ランダムウォークへの挑戦⑦ | Pythonでエントリー曜日と決済曜日から零号機を分析する（2026/7/23）

import os
import pandas as pd
import matplotlib.pyplot as plt


# ==============================
# resultsフォルダを作成
# ==============================

os.makedirs(
    "results",
    exist_ok=True
)


# ==============================
# CSV読み込み
# ==============================

df = pd.read_csv(
    "/mnt/c/MT4/report.csv",
    encoding="utf-8-sig"
)


# ==============================
# エントリー行を抽出
# ==============================

# buy・sellがエントリー行
entries = df[
    df["取引種別"].isin(
        [
            "buy",
            "sell"
        ]
    )
].copy()


# エントリー分析に必要な列だけ残す
entries = entries[
    [
        "注文番号",
        "時間"
    ]
].copy()


# 「時間」を「エントリー時間」に変更
entries = entries.rename(
    columns={
        "時間": "エントリー時間"
    }
)


# ==============================
# 決済行を抽出
# ==============================

# closeが決済行
trades = df[
    df["取引種別"] == "close"
].copy()


# 「時間」を「決済時間」に変更
trades = trades.rename(
    columns={
        "時間": "決済時間"
    }
)


# ==============================
# エントリーと決済を結合
# ==============================

# 注文番号を使って1つの取引として結合
trades = trades.merge(
    entries,
    on="注文番号",
    how="left",
    validate="one_to_one"
)


# ==============================
# 日時型へ変換
# ==============================

date_columns = [
    "エントリー時間",
    "決済時間"
]


for col in date_columns:

    trades[col] = pd.to_datetime(
        trades[col],
        format="%Y.%m.%d %H:%M",
        errors="coerce"
    )


# 日時変換に失敗していないか確認
print(
    "\n===== 日時変換エラー確認 ====="
)

print(
    trades[
        [
            "エントリー時間",
            "決済時間"
        ]
    ].isna().sum()
)


# ==============================
# 年月列を作成
# ==============================

# 月別分析は決済した年月を使用
trades["年月"] = (
    trades["決済時間"]
    .dt.strftime("%Y-%m")
)


# ==============================
# 基本集計
# ==============================

total_trades = len(trades)


# 勝ち取引
wins = trades[
    trades["損益"] > 0
]


# 負け取引
losses = trades[
    trades["損益"] < 0
]


# 勝ち数・負け数
win_count = len(wins)
loss_count = len(losses)


# 勝率
win_rate = (
    win_count
    / total_trades
    * 100
)


# 総利益・総損失
gross_profit = (
    wins["損益"].sum()
)

gross_loss = (
    losses["損益"].sum()
)


# プロフィットファクタ
profit_factor = (
    gross_profit / abs(gross_loss)
    if gross_loss != 0
    else float("nan")
)


# 総損益
total_profit = (
    trades["損益"].sum()
)


# 平均利益
average_profit = (
    wins["損益"].mean()
)


# 平均損失
average_loss = (
    losses["損益"].mean()
)


# ==============================
# 最大ドローダウン
# ==============================

# その時点までの最高残高
trades["過去最高残高"] = (
    trades["残高"].cummax()
)


# 現在残高と過去最高残高との差
trades["ドローダウン"] = (
    trades["残高"]
    - trades["過去最高残高"]
)


# 最大ドローダウン
max_balance_drawdown = (
    trades["ドローダウン"].min()
)


# ==============================
# 基本分析結果を表示
# ==============================

print(
    "\n===== EA分析結果 ====="
)

print(
    f"総取引数: {total_trades}"
)

print(
    f"勝ち取引: {win_count}"
)

print(
    f"負け取引: {loss_count}"
)

print(
    f"勝率: {win_rate:.2f}%"
)

print(
    f"総損益: {total_profit:.2f}"
)

print(
    f"総利益: {gross_profit:.2f}"
)

print(
    f"総損失: {gross_loss:.2f}"
)

print(
    f"PF: {profit_factor:.2f}"
)

print(
    f"平均利益: {average_profit:.2f}"
)

print(
    f"平均損失: {average_loss:.2f}"
)

print(
    f"最大ドローダウン: "
    f"{max_balance_drawdown:.2f}"
)

print(
    f"最終残高: "
    f"{trades['残高'].iloc[-1]:.2f}"
)


# ==============================
# 月別分析
# ==============================

monthly_profit = (
    trades
    .groupby("年月")["損益"]
    .sum()
    .reset_index()
)


print(
    "\n===== 月別損益 ====="
)

print(
    monthly_profit
)


# ==============================
# 月別損益グラフ
# ==============================

plt.figure(
    figsize=(14, 6)
)

plt.bar(
    monthly_profit["年月"],
    monthly_profit["損益"]
)

plt.title(
    "Monthly Profit"
)

plt.xlabel(
    "Month"
)

plt.ylabel(
    "Profit"
)

plt.xticks(
    rotation=90
)

plt.axhline(
    0,
    linewidth=1
)

plt.tight_layout()

plt.savefig(
    "results/monthly_profit.png",
    dpi=150
)

plt.close()


# ==============================
# エントリー時間帯・決済時間帯を作成
# ==============================

trades["エントリー時間帯"] = (
    trades["エントリー時間"]
    .dt.hour
)


trades["決済時間帯"] = (
    trades["決済時間"]
    .dt.hour
)


# ==============================
# 時間帯別分析用の関数
# ==============================

def analyze_by_hour(
    data,
    hour_column
):

    analysis = (
        data
        .groupby(hour_column)
        .agg(
            取引回数=(
                "損益",
                "count"
            ),
            勝ち数=(
                "損益",
                lambda x: (x > 0).sum()
            ),
            負け数=(
                "損益",
                lambda x: (x < 0).sum()
            ),
            総損益=(
                "損益",
                "sum"
            ),
            平均損益=(
                "損益",
                "mean"
            ),
            総利益=(
                "損益",
                lambda x: x[x > 0].sum()
            ),
            総損失=(
                "損益",
                lambda x: x[x < 0].sum()
            )
        )
    )


    # 勝率
    analysis["勝率"] = (
        analysis["勝ち数"]
        / analysis["取引回数"]
        * 100
    )


    # PF
    analysis["PF"] = (
        analysis["総利益"]
        / analysis["総損失"]
        .abs()
        .replace(
            0,
            float("nan")
        )
    )


    # 必要な列だけ残す
    analysis = analysis[
        [
            "取引回数",
            "勝ち数",
            "負け数",
            "勝率",
            "総損益",
            "平均損益",
            "PF"
        ]
    ]


    return analysis


# ==============================
# エントリー時間帯別分析
# ==============================

entry_hour_analysis = analyze_by_hour(
    trades,
    "エントリー時間帯"
)


print(
    "\n===== エントリー時間帯別分析 ====="
)

print(
    entry_hour_analysis.round(2)
)


entry_hour_analysis.to_csv(
    "results/entry_hour_analysis.csv",
    encoding="utf-8-sig"
)


# ==============================
# 決済時間帯別分析
# ==============================

exit_hour_analysis = analyze_by_hour(
    trades,
    "決済時間帯"
)


print(
    "\n===== 決済時間帯別分析 ====="
)

print(
    exit_hour_analysis.round(2)
)


exit_hour_analysis.to_csv(
    "results/exit_hour_analysis.csv",
    encoding="utf-8-sig"
)


# ==============================
# エントリー時間帯別グラフ
# ==============================

plt.figure(
    figsize=(12, 6)
)

plt.bar(
    entry_hour_analysis.index,
    entry_hour_analysis["総損益"]
)

plt.title(
    "Profit by Entry Hour"
)

plt.xlabel(
    "Entry Hour"
)

plt.ylabel(
    "Profit"
)

plt.xticks(
    range(24)
)

plt.axhline(
    0,
    linewidth=1
)

plt.tight_layout()

plt.savefig(
    "results/entry_hour_profit.png",
    dpi=150
)

plt.close()


# ==============================
# 決済時間帯別グラフ
# ==============================

plt.figure(
    figsize=(12, 6)
)

plt.bar(
    exit_hour_analysis.index,
    exit_hour_analysis["総損益"]
)

plt.title(
    "Profit by Exit Hour"
)

plt.xlabel(
    "Exit Hour"
)

plt.ylabel(
    "Profit"
)

plt.xticks(
    range(24)
)

plt.axhline(
    0,
    linewidth=1
)

plt.tight_layout()

plt.savefig(
    "results/exit_hour_profit.png",
    dpi=150
)

plt.close()


# ==============================
# エントリー曜日・決済曜日を作成
# ==============================

# Monday=0 ～ Sunday=6
trades["エントリー曜日番号"] = (
    trades["エントリー時間"]
    .dt.dayofweek
)

trades["決済曜日番号"] = (
    trades["決済時間"]
    .dt.dayofweek
)


# 曜日番号と英語曜日名の対応表
weekday_names = {
    0: "Monday",
    1: "Tuesday",
    2: "Wednesday",
    3: "Thursday",
    4: "Friday",
    5: "Saturday",
    6: "Sunday"
}


# 曜日番号を曜日名へ変換
trades["エントリー曜日"] = (
    trades["エントリー曜日番号"]
    .map(weekday_names)
)

trades["決済曜日"] = (
    trades["決済曜日番号"]
    .map(weekday_names)
)


# ==============================
# 曜日別分析用の関数
# ==============================

def analyze_by_weekday(
    data,
    weekday_number_column,
    weekday_column
):

    # 曜日ごとに集計
    analysis = (
        data
        .groupby(
            [
                weekday_number_column,
                weekday_column
            ]
        )
        .agg(
            取引回数=(
                "損益",
                "count"
            ),
            勝ち数=(
                "損益",
                lambda x: (x > 0).sum()
            ),
            負け数=(
                "損益",
                lambda x: (x < 0).sum()
            ),
            総損益=(
                "損益",
                "sum"
            ),
            平均損益=(
                "損益",
                "mean"
            ),
            総利益=(
                "損益",
                lambda x: x[x > 0].sum()
            ),
            総損失=(
                "損益",
                lambda x: x[x < 0].sum()
            )
        )
    )


    # 勝率
    analysis["勝率"] = (
        analysis["勝ち数"]
        / analysis["取引回数"]
        * 100
    )


    # PF
    analysis["PF"] = (
        analysis["総利益"]
        / analysis["総損失"]
        .abs()
        .replace(
            0,
            float("nan")
        )
    )


    # groupbyでindexになった曜日を通常の列へ戻す
    analysis = (
        analysis
        .reset_index()
    )


    # Monday → Sunday の順番に並べる
    analysis = (
        analysis
        .sort_values(
            weekday_number_column
        )
    )


    # 必要な列だけ残す
    analysis = analysis[
        [
            weekday_number_column,
            weekday_column,
            "取引回数",
            "勝ち数",
            "負け数",
            "勝率",
            "総損益",
            "平均損益",
            "PF"
        ]
    ]


    return analysis


# ==============================
# エントリー曜日別分析
# ==============================

entry_weekday_analysis = analyze_by_weekday(
    trades,
    "エントリー曜日番号",
    "エントリー曜日"
)


print(
    "\n===== エントリー曜日別分析 ====="
)

print(
    entry_weekday_analysis[
        [
            "エントリー曜日",
            "取引回数",
            "勝ち数",
            "負け数",
            "勝率",
            "総損益",
            "平均損益",
            "PF"
        ]
    ].round(2)
)


# ==============================
# エントリー曜日別CSV
# ==============================

entry_weekday_analysis[
    [
        "エントリー曜日",
        "取引回数",
        "勝ち数",
        "負け数",
        "勝率",
        "総損益",
        "平均損益",
        "PF"
    ]
].to_csv(
    "results/entry_weekday_analysis.csv",
    index=False,
    encoding="utf-8-sig"
)


# ==============================
# 決済曜日別分析
# ==============================

exit_weekday_analysis = analyze_by_weekday(
    trades,
    "決済曜日番号",
    "決済曜日"
)


print(
    "\n===== 決済曜日別分析 ====="
)

print(
    exit_weekday_analysis[
        [
            "決済曜日",
            "取引回数",
            "勝ち数",
            "負け数",
            "勝率",
            "総損益",
            "平均損益",
            "PF"
        ]
    ].round(2)
)


# ==============================
# 決済曜日別CSV
# ==============================

exit_weekday_analysis[
    [
        "決済曜日",
        "取引回数",
        "勝ち数",
        "負け数",
        "勝率",
        "総損益",
        "平均損益",
        "PF"
    ]
].to_csv(
    "results/exit_weekday_analysis.csv",
    index=False,
    encoding="utf-8-sig"
)


# ==============================
# エントリー曜日別損益グラフ
# ==============================

plt.figure(
    figsize=(10, 6)
)

plt.bar(
    entry_weekday_analysis[
        "エントリー曜日"
    ],
    entry_weekday_analysis[
        "総損益"
    ]
)

plt.title(
    "Profit by Entry Weekday"
)

plt.xlabel(
    "Entry Weekday"
)

plt.ylabel(
    "Profit"
)

plt.axhline(
    0,
    linewidth=1
)

plt.tight_layout()

plt.savefig(
    "results/entry_weekday_profit.png",
    dpi=150
)

plt.close()


print(
    "エントリー曜日別損益グラフを "
    "results/entry_weekday_profit.png "
    "として保存しました。"
)


# ==============================
# 決済曜日別損益グラフ
# ==============================

plt.figure(
    figsize=(10, 6)
)

plt.bar(
    exit_weekday_analysis[
        "決済曜日"
    ],
    exit_weekday_analysis[
        "総損益"
    ]
)

plt.title(
    "Profit by Exit Weekday"
)

plt.xlabel(
    "Exit Weekday"
)

plt.ylabel(
    "Profit"
)

plt.axhline(
    0,
    linewidth=1
)

plt.tight_layout()

plt.savefig(
    "results/exit_weekday_profit.png",
    dpi=150
)

plt.close()


print(
    "決済曜日別損益グラフを "
    "results/exit_weekday_profit.png "
    "として保存しました。"
)