Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,40 +1,204 @@
-import altair as alt
-import numpy as np
-import pandas as pd
 import streamlit as st
+import yfinance as yf
+import pandas as pd
+import numpy as np
+import plotly.express as px
+
+# --- 1. 页面配置 ---
+st.set_page_config(page_title="Permanent Portfolio Backtester", layout="wide")
+st.title("🧩 永久组合 (Permanent Portfolio) 阈值策略回测")
+
+# --- 2. 数据获取与缓存 ---
+# SHV (短债) 是2007年才有的，所以回测通常从2007年开始
+@st.cache_data
+def get_data():
+    tickers = ['SPY', 'TLT', 'GLD', 'SHV']
+    # 下载数据，使用 Adj Close 以包含分红
+    data = yf.download(tickers, start="2007-01-01")['Adj Close']
+    data = data.dropna() # 确保所有ETF都有数据的日期才开始
+    return data
+
+try:
+    with st.spinner('正在从 Yahoo Finance 获取数据...'):
+        df = get_data()
+    st.success(f"数据获取成功! 范围: {df.index[0].date()} 到 {df.index[-1].date()}")
+except Exception as e:
+    st.error(f"数据获取失败: {e}")
+    st.stop()
+
+# --- 3. 回测逻辑核心函数 ---
+def run_backtest(data, threshold=0.05, check_freq='Daily'):
+    """
+    threshold: 偏离阈值 (例如 0.05 代表 ±5%)
+    check_freq: 检测频率 ('Daily', 'Monthly', 'Quarterly', 'Annually')
+    """
+    
+    # 初始资金
+    initial_cash = 10000.0
+    
+    # 资产列表
+    assets = ['SPY', 'TLT', 'GLD', 'SHV']
+    
+    # 初始化持仓 (股数)
+    holdings = {asset: 0.0 for asset in assets}
+    
+    # 初始化记录
+    portfolio_history = []
+    rebalance_log = []
+    
+    # 初始建仓：Day 1 强制 4等分
+    first_prices = data.iloc[0]
+    for asset in assets:
+        holdings[asset] = (initial_cash * 0.25) / first_prices[asset]
+        
+    # 定义检测频率的逻辑
+    # 如果是 Daily，每天都检测；如果是 Monthly，只有月底检测，以此类推
+    is_rebalance_day = pd.Series(False, index=data.index)
+    if check_freq == 'Daily':
+        is_rebalance_day[:] = True
+    elif check_freq == 'Monthly':
+        is_rebalance_day = data.reset_index().groupby(data.index.to_period('M'))['Date'].idxmax().isin(data.index)
+        # 上面这行稍复杂，简化处理：取重采样后的索引
+        resampled_dates = data.resample('ME').last().index
+        # 找到最接近的交易日
+        search_indexer = data.index.searchsorted(resampled_dates)
+        search_indexer = search_indexer[search_indexer < len(data)]
+        valid_dates = data.index[search_indexer]
+        is_rebalance_day = pd.Series(data.index.isin(valid_dates), index=data.index)
+    # (Quarterly, Annually 逻辑类似，此处为了demo简化，重点放在阈值逻辑)
+
+    # --- 每日循环 ---
+    # 为了性能，这里不使用iterrows，而是简单循环
+    # 注意：这里模拟的是 "Close 价格触发，Close 价格成交" (简化模型)
+    
+    dates = data.index
+    prices_values = data.values # Numpy array 更快
+    
+    # 映射 column index
+    asset_idx = {asset: i for i, asset in enumerate(data.columns)}
+    
+    last_rebalance_val = initial_cash
+    
+    for i in range(len(dates)):
+        current_date = dates[i]
+        current_prices = prices_values[i]
+        
+        # 1. 计算当前市值
+        current_vals = {}
+        total_value = 0.0
+        for asset in assets:
+            val = holdings[asset] * current_prices[asset_idx[asset]]
+            current_vals[asset] = val
+            total_value += val
+            
+        portfolio_history.append({'Date': current_date, 'Total Value': total_value})
+        
+        # 2. 检查是否需要检测
+        # 如果不是检测日，直接跳过
+        if check_freq != 'Daily' and not is_rebalance_day[i]:
+            continue
+            
+        # 3. 计算权重并检查阈值
+        needs_rebalance = False
+        
+        # 绝对阈值逻辑 (Absolute Band)
+        # 比如 25% ± 5% -> [20%, 30%]
+        lower_bound = 0.25 - threshold
+        upper_bound = 0.25 + threshold
+        
+        for asset in assets:
+            weight = current_vals[asset] / total_value
+            if weight < lower_bound or weight > upper_bound:
+                needs_rebalance = True
+                break
+        
+        # 4. 执行再平衡
+        if needs_rebalance:
+            target_val = total_value * 0.25
+            for asset in assets:
+                price = current_prices[asset_idx[asset]]
+                holdings[asset] = target_val / price
+            
+            rebalance_log.append({
+                'Date': current_date, 
+                'Type': 'Rebalance', 
+                'Value': total_value
+            })
+
+    return pd.DataFrame(portfolio_history).set_index('Date'), pd.DataFrame(rebalance_log)
+
+# --- 4. 侧边栏控制面板 ---
+st.sidebar.header("⚙️ 回测参数设置")
+
+# 用户选择阈值
+selected_thresholds = st.sidebar.multiselect(
+    "选择要对比的阈值 (Bands)",
+    options=[0.0, 0.05, 0.10, 0.15, 0.20, 0.25],
+    default=[0.05, 0.10, 0.15],
+    format_func=lambda x: f"±{int(x*100)}% ({25-int(x*100)}%-{25+int(x*100)}%)"
+)
+
+# 频率选择 (简化版，强制 Daily Check 以配合阈值策略)
+st.sidebar.info("注意：阈值策略通常配合 '每日检测' 使用，以捕捉瞬间波动。")
+
+# --- 5. 执行对比 ---
+if st.button("开始枚举回测 (Run Benchmark)"):
+    
+    results_summary = []
+    all_equity_curves = pd.DataFrame()
+    
+    for thresh in selected_thresholds:
+        # 运行回测
+        equity_df, log_df = run_backtest(df, threshold=thresh, check_freq='Daily')
+        
+        # 计算指标
+        start_val = equity_df['Total Value'].iloc[0]
+        end_val = equity_df['Total Value'].iloc[-1]
+        years = (equity_df.index[-1] - equity_df.index[0]).days / 365.25
+        cagr = (end_val / start_val) ** (1/years) - 1
+        
+        # 计算最大回撤
+        roll_max = equity_df['Total Value'].cummax()
+        drawdown = (equity_df['Total Value'] - roll_max) / roll_max
+        max_dd = drawdown.min()
+        
+        # 交易次数
+        num_trades = len(log_df)
+        
+        name = f"Threshold ±{int(thresh*100)}%"
+        
+        # 存结果
+        results_summary.append({
+            "策略名称": name,
+            "总收益率": (end_val/start_val) - 1,
+            "年化收益 (CAGR)": cagr,
+            "最大回撤": max_dd,
+            "再平衡次数": num_trades,
+            "最终资金": end_val
+        })
+        
+        all_equity_curves[name] = equity_df['Total Value']
 
-"""
-# Welcome to Streamlit!
-
-Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:.
-If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
-forums](https://discuss.streamlit.io).
-
-In the meantime, below is an example of what you can do with just a few lines of code:
-"""
-
-num_points = st.slider("Number of points in spiral", 1, 10000, 1100)
-num_turns = st.slider("Number of turns in spiral", 1, 300, 31)
-
-indices = np.linspace(0, 1, num_points)
-theta = 2 * np.pi * num_turns * indices
-radius = indices
-
-x = radius * np.cos(theta)
-y = radius * np.sin(theta)
-
-df = pd.DataFrame({
-    "x": x,
-    "y": y,
-    "idx": indices,
-    "rand": np.random.randn(num_points),
-})
-
-st.altair_chart(alt.Chart(df, height=700, width=700)
-    .mark_point(filled=True)
-    .encode(
-        x=alt.X("x", axis=None),
-        y=alt.Y("y", axis=None),
-        color=alt.Color("idx", legend=None, scale=alt.Scale()),
-        size=alt.Size("rand", legend=None, scale=alt.Scale(range=[1, 150])),
-    ))
+    # --- 6. 展示结果 ---
+    st.subheader("📊 资金曲线对比")
+    st.line_chart(all_equity_curves)
+    
+    st.subheader("🏆 绩效统计表")
+    summary_df = pd.DataFrame(results_summary)
+    
+    # 格式化表格显示
+    st.dataframe(summary_df.style.format({
+        "总收益率": "{:.2%}",
+        "年化收益 (CAGR)": "{:.2%}",
+        "最大回撤": "{:.2%}",
+        "最终资金": "${:,.2f}"
+    }))
+    
+    st.markdown("""
+    **解读指南:**
+    * **再平衡次数**: 如果次数太少（比如0次），说明阈值设得太宽了。如果次数太多（比如几百次），交易成本（滑点、佣金、税）会吃掉利润。
+    * **最大回撤**: 观察哪个阈值在极端行情（如2008, 2020）下保护得最好。
+    """)
+    
+else:
+    st.info("👈 请在左侧选择参数并点击 '开始回测'")