whatif_analysis.py

"""
What-if 分析：投資人各種假設情境的損益比較
===========================================
使用 0050 實際價格資料，計算：
1. 實際交易 vs 策略執行 的差異
2. 如果晚進場 / 早進場的影響
3. 如果嚴格按策略出場的損益
"""
import warnings
warnings.filterwarnings("ignore")
import sys, os
import numpy as np
import pandas as pd

sys.stdout.reconfigure(encoding="utf-8")

SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
OUTPUT_DIR = os.path.join(SCRIPT_DIR, "output")

def download_0050():
    import yfinance as yf
    df = yf.download("0050.TW", start="2023-09-01", end="2026-02-18", auto_adjust=True, progress=False)
    if isinstance(df.columns, pd.MultiIndex):
        df.columns = df.columns.get_level_values(0)
    df.index = pd.to_datetime(df.index).tz_localize(None)
    return df

def main():
    os.makedirs(OUTPUT_DIR, exist_ok=True)
    sep = "=" * 72

    print(f"\n{sep}")
    print("  What-If 情境分析：0050 實際價格驗證")
    print(sep)

    df = download_0050()
    print(f"  0050 日線資料: {len(df)} 筆 ({df.index[0].date()} ~ {df.index[-1].date()})")

    # ── 2023 Cycle Analysis ──
    print(f"\n{sep}")
    print("  【2023 年週期】投資人實際 vs 多種假設情境")
    print(sep)

    buy_date = pd.Timestamp("2023-10-24")
    buy_price = 123.65
    shares = 2000
    sell_date = pd.Timestamp("2024-01-02")
    sell_price = 134.70

    # Actual
    actual_pnl = (sell_price - buy_price) * shares
    actual_ret = (sell_price - buy_price) / buy_price
    print(f"\n  實際交易:")
    print(f"    買入: {buy_date.date()} @ {buy_price} x {shares}")
    print(f"    賣出: {sell_date.date()} @ {sell_price}")
    print(f"    報酬率: {actual_ret:.2%}")
    print(f"    獲利: TWD {actual_pnl:,.0f}")

    # What-if: Hold to +30%
    target_30 = buy_price * 1.30
    reached_30 = df[df.index > buy_date][df["Close"] >= target_30]
    if len(reached_30) > 0:
        date_30 = reached_30.index[0]
        price_30 = reached_30.iloc[0]["Close"]
        sell_70_shares = int(shares * 0.70 / 1000) * 1000  # 1000
        remain = shares - sell_70_shares
        pnl_30 = (price_30 - buy_price) * sell_70_shares
        print(f"\n  假設A：嚴格執行 +30% 賣 70%")
        print(f"    +30% 目標價: {target_30:.2f}")
        print(f"    達到日期: {date_30.date()} @ {price_30:.2f}")
        print(f"    賣出 {sell_70_shares} 股，獲利: TWD {pnl_30:,.0f}")
        print(f"    剩餘 {remain} 股繼續持有")

        # Then check +50%
        target_50 = buy_price * 1.50
        reached_50 = df[df.index > date_30][df["Close"] >= target_50]
        if len(reached_50) > 0:
            date_50 = reached_50.index[0]
            price_50 = reached_50.iloc[0]["Close"]
            pnl_50 = (price_50 - buy_price) * remain
            total_strat = pnl_30 + pnl_50
            print(f"    +50% 目標價: {target_50:.2f}")
            print(f"    達到日期: {date_50.date()} @ {price_50:.2f}")
            print(f"    賣出剩餘 {remain} 股，獲利: TWD {pnl_50:,.0f}")
            print(f"    策略總獲利: TWD {total_strat:,.0f}")
            print(f"    vs 實際獲利: TWD {actual_pnl:,.0f}")
            print(f"    差異: TWD {total_strat - actual_pnl:+,.0f} ({(total_strat/actual_pnl - 1):.0%} more)")
        else:
            # Check latest price for remaining
            latest = df.iloc[-1]["Close"]
            unrealized = (latest - buy_price) * remain
            total_strat = pnl_30 + unrealized
            print(f"    +50% 未達到，剩餘 {remain} 股持至今 @ {latest:.2f}")
            print(f"    未實現損益: TWD {unrealized:,.0f}")
            print(f"    策略總損益: TWD {total_strat:,.0f}")
    else:
        print(f"\n  假設A：0050 從未達到 +30% ({target_30:.2f})")

    # What-if: 3 tranches
    print(f"\n  假設B：嚴格 3 碼分批（100萬資金）")
    capital = 1_000_000
    tranche1_pct = 0.30
    tranche1_budget = capital * tranche1_pct
    t1_shares = int(tranche1_budget / buy_price / 1000) * 1000
    t1_cost = t1_shares * buy_price
    print(f"    第1碼: {buy_date.date()} @ {buy_price} x {t1_shares} = TWD {t1_cost:,.0f}")

    # 2nd tranche: next KD < 50 signal or -10% drawdown (approximate)
    # From backtest, 2nd signal was ~2 weeks later
    t2_date = pd.Timestamp("2023-11-03")  # approximate
    if t2_date in df.index:
        t2_price = df.loc[t2_date, "Close"]
    else:
        nearest = df.index[df.index >= t2_date][0]
        t2_price = df.loc[nearest, "Close"]
        t2_date = nearest
    t2_budget = capital * 0.30
    t2_shares = int(t2_budget / t2_price / 1000) * 1000
    t2_cost = t2_shares * t2_price
    print(f"    第2碼: {t2_date.date()} @ {t2_price:.2f} x {t2_shares} = TWD {t2_cost:,.0f}")

    t3_date = pd.Timestamp("2023-11-17")
    if t3_date in df.index:
        t3_price = df.loc[t3_date, "Close"]
    else:
        nearest = df.index[df.index >= t3_date][0]
        t3_price = df.loc[nearest, "Close"]
        t3_date = nearest
    t3_budget = capital * 0.40
    t3_shares = int(t3_budget / t3_price / 1000) * 1000
    t3_cost = t3_shares * t3_price
    print(f"    第3碼: {t3_date.date()} @ {t3_price:.2f} x {t3_shares} = TWD {t3_cost:,.0f}")

    total_shares = t1_shares + t2_shares + t3_shares
    avg_cost = (t1_cost + t2_cost + t3_cost) / total_shares
    print(f"    總股數: {total_shares:,}  平均成本: {avg_cost:.2f}")

    if len(reached_30) > 0:
        sell_70 = int(total_shares * 0.70 / 1000) * 1000
        pnl_b30 = (price_30 - avg_cost) * sell_70
        remain_b = total_shares - sell_70
        print(f"    +30% ({target_30:.2f})? → 均價 +30% = {avg_cost*1.30:.2f}")
        # Use avg cost for target
        target_30b = avg_cost * 1.30
        reached_30b = df[df.index > t3_date][df["Close"] >= target_30b]
        if len(reached_30b) > 0:
            d30b = reached_30b.index[0]
            p30b = reached_30b.iloc[0]["Close"]
            pnl_b30 = (p30b - avg_cost) * sell_70
            print(f"    均價+30%達到: {d30b.date()} @ {p30b:.2f}")
            print(f"    賣 70% ({sell_70}股): TWD {pnl_b30:,.0f}")

            target_50b = avg_cost * 1.50
            reached_50b = df[df.index > d30b][df["Close"] >= target_50b]
            if len(reached_50b) > 0:
                d50b = reached_50b.index[0]
                p50b = reached_50b.iloc[0]["Close"]
                pnl_b50 = (p50b - avg_cost) * remain_b
                total_b = pnl_b30 + pnl_b50
                print(f"    均價+50%達到: {d50b.date()} @ {p50b:.2f}")
                print(f"    賣 30% ({remain_b}股): TWD {pnl_b50:,.0f}")
                print(f"    3碼策略總獲利: TWD {total_b:,.0f}")
                print(f"    vs 實際獲利: TWD {actual_pnl:,.0f}")
                print(f"    差異: TWD {total_b - actual_pnl:+,.0f}")

    # What-if: Late entry (1 week, 2 weeks, 1 month late)
    print(f"\n  假設C：延遲進場的影響")
    delays = [("1週後", 7), ("2週後", 14), ("1個月後", 30)]
    for label, days in delays:
        late_date = buy_date + pd.Timedelta(days=days)
        closest = df.index[df.index >= late_date]
        if len(closest) > 0:
            late_d = closest[0]
            late_p = df.loc[late_d, "Close"]
            late_ret_at_sell = (sell_price - late_p) / late_p
            late_pnl = (sell_price - late_p) * shares
            # Also check what +30% would be
            late_target30 = late_p * 1.30
            print(f"    {label} ({late_d.date()}): 買入 @ {late_p:.2f}")
            print(f"      以 134.70 賣出: 報酬 {late_ret_at_sell:.2%}, 獲利 TWD {late_pnl:,.0f}")
            print(f"      +30% 目標: {late_target30:.2f}")

    # ── 2024 Cycle ──
    print(f"\n{sep}")
    print("  【2024 年週期】0050 進場分析")
    print(sep)

    buy2_date = pd.Timestamp("2024-10-07")
    buy2_price = 187.15
    buy2_shares = 2000

    # Price trajectory after buy
    after_buy2 = df[df.index >= buy2_date].copy()
    if len(after_buy2) > 0:
        peak_after = after_buy2["Close"].max()
        peak_date = after_buy2["Close"].idxmax()
        trough_after = after_buy2["Close"].min()
        trough_date = after_buy2["Close"].idxmin()
        latest = after_buy2.iloc[-1]
        current_ret = (latest["Close"] - buy2_price) / buy2_price

        print(f"  買入: {buy2_date.date()} @ {buy2_price} x {buy2_shares}")
        print(f"  期間最高: {peak_date.date()} @ {peak_after:.2f} (+{(peak_after/buy2_price-1):.2%})")
        print(f"  期間最低: {trough_date.date()} @ {trough_after:.2f} ({(trough_after/buy2_price-1):.2%})")
        print(f"  最新: {latest.name.date()} @ {latest['Close']:.2f} ({current_ret:+.2%})")
        print(f"  +30% 目標: {buy2_price*1.30:.2f}")

        # Check if +30% was reached
        target30_2 = buy2_price * 1.30
        reached30_2 = after_buy2[after_buy2["Close"] >= target30_2]
        if len(reached30_2) > 0:
            print(f"  +30% 達到: {reached30_2.index[0].date()} @ {reached30_2.iloc[0]['Close']:.2f}")
        else:
            print(f"  +30% 尚未達到")

    # ── 2025 Cycle (from 1.png) ──
    print(f"\n{sep}")
    print("  【2025 年週期】1.png KD 信號分析")
    print(sep)

    # The investor circled area around late Oct - mid Nov 2025
    # 0050 was around 60-63 (after stock split, prices are adjusted)
    signal_start = pd.Timestamp("2025-10-20")
    signal_end = pd.Timestamp("2025-11-15")
    signal_range = df[(df.index >= signal_start) & (df.index <= signal_end)]
    if len(signal_range) > 0:
        low_price = signal_range["Close"].min()
        low_date = signal_range["Close"].idxmin()
        print(f"  紅框區間: {signal_start.date()} ~ {signal_end.date()}")
        print(f"  區間最低: {low_date.date()} @ {low_price:.2f}")

        after_signal = df[df.index > signal_end]
        if len(after_signal) > 0:
            latest = after_signal.iloc[-1]
            gain = (latest["Close"] - low_price) / low_price
            print(f"  最新價格: {latest.name.date()} @ {latest['Close']:.2f}")
            print(f"  從低點漲幅: {gain:.2%}")
            print(f"  +30% 目標: {low_price * 1.30:.2f}")
            if gain >= 0.30:
                print(f"  ✓ 已達 +30% 停利門檻！")
            else:
                print(f"  距離 +30% 還差: {0.30 - gain:.2%}")

    # ── Summary table ──
    print(f"\n{sep}")
    print("  情境損益比較摘要")
    print(sep)

    results = []
    results.append({"情境": "實際交易（1碼+8.94%全出）", "獲利TWD": f"{actual_pnl:,.0f}", "報酬率": f"{actual_ret:.2%}"})
    if len(reached_30) > 0:
        results.append({"情境": "假設A: 1碼+嚴格停利", "獲利TWD": f"{total_strat:,.0f}", "報酬率": f"{total_strat/(buy_price*shares):.2%}"})
    results.append({"情境": "假設B: 3碼+嚴格停利", "獲利TWD": "見上方計算", "報酬率": "—"})

    rdf = pd.DataFrame(results)
    print(rdf.to_string(index=False))

    rdf.to_csv(os.path.join(OUTPUT_DIR, "whatif_summary.csv"), index=False, encoding="utf-8-sig")
    print(f"\n  已儲存: {os.path.join(OUTPUT_DIR, 'whatif_summary.csv')}")
    print(sep)


if __name__ == "__main__":
    main()