Refactor code structure for improved readability and maintainability

data/data_quality_report.txt

@@ -0,0 +1,37 @@
+============================================================
+数据质量报告 / Data Quality Report
+生成时间: 2026-06-07T16:05:29.175368
+数据源: AKShare (fund_etf_hist_em)
+============================================================
+
+159995 (芯片ETF):
+  日期范围: 2021-02-04 ~ 2025-05-30 (1043 天)
+  原始缺失: 265 天 (17.1%)
+  日均收益率: +0.000172
+  日波动率:   0.021145
+  最大日涨幅: +10.01%
+  最大日跌幅: -9.05%
+
+159819 (人工智能ETF):
+  日期范围: 2021-02-04 ~ 2025-05-30 (1043 天)
+  原始缺失: 421 天 (27.1%)
+  日均收益率: +0.000163
+  日波动率:   0.019471
+  最大日涨幅: +10.07%
+  最大日跌幅: -9.95%
+
+516160 (新能源ETF):
+  日期范围: 2021-02-04 ~ 2025-05-30 (1043 天)
+  原始缺失: 510 天 (32.8%)
+  日均收益率: -0.000343
+  日波动率:   0.020161
+  最大日涨幅: +10.07%
+  最大日跌幅: -9.98%
+
+159928 (消费ETF):
+  日期范围: 2021-02-04 ~ 2025-05-30 (1043 天)
+  原始缺失: 0 天 (0.0%)
+  日均收益率: -0.000380
+  日波动率:   0.015514
+  最大日涨幅: +10.01%
+  最大日跌幅: -9.94%

demo_akshare_data.py

@@ -32,12 +32,20 @@ from datetime import datetime, timedelta
 import time
 import os
 
+# ── 绕过系统代理，直连东方财富 ──
+# macOS 上的 Clash/小火箭等代理工具的出口 IP 容易被东方财富反爬封禁。
+# 设置环境变量让 requests 库对 eastmoney.com 不走代理。
+os.environ["NO_PROXY"] = os.environ.get("NO_PROXY", "") + ",eastmoney.com,eastmoneyfutures.com,10jqka.com.cn"
+os.environ["no_proxy"] = os.environ["NO_PROXY"]
+
 # ── 检查 akshare 是否安装 ──
 try:
     import akshare as ak
     print(f"✓ AKShare 版本: {ak.__version__}")
 except ImportError:
-    print("请先安装 AKShare: pip install akshare")
+    print("❌ AKShare 未安装。请在 trading conda 环境中执行:")
+    print("   conda activate trading")
+    print("   pip install akshare")
     raise
 
 print("=" * 68)
@@ -118,18 +126,26 @@ def fetch_etf_safe(code: str, start: str, end: str, max_retries: int = 3):
                 end_date=end,
                 adjust="qfq",     # 前复权
             )
-            time.sleep(0.5)       # 礼貌限速，避免被封 IP
             return df
         except Exception as e:
-            print(f"  ⚠ {code} 第 {attempt+1} 次获取失败: {e}")
+            err_msg = str(e)
+            if "ProxyError" in err_msg or "RemoteDisconnected" in err_msg:
+                wait = (attempt + 1) * 5  # 代理/连接问题，等更久
+                print(f"\n  ⚠ {code} 连接被拒绝 (代理/反爬)，{wait}s 后重试...")
+            else:
+                wait = 2.0
+                print(f"\n  ⚠ {code} 第 {attempt+1} 次获取失败: {e}")
             if attempt < max_retries - 1:
-                time.sleep(2.0)
-    print(f"  ✗ {code} 全部 {max_retries} 次获取失败，跳过")
+                time.sleep(wait)
+    print(f"\n  ✗ {code} 全部 {max_retries} 次获取失败，跳过")
     return None
 
 
 etf_data = {}
-for code, name in ETF_UNIVERSE.items():
+for i, (code, name) in enumerate(ETF_UNIVERSE.items()):
+    if i > 0:
+        # 请求间隔 + 随机抖动，避免触发东方财富反爬
+        time.sleep(2.0 + np.random.uniform(0, 1.5))
     print(f"  获取 {code} ({name})...", end=" ")
     df = fetch_etf_safe(code, START_DATE, END_DATE)
     if df is not None:
@@ -234,26 +250,17 @@ if index_data:
 print("\n[§4] 获取北向资金数据")
 
 try:
-    # 北向资金日频净流入
-    north_flow = ak.stock_hsgt_north_net_flow_in_em(symbol="北上")
-    if north_flow is not None and len(north_flow) > 0:
-        print(f"  北向资金净流入: {north_flow.shape}")
-        print(f"  列名: {list(north_flow.columns)}")
-        # 标准化为通用格式
-        if "date" in north_flow.columns and "value" in north_flow.columns:
-            north_flow = north_flow.rename(columns={"date": "日期", "value": "净流入_亿"})
-        print(f"  最新数据:\n{north_flow.tail(3)}")
-    else:
-        print("  未获取到北向资金数据 (接口可能已变更)")
-
-    time.sleep(0.5)
-
-    # 沪深港通历史资金流向 (更详细)
-    hsgt = ak.stock_hsgt_hist_em(symbol="沪股通")
-    if hsgt is not None and len(hsgt) > 0:
-        print(f"\n  沪深港通历史资金: {hsgt.shape}")
-        print(f"  列名: {list(hsgt.columns)}")
-        print(f"  最新 3 行:\n{hsgt.tail(3)}")
+    # 北向资金: stock_hsgt_hist_em 获取沪股通/深股通历史资金流向
+    for market, label in [("沪股通", "沪股通(北向)"), ("深股通", "深股通(北向)")]:
+        try:
+            hsgt = ak.stock_hsgt_hist_em(symbol=market)
+            if hsgt is not None and len(hsgt) > 0:
+                print(f"  {label}: {hsgt.shape}")
+                print(f"  列名: {list(hsgt.columns)}")
+                print(f"  最新 3 行:\n{hsgt.tail(3)}")
+            time.sleep(1.5)  # 避免触发反爬
+        except Exception as e:
+            print(f"  {label} 获取失败: {e}")
 
 except Exception as e:
     print(f"  北向资金获取失败: {e}")

demo_tushare_data.py

@@ -47,7 +47,9 @@ try:
     import tushare as ts
     print(f"✓ Tushare 版本: {ts.__version__}")
 except ImportError:
-    print("请先安装 Tushare: pip install tushare")
+    print("❌ Tushare 未安装。请在 trading conda 环境中执行:")
+    print("   conda activate trading")
+    print("   pip install tushare")
     raise
 
 print("=" * 68)

requirements.txt

@@ -2,3 +2,6 @@ numpy
 pandas
 matplotlib
 scipy
+scikit-learn
+akshare
+tushare