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


# ============================================================
# 設定
# ============================================================

INPUT_FILE = "results/trade_base.csv"

OUTPUT_DIR = "results"

os.makedirs(
    OUTPUT_DIR,
    exist_ok=True
)


CANDIDATES = {
    "A_0.44_0.88": (
        0.44,
        0.88
    ),
    "B_0.44_1.00": (
        0.44,
        1.00
    ),
    "C_0.50_1.00": (
        0.50,
        1.00
    ),
    "D_0.60_1.00": (
        0.60,
        1.00
    ),
}


PERIOD_LABELS = [
    "Period1",
    "Period2",
    "Period3",
    "Period4"
]


HALF_LABELS = [
    "First Half",
    "Second Half"
]


# ============================================================
# 出力ファイル
# ============================================================

OVERALL_CSV = os.path.join(
    OUTPUT_DIR,
    "candidate_overall_analysis.csv"
)

PERIOD_CSV = os.path.join(
    OUTPUT_DIR,
    "candidate_period_analysis.csv"
)

HALF_CSV = os.path.join(
    OUTPUT_DIR,
    "candidate_half_analysis.csv"
)

YEAR_CSV = os.path.join(
    OUTPUT_DIR,
    "candidate_year_analysis.csv"
)

STABILITY_CSV = os.path.join(
    OUTPUT_DIR,
    "candidate_time_stability_summary.csv"
)

PERIOD_PF_GRAPH = os.path.join(
    OUTPUT_DIR,
    "candidate_period_pf.png"
)

HALF_PF_GRAPH = os.path.join(
    OUTPUT_DIR,
    "candidate_half_pf.png"
)

YEAR_PF_GRAPH = os.path.join(
    OUTPUT_DIR,
    "candidate_year_pf.png"
)

STABILITY_GRAPH = os.path.join(
    OUTPUT_DIR,
    "candidate_stability_pf.png"
)


# ============================================================
# データ読み込み
# ============================================================

df = pd.read_csv(
    INPUT_FILE
)


print(
    "========================================"
)

print(
    "RELATIVE BAND TIME STABILITY ANALYSIS"
)

print(
    "========================================"
)

print(
    f"\n読み込み件数: {len(df)}"
)


# ============================================================
# 必要列確認
# ============================================================

required_columns = [
    "entry_time",
    "entry_band_distance_pips",
    "entry_atr14_pips",
    "profit"
]


missing_columns = [
    column
    for column in required_columns
    if column not in df.columns
]


if missing_columns:

    raise ValueError(
        "必要な列がありません: "
        + ", ".join(
            missing_columns
        )
    )


# ============================================================
# 型変換
# ============================================================

df[
    "entry_time"
] = pd.to_datetime(
    df[
        "entry_time"
    ],
    errors="coerce"
)


numeric_columns = [
    "entry_band_distance_pips",
    "entry_atr14_pips",
    "profit"
]


for column in numeric_columns:

    df[
        column
    ] = pd.to_numeric(
        df[
            column
        ],
        errors="coerce"
    )


# ============================================================
# 分析対象データ
# ============================================================

analysis_df = df[
    df[
        "entry_time"
    ].notna()
    &
    df[
        "entry_band_distance_pips"
    ].notna()
    &
    df[
        "entry_atr14_pips"
    ].notna()
    &
    df[
        "profit"
    ].notna()
    &
    (
        df[
            "entry_atr14_pips"
        ]
        > 0
    )
].copy()


if analysis_df.empty:

    raise ValueError(
        "分析対象データがありません。"
    )


# ============================================================
# Relative Band Distance
# ============================================================

analysis_df[
    "relative_band_distance"
] = (
    analysis_df[
        "entry_band_distance_pips"
    ]
    /
    analysis_df[
        "entry_atr14_pips"
    ]
)


analysis_df[
    "relative_band_distance"
] = (
    analysis_df[
        "relative_band_distance"
    ]
    .replace(
        [
            np.inf,
            -np.inf
        ],
        np.nan
    )
)


analysis_df = analysis_df[
    analysis_df[
        "relative_band_distance"
    ].notna()
].copy()


# ============================================================
# 時系列順
# ============================================================

analysis_df = (
    analysis_df
    .sort_values(
        "entry_time"
    )
    .reset_index(
        drop=True
    )
)


print(
    f"分析対象件数: "
    f"{len(analysis_df)}"
)


print(
    f"分析開始日: "
    f"{analysis_df['entry_time'].min()}"
)

print(
    f"分析終了日: "
    f"{analysis_df['entry_time'].max()}"
)


# ============================================================
# Period1～4を作成
# ============================================================

period_numbers = pd.qcut(
    np.arange(
        len(
            analysis_df
        )
    ),
    q=4,
    labels=PERIOD_LABELS
)


analysis_df[
    "period"
] = period_numbers


# ============================================================
# 前半・後半を作成
# ============================================================

half_numbers = pd.qcut(
    np.arange(
        len(
            analysis_df
        )
    ),
    q=2,
    labels=HALF_LABELS
)


analysis_df[
    "half"
] = half_numbers


# ============================================================
# 年
# ============================================================

analysis_df[
    "year"
] = (
    analysis_df[
        "entry_time"
    ]
    .dt
    .year
)


# ============================================================
# 期間情報表示
# ============================================================

print(
    "\n========================================"
)

print(
    "4期間の日時範囲"
)

print(
    "========================================"
)


for period in PERIOD_LABELS:

    period_df = analysis_df[
        analysis_df[
            "period"
        ]
        == period
    ]

    print(
        f"\n{period}"
    )

    print(
        f"開始: "
        f"{period_df['entry_time'].min()}"
    )

    print(
        f"終了: "
        f"{period_df['entry_time'].max()}"
    )

    print(
        f"元データ件数: "
        f"{len(period_df)}"
    )


# ============================================================
# 指標計算関数
# ============================================================

def calculate_metrics(
    selected
):

    trade_count = len(
        selected
    )


    if trade_count == 0:

        return {
            "trade_count": 0,
            "win_count": 0,
            "loss_count": 0,
            "win_rate": np.nan,
            "total_profit": np.nan,
            "average_profit": np.nan,
            "median_profit": np.nan,
            "gross_profit": np.nan,
            "gross_loss": np.nan,
            "profit_factor": np.nan
        }


    win_count = (
        selected[
            "profit"
        ]
        > 0
    ).sum()


    loss_count = (
        selected[
            "profit"
        ]
        < 0
    ).sum()


    win_rate = (
        win_count
        /
        trade_count
        *
        100
    )


    total_profit = (
        selected[
            "profit"
        ]
        .sum()
    )


    average_profit = (
        selected[
            "profit"
        ]
        .mean()
    )


    median_profit = (
        selected[
            "profit"
        ]
        .median()
    )


    gross_profit = (
        selected.loc[
            selected[
                "profit"
            ]
            > 0,
            "profit"
        ]
        .sum()
    )


    gross_loss = -(
        selected.loc[
            selected[
                "profit"
            ]
            < 0,
            "profit"
        ]
        .sum()
    )


    if gross_loss > 0:

        profit_factor = (
            gross_profit
            /
            gross_loss
        )

    else:

        profit_factor = np.nan


    return {
        "trade_count": trade_count,
        "win_count": win_count,
        "loss_count": loss_count,
        "win_rate": win_rate,
        "total_profit": total_profit,
        "average_profit": average_profit,
        "median_profit": median_profit,
        "gross_profit": gross_profit,
        "gross_loss": gross_loss,
        "profit_factor": profit_factor
    }


# ============================================================
# 候補抽出関数
# ============================================================

def select_candidate(
    source_df,
    lower,
    upper
):

    selected = source_df[
        (
            source_df[
                "relative_band_distance"
            ]
            >= lower
        )
        &
        (
            source_df[
                "relative_band_distance"
            ]
            <= upper
        )
    ].copy()


    return selected


# ============================================================
# 全期間分析
# ============================================================

overall_results = []


for candidate_name, thresholds in CANDIDATES.items():

    lower = thresholds[0]
    upper = thresholds[1]


    selected = select_candidate(
        analysis_df,
        lower,
        upper
    )


    metrics = calculate_metrics(
        selected
    )


    row = {
        "candidate": candidate_name,
        "lower": lower,
        "upper": upper
    }


    row.update(
        metrics
    )


    overall_results.append(
        row
    )


overall_df = pd.DataFrame(
    overall_results
)


overall_df.to_csv(
    OVERALL_CSV,
    index=False,
    encoding="utf-8-sig"
)


print(
    "\n========================================"
)

print(
    "全期間分析"
)

print(
    "========================================"
)


print(
    overall_df[
        [
            "candidate",
            "lower",
            "upper",
            "trade_count",
            "win_rate",
            "total_profit",
            "average_profit",
            "median_profit",
            "profit_factor"
        ]
    ]
    .to_string(
        index=False
    )
)


# ============================================================
# 4期間分析
# ============================================================

period_results = []


for candidate_name, thresholds in CANDIDATES.items():

    lower = thresholds[0]
    upper = thresholds[1]


    for period in PERIOD_LABELS:

        source_df = analysis_df[
            analysis_df[
                "period"
            ]
            == period
        ]


        selected = select_candidate(
            source_df,
            lower,
            upper
        )


        metrics = calculate_metrics(
            selected
        )


        row = {
            "candidate": candidate_name,
            "lower": lower,
            "upper": upper,
            "period": period,
            "period_start": (
                source_df[
                    "entry_time"
                ].min()
            ),
            "period_end": (
                source_df[
                    "entry_time"
                ].max()
            )
        }


        row.update(
            metrics
        )


        period_results.append(
            row
        )


period_df = pd.DataFrame(
    period_results
)


period_df.to_csv(
    PERIOD_CSV,
    index=False,
    encoding="utf-8-sig"
)


print(
    "\n========================================"
)

print(
    "4期間分析"
)

print(
    "========================================"
)


print(
    period_df[
        [
            "candidate",
            "period",
            "trade_count",
            "win_rate",
            "total_profit",
            "average_profit",
            "median_profit",
            "profit_factor"
        ]
    ]
    .to_string(
        index=False
    )
)


# ============================================================
# 前半・後半分析
# ============================================================

half_results = []


for candidate_name, thresholds in CANDIDATES.items():

    lower = thresholds[0]
    upper = thresholds[1]


    for half in HALF_LABELS:

        source_df = analysis_df[
            analysis_df[
                "half"
            ]
            == half
        ]


        selected = select_candidate(
            source_df,
            lower,
            upper
        )


        metrics = calculate_metrics(
            selected
        )


        row = {
            "candidate": candidate_name,
            "lower": lower,
            "upper": upper,
            "half": half,
            "period_start": (
                source_df[
                    "entry_time"
                ].min()
            ),
            "period_end": (
                source_df[
                    "entry_time"
                ].max()
            )
        }


        row.update(
            metrics
        )


        half_results.append(
            row
        )


half_df = pd.DataFrame(
    half_results
)


half_df.to_csv(
    HALF_CSV,
    index=False,
    encoding="utf-8-sig"
)


print(
    "\n========================================"
)

print(
    "前半・後半分析"
)

print(
    "========================================"
)


print(
    half_df[
        [
            "candidate",
            "half",
            "trade_count",
            "win_rate",
            "total_profit",
            "average_profit",
            "median_profit",
            "profit_factor"
        ]
    ]
    .to_string(
        index=False
    )
)


# ============================================================
# 年別分析
# ============================================================

year_results = []


years = sorted(
    analysis_df[
        "year"
    ]
    .dropna()
    .unique()
)


for candidate_name, thresholds in CANDIDATES.items():

    lower = thresholds[0]
    upper = thresholds[1]


    for year in years:

        source_df = analysis_df[
            analysis_df[
                "year"
            ]
            == year
        ]


        selected = select_candidate(
            source_df,
            lower,
            upper
        )


        metrics = calculate_metrics(
            selected
        )


        row = {
            "candidate": candidate_name,
            "lower": lower,
            "upper": upper,
            "year": int(
                year
            )
        }


        row.update(
            metrics
        )


        year_results.append(
            row
        )


year_df = pd.DataFrame(
    year_results
)


year_df.to_csv(
    YEAR_CSV,
    index=False,
    encoding="utf-8-sig"
)


print(
    "\n========================================"
)

print(
    "年別PF"
)

print(
    "========================================"
)


year_pf_table = (
    year_df
    .pivot(
        index="year",
        columns="candidate",
        values="profit_factor"
    )
)


print(
    year_pf_table
    .round(
        3
    )
    .to_string()
)


# ============================================================
# 時間安定性サマリー
# ============================================================

stability_results = []


for candidate_name in CANDIDATES.keys():

    candidate_period_df = period_df[
        period_df[
            "candidate"
        ]
        == candidate_name
    ].copy()


    valid_pf = (
        candidate_period_df[
            "profit_factor"
        ]
        .dropna()
    )


    if len(
        valid_pf
    ) > 0:

        mean_period_pf = (
            valid_pf
            .mean()
        )

        median_period_pf = (
            valid_pf
            .median()
        )

        min_period_pf = (
            valid_pf
            .min()
        )

        max_period_pf = (
            valid_pf
            .max()
        )

        period_pf_std = (
            valid_pf
            .std(
                ddof=0
            )
        )

        periods_above_1 = (
            valid_pf
            > 1.0
        ).sum()

    else:

        mean_period_pf = np.nan
        median_period_pf = np.nan
        min_period_pf = np.nan
        max_period_pf = np.nan
        period_pf_std = np.nan
        periods_above_1 = 0


    candidate_year_df = year_df[
        year_df[
            "candidate"
        ]
        == candidate_name
    ].copy()


    valid_year_pf = (
        candidate_year_df[
            "profit_factor"
        ]
        .dropna()
    )


    years_with_trades = (
        candidate_year_df[
            "trade_count"
        ]
        > 0
    ).sum()


    years_above_1 = (
        candidate_year_df[
            "profit_factor"
        ]
        > 1.0
    ).sum()


    if years_with_trades > 0:

        year_pf_above_1_rate = (
            years_above_1
            /
            years_with_trades
            *
            100
        )

    else:

        year_pf_above_1_rate = np.nan


    if len(
        valid_year_pf
    ) > 0:

        min_year_pf = (
            valid_year_pf
            .min()
        )

        mean_year_pf = (
            valid_year_pf
            .mean()
        )

        year_pf_std = (
            valid_year_pf
            .std(
                ddof=0
            )
        )

    else:

        min_year_pf = np.nan
        mean_year_pf = np.nan
        year_pf_std = np.nan


    overall_row = overall_df[
        overall_df[
            "candidate"
        ]
        == candidate_name
    ].iloc[0]


    stability_results.append(
        {
            "candidate": candidate_name,
            "lower": overall_row[
                "lower"
            ],
            "upper": overall_row[
                "upper"
            ],
            "overall_trade_count": overall_row[
                "trade_count"
            ],
            "overall_total_profit": overall_row[
                "total_profit"
            ],
            "overall_profit_factor": overall_row[
                "profit_factor"
            ],
            "mean_period_pf": mean_period_pf,
            "median_period_pf": median_period_pf,
            "min_period_pf": min_period_pf,
            "max_period_pf": max_period_pf,
            "period_pf_std": period_pf_std,
            "periods_above_1": periods_above_1,
            "period_count": len(
                valid_pf
            ),
            "years_with_trades": years_with_trades,
            "years_above_1": years_above_1,
            "year_pf_above_1_rate": (
                year_pf_above_1_rate
            ),
            "mean_year_pf": mean_year_pf,
            "min_year_pf": min_year_pf,
            "year_pf_std": year_pf_std
        }
    )


stability_df = pd.DataFrame(
    stability_results
)


stability_df.to_csv(
    STABILITY_CSV,
    index=False,
    encoding="utf-8-sig"
)


print(
    "\n========================================"
)

print(
    "時間安定性サマリー"
)

print(
    "========================================"
)


print(
    stability_df[
        [
            "candidate",
            "overall_trade_count",
            "overall_total_profit",
            "overall_profit_factor",
            "mean_period_pf",
            "min_period_pf",
            "period_pf_std",
            "periods_above_1",
            "years_above_1",
            "years_with_trades",
            "year_pf_above_1_rate",
            "min_year_pf"
        ]
    ]
    .round(
        3
    )
    .to_string(
        index=False
    )
)


# ============================================================
# グラフ1
# 4期間PF
# ============================================================

plt.figure(
    figsize=(
        12,
        7
    )
)


for candidate_name in CANDIDATES.keys():

    candidate_df = period_df[
        period_df[
            "candidate"
        ]
        == candidate_name
    ].copy()


    candidate_df[
        "period"
    ] = pd.Categorical(
        candidate_df[
            "period"
        ],
        categories=PERIOD_LABELS,
        ordered=True
    )


    candidate_df = (
        candidate_df
        .sort_values(
            "period"
        )
    )


    plt.plot(
        candidate_df[
            "period"
        ],
        candidate_df[
            "profit_factor"
        ],
        marker="o",
        label=candidate_name
    )


plt.axhline(
    1.0,
    linestyle="--"
)


plt.title(
    "Profit Factor by Time Period"
)

plt.xlabel(
    "Time Period"
)

plt.ylabel(
    "Profit Factor"
)

plt.legend()

plt.grid()

plt.tight_layout()


plt.savefig(
    PERIOD_PF_GRAPH,
    dpi=150
)


plt.close()


# ============================================================
# グラフ2
# 前半・後半PF
# ============================================================

plt.figure(
    figsize=(
        11,
        7
    )
)


for candidate_name in CANDIDATES.keys():

    candidate_df = half_df[
        half_df[
            "candidate"
        ]
        == candidate_name
    ].copy()


    candidate_df[
        "half"
    ] = pd.Categorical(
        candidate_df[
            "half"
        ],
        categories=HALF_LABELS,
        ordered=True
    )


    candidate_df = (
        candidate_df
        .sort_values(
            "half"
        )
    )


    plt.plot(
        candidate_df[
            "half"
        ],
        candidate_df[
            "profit_factor"
        ],
        marker="o",
        label=candidate_name
    )


plt.axhline(
    1.0,
    linestyle="--"
)


plt.title(
    "Profit Factor: First Half vs Second Half"
)

plt.xlabel(
    "Period"
)

plt.ylabel(
    "Profit Factor"
)

plt.legend()

plt.grid()

plt.tight_layout()


plt.savefig(
    HALF_PF_GRAPH,
    dpi=150
)


plt.close()


# ============================================================
# グラフ3
# 年別PF
# ============================================================

plt.figure(
    figsize=(
        14,
        8
    )
)


for candidate_name in CANDIDATES.keys():

    candidate_df = year_df[
        year_df[
            "candidate"
        ]
        == candidate_name
    ].copy()


    candidate_df = (
        candidate_df
        .sort_values(
            "year"
        )
    )


    plt.plot(
        candidate_df[
            "year"
        ],
        candidate_df[
            "profit_factor"
        ],
        marker="o",
        label=candidate_name
    )


plt.axhline(
    1.0,
    linestyle="--"
)


plt.title(
    "Yearly Profit Factor by Candidate"
)

plt.xlabel(
    "Year"
)

plt.ylabel(
    "Profit Factor"
)

plt.legend()

plt.grid()

plt.tight_layout()


plt.savefig(
    YEAR_PF_GRAPH,
    dpi=150
)


plt.close()


# ============================================================
# グラフ4
# 平均期間PFと最低期間PF
# ============================================================

plot_df = (
    stability_df
    .set_index(
        "candidate"
    )[
        [
            "mean_period_pf",
            "min_period_pf"
        ]
    ]
)


ax = plot_df.plot(
    kind="bar",
    figsize=(
        12,
        7
    )
)


ax.axhline(
    1.0,
    linestyle="--"
)


ax.set_title(
    "Mean and Minimum Profit Factor "
    "Across Time Periods"
)

ax.set_xlabel(
    "Candidate"
)

ax.set_ylabel(
    "Profit Factor"
)


plt.xticks(
    rotation=0
)

plt.grid(
    axis="y"
)

plt.tight_layout()


plt.savefig(
    STABILITY_GRAPH,
    dpi=150
)


plt.close()


# ============================================================
# 各候補の期間別詳細
# ============================================================

print(
    "\n========================================"
)

print(
    "候補別詳細"
)

print(
    "========================================"
)


for candidate_name in CANDIDATES.keys():

    print(
        "\n----------------------------------------"
    )

    print(
        candidate_name
    )

    print(
        "----------------------------------------"
    )


    candidate_overall = overall_df[
        overall_df[
            "candidate"
        ]
        == candidate_name
    ]


    print(
        "\n全期間"
    )


    print(
        candidate_overall[
            [
                "trade_count",
                "win_rate",
                "total_profit",
                "average_profit",
                "median_profit",
                "profit_factor"
            ]
        ]
        .round(
            3
        )
        .to_string(
            index=False
        )
    )


    candidate_period = period_df[
        period_df[
            "candidate"
        ]
        == candidate_name
    ]


    print(
        "\n4期間"
    )


    print(
        candidate_period[
            [
                "period",
                "trade_count",
                "win_rate",
                "total_profit",
                "average_profit",
                "median_profit",
                "profit_factor"
            ]
        ]
        .round(
            3
        )
        .to_string(
            index=False
        )
    )


    candidate_half = half_df[
        half_df[
            "candidate"
        ]
        == candidate_name
    ]


    print(
        "\n前半・後半"
    )


    print(
        candidate_half[
            [
                "half",
                "trade_count",
                "win_rate",
                "total_profit",
                "average_profit",
                "median_profit",
                "profit_factor"
            ]
        ]
        .round(
            3
        )
        .to_string(
            index=False
        )
    )


# ============================================================
# 出力ファイル一覧
# ============================================================

print(
    "\n========================================"
)

print(
    "分析完了"
)

print(
    "========================================"
)


print(
    "\nCSV:"
)

print(
    OVERALL_CSV
)

print(
    PERIOD_CSV
)

print(
    HALF_CSV
)

print(
    YEAR_CSV
)

print(
    STABILITY_CSV
)


print(
    "\nグラフ:"
)

print(
    PERIOD_PF_GRAPH
)

print(
    HALF_PF_GRAPH
)

print(
    YEAR_PF_GRAPH
)

print(
    STABILITY_GRAPH
)