloc[i] < df['vah'].iloc[i-1]:
            df.loc[df.index[i], 'is_sell'] = True
            df.loc[df.index[i], 'signal'] = -1
            position = -1
        else:
            # ポジション維持
            df.loc[df.index[i], 'signal'] = position
    
    # NaN値を0で埋める
    df['signal'] = df['signal'].fillna(0)
    
    # エグジットシグナル
    df['exit_signal'] = 0
    
    # Value Area外でのエグジット
    df.loc[(df['signal'] == 1) & (df['close'] > df['vah']), 'exit_signal'] = -1
    df.loc[(df['signal'] == -1) & (df['close'] < df['val']), 'exit_signal'] = 1
    
    return df


def get_strategy_name() -> str:
    """戦略名を返す"""
    return "Market Profile Value"


def get_strategy_description() -> str:
    """戦略の説明を返す"""
    return "マーケットプロファイルのバリューエリアとPOCを使用した戦略"