{"id":1767,"date":"2025-01-03T07:51:00","date_gmt":"2025-01-02T23:51:00","guid":{"rendered":"https:\/\/blog.laoyulaoyu.top\/?p=1767"},"modified":"2025-01-01T12:18:34","modified_gmt":"2025-01-01T04:18:34","slug":"%e3%80%82%e3%80%82%e3%80%82%e5%ae%9e%e6%88%98%e6%95%99%e5%ad%a6%ef%bc%9a%e6%9e%84%e5%bb%ba%e5%8f%af%e8%a7%a3%e9%87%8a%e7%9a%84%e5%8f%98%e6%8d%a2%e5%99%a8%e6%a8%a1%e5%9e%8b%ef%bc%8c%e7%b2%be%e5%87%86","status":"publish","type":"post","link":"https:\/\/www.laoyulaoyu.com\/index.php\/2025\/01\/03\/%e3%80%82%e3%80%82%e3%80%82%e5%ae%9e%e6%88%98%e6%95%99%e5%ad%a6%ef%bc%9a%e6%9e%84%e5%bb%ba%e5%8f%af%e8%a7%a3%e9%87%8a%e7%9a%84%e5%8f%98%e6%8d%a2%e5%99%a8%e6%a8%a1%e5%9e%8b%ef%bc%8c%e7%b2%be%e5%87%86\/","title":{"rendered":"\u5b9e\u6218\u6559\u5b66\uff1a\u6784\u5efa\u53ef\u89e3\u91ca\u7684\u53d8\u6362\u5668\u6a21\u578b\uff0c\u7cbe\u51c6\u9884\u6d4b\u80a1\u4ef7\u6ce2\u52a8\uff08\u4e8c\uff09"},"content":{"rendered":"\n

\u4f5c\u8005\uff1a\u8001\u4f59\u635e\u9c7c<\/a><\/p>\n\n\n\n

\u539f\u521b\u4e0d\u6613\uff0c\u8f6c\u8f7d\u8bf7\u6807\u660e\u51fa\u5904\u53ca\u539f\u4f5c\u8005\u3002<\/mark><\/strong><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\n
\u5199\u5728\u524d\u9762\u7684\u8bdd\uff1a<\/strong>\u5728\u672c\u6587\u4e2d\uff0c\u6211\u5c06\u7ee7\u7eed\u5e26\u4f60\u6df1\u5165\u4e86\u89e3\u5982\u4f55\u6784\u5efa\u4e00\u4e2a\u65f6\u6001\u878d\u5408\u8f6c\u6362\u5668\uff08TFT\uff09\u6a21\u578b\uff0c\u7528\u4ee5\u9884\u6d4b1\u5206\u949f\u7ea7\u522b\u7684\u80a1\u7968\u4ef7\u683c\u3002\u8fd9\u662f\u6574\u4e2a\u4e3b\u9898\u7684\u7b2c\u4e8c\u90e8\u5206\uff0c\u6211\u4eec\u5c06\u63a2\u8ba8\u5e02\u573a\u5fae\u89c2\u7ed3\u6784\u3001\u80a1\u7968\u6ce2\u52a8\u6027\u7684\u591a\u7ef4\u8861\u91cf\u65b9\u6cd5\uff0c\u4ee5\u53ca\u76c8\u5229\u4e8b\u4ef6\u5bf9\u80a1\u4ef7\u6ce2\u52a8\u6027\u7684\u663e\u8457\u5f71\u54cd<\/mark>\u3002\u6b64\u5916\uff0c\u6211\u8fd8\u5c06\u5206\u4eab\u5982\u4f55\u901a\u8fc7\u6210\u4ea4\u91cf\u52a8\u6001\u6765\u8bc4\u4f30\u4ea4\u6613\u6d3b\u52a8\u7684\u5f3a\u5ea6\uff0c\u4ee5\u53ca\u8fd9\u4e9b\u6d1e\u5bdf\u5982\u4f55\u8f6c\u5316\u4e3a\u673a\u5668\u5b66\u4e60\u6a21\u578b\uff08\u5c01\u9762\u56fe\u4e3aTFT \u9884\u6d4b AMD \u80a1\u7968\u7684\u6210\u529f\u793a\u610f\uff09<\/em>\u3002<\/pre>\n<\/blockquote>\n\n\n\n
\u6b22\u8fce\u56de\u5230\u6211\u4eec\u4f7f\u7528\u65f6\u6001\u878d\u5408\u8f6c\u6362\u5668\u5efa\u7acb\u9ad8\u9891\u80a1\u4ef7\u9884\u6d4b\u6a21\u578b\u7684\u7cfb\u5217\uff01\u5728\u7b2c\u4e00\u90e8\u5206\u4e2d\uff0c\u6211\u4eec\u4ecb\u7ecd\u4e86\u6570\u636e\u6536\u96c6\u548c\u51c6\u5907\u7684\u57fa\u7840\u77e5\u8bc6\uff0c\u91cd\u70b9\u662f\u6536\u96c6\u9ad8\u6d41\u52a8\u6027\u80a1\u7968\u7684\u5206\u949f\u7ea7\u6570\u636e\u3002\u73b0\u5728\uff0c\u6211\u4eec\u5c06\u6df1\u5165\u7814\u7a76\u5e02\u573a\u5fae\u89c2\u7ed3\u6784\uff0c\u63a2\u7d22\u76c8\u5229\u4e8b\u4ef6\u3001\u6210\u4ea4\u91cf\u6a21\u5f0f\u548c\u6ce2\u52a8\u7387\u7b49\u5404\u79cd\u56e0\u7d20\u662f\u5982\u4f55\u76f8\u4e92\u4f5c\u7528\u521b\u9020\u4ea4\u6613\u673a\u4f1a\u7684\u3002\u901a\u8fc7\u4e86\u89e3\u8fd9\u4e9b\u5173\u7cfb\uff0c\u6211\u4eec\u5c31\u80fd\u4e3a\u6211\u4eec\u7684 TFT \u6a21\u578b\u8bbe\u8ba1\u51fa\u66f4\u6709\u610f\u4e49\u7684\u7279\u5f81\u3002\u5982\u679c\u60a8\u8fd8\u6ca1\u6709\u9605\u8bfb\u7b2c 1 \u90e8\u5206\uff0c\u6211\u5efa\u8bae\u60a8\u5148\u9605\u8bfb\u8fd9\u4e00\u90e8\u5206\uff0c\u4ee5\u4e86\u89e3\u6211\u4eec\u7684\u6570\u636e\u7ba1\u9053\u548c\u521d\u59cb\u9884\u5904\u7406\u6b65\u9aa4\u3002\u8ba9\u6211\u4eec\u8fdb\u5165\u5e02\u573a\u5fae\u89c2\u7ed3\u6784\u548c\u6280\u672f\u5206\u6790\u7684\u8ff7\u4eba\u4e16\u754c\uff01<\/p>\n\n\n\n
\u7b2c\u4e00\u90e8\u5206\u770b\u8fd9\u91cc\uff1a\u300a\u5b9e\u6218\u6559\u5b66\uff1a\u6784\u5efa\u53ef\u89e3\u91ca\u7684\u53d8\u6362\u5668\u6a21\u578b\uff0c\u7cbe\u51c6\u9884\u6d4b\u80a1\u4ef7\u6ce2\u52a8\uff08\u4e00\uff09<\/a>\u300b<\/em><\/p>\n\n\n\n
\u4e00\u3001\u5e02\u573a\u7ed3\u6784\u6982\u89c8<\/strong><\/h2>\n\n\n\n
\u5728\u5f00\u59cb\u6784\u5efa\u590d\u6742\u7684\u7279\u5f81\u4e4b\u524d\uff0c\u5148\u660e\u786e\u6211\u4eec\u6240\u9009\u80a1\u7968\u7ec4\u5408\u7684\u7ed3\u6784\u662f\u5173\u952e\u4e00\u6b65\u3002\u5728\u7cfb\u5217\u6587\u7ae0\u7684\u7b2c\u4e00\u7bc7\u4e2d\uff0c\u6211\u4eec\u5df2\u7ecf\u5bf9\u521d\u6b65\u6570\u636e\u96c6\u8fdb\u884c\u4e86\u7b5b\u9009\uff0c\u4e13\u6ce8\u4e8e\u90a3\u4e9b\u5177\u6709\u6700\u9ad8\u6d41\u52a8\u6027\u7684\u80a1\u7968\u3002\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u5c06\u5bf9\u8fd9\u4e9b\u9009\u5b9a\u516c\u53f8\u7684\u6838\u5fc3\u7279\u5f81\u8fdb\u884c\u8be6\u7ec6\u5ba1\u89c6\u3002<\/p>\n\n\n\n
1.1 \u83b7\u53d6\u57fa\u672c\u6570\u636e<\/strong><\/h3>\n\n\n\n
\u9996\u5148\uff0c\u6211\u4eec\u5c06\u4f7f\u7528 yfinance \u5e93\u4e3a\u6211\u4eec\u5b87\u5b99\u4e2d\u7684\u6bcf\u53ea\u80a1\u7968\u6536\u96c6\u57fa\u672c\u6570\u636e\u3002\u5176\u4e2d\u5305\u62ec\u884c\u4e1a\u5206\u7c7b\u3001\u80a1\u6743\u7ed3\u6784\u548c\u6536\u76ca\u65e5\u671f\uff1a<\/p>\n\n\n\n
import yfinance as yf\nimport json\nfrom tqdm import tqdm\n\nfundamental_data = {}\nmaybe_delisted = []\nfor stock in tqdm(df[\"symbol\"].unique(), desc=\"Fetching fundamental data...\"):\n    ticker = yf.Ticker(stock)\n    try:\n        ticker_info = ticker.info\n        fundamental_data[stock] = {\n            \"earning_dates\": [date for date in ticker.earnings_dates.index if DATA_START.date() <= date.date() < DATA_CUTOFF.date()],\n            \"sector\": ticker_info[\"sector\"],\n            \"industry\": ticker_info[\"industry\"],\n            \"total_shares\": ticker_info[\"sharesOutstanding\"],\n            \"float_shares\": ticker_info[\"floatShares\"],\n        }\n    except Exception as e:\n        maybe_delisted.append(stock)\n        continue\nfor stock in fundamental_data.keys():\n    fundamental_data[stock][\"earning_dates\"] = [date.strftime(\"%Y-%m-%d %H:%M:%S\") for date in fundamental_data[stock][\"earning_dates\"]]\nwith open(\"fundamental_data.json\", \"w\") as f:\n    json.dump(fundamental_data, f)\n\ndf = df[df[\"symbol\"].isin(list(fundamental_data.keys()))] # df is the same dataframe from part 1. It holds the minute level bars.\ndf = df[~(df[\"symbol\"] == \"GOOGL\")] # Google has 2 classes of shares, their price is identical, so we will remove one<\/code><\/pre>\n\n\n\n
\u8fd0\u884c\u4e0a\u9762\u7684\u4ee3\u7801\u540e\uff0c\u6211\u4eec\u5c06\u5f97\u5230\u4e00\u4e2a\u5305\u542b\u7b26\u53f7\u57fa\u672c\u9762\u6570\u636e\u7684\u5b57\u5178\u3002\u4f8b\u5982\uff0c\u8981\u67e5\u770b $AAPL \u7684\u57fa\u672c\u9762\u6570\u636e\uff0c\u6211\u4eec\u53ef\u4ee5\u8fd9\u6837\u505a\uff1a<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u6ce8\u610f\uff1a<\/strong>\u5728\u672c\u6559\u7a0b\u4e2d\uff0c\u6211\u4eec\u53ea\u4f7f\u7528\u4ece 2024-01-01 \u5f00\u59cb\u7684\u6570\u636e\uff0c\u56e0\u6b64\u6211\u4eec\u76f8\u5e94\u5730\u8fc7\u6ee4\u4e86\u6536\u5165\u65e5\u671f\u3002<\/p>\n\n\n\n
1.2 \u5206\u6790\u884c\u4e1a\u5206\u5e03\u60c5\u51b5<\/strong><\/h3>\n\n\n\n
\u73b0\u5728\uff0c\u6211\u4eec\u6709\u4e86\u57fa\u672c\u9762\u6570\u636e\uff0c\u8ba9\u6211\u4eec\u6765\u5206\u6790\u4e00\u4e0b\u80a1\u7968\u5e02\u573a\u7684\u884c\u4e1a\u5206\u5e03\uff1a<\/p>\n\n\n\n
import pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n\nfundamental_df = pd.DataFrame.from_dict(fundamental_data, orient='index')\nsector_industry_counts = fundamental_df.groupby(['sector', 'industry']).size().unstack(fill_value=0)\nsector_totals = sector_industry_counts.sum(axis=1)\nsector_industry_counts = sector_industry_counts.loc[sector_totals.sort_values(ascending=False).index]\nplt.figure(figsize=(12, 8))\nsector_industry_counts.plot(kind='bar', stacked=True, colormap='viridis', figsize=(12, 8))\n\nplt.title('Number of Stocks per Sector and Industry')\nplt.xlabel('Sector')\nplt.ylabel('Number of Stocks')\nplt.legend(title='Industry', bbox_to_anchor=(1.05, 1), loc='upper left', fontsize='small')\nplt.tight_layout()\nplt.show()<\/code><\/pre>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u8fd9\u9879\u5206\u6790\u8868\u660e\uff0c\u6211\u4eec\u9009\u62e9\u7684\u80a1\u7968\u8303\u56f4\u5728\u5404\u4e3b\u8981\u884c\u4e1a\u90fd\u6709\u5f88\u597d\u7684\u5206\u6563\u6027\uff0c\u5c24\u5176\u96c6\u4e2d\u5728\u79d1\u6280\u3001\u533b\u7597\u4fdd\u5065\u548c\u91d1\u878d\u670d\u52a1\u9886\u57df\u3002\u8fd9\u79cd\u591a\u6837\u6027\u5bf9\u6211\u4eec\u7684\u6a21\u578b\u975e\u5e38\u91cd\u8981\uff0c\u56e0\u4e3a\u5b83\u9700\u8981\u5b66\u4e60\u4e0d\u540c\u7c7b\u578b\u80a1\u7968\u7684\u901a\u7528\u6a21\u5f0f\u3002<\/p>\n\n\n\n
1.3 \u8bc4\u4f30\u80a1\u7968\u89c4\u6a21<\/strong><\/h3>\n\n\n\n
\u5e02\u503c\u662f\u4ea4\u6613\u8005\u7528\u6765\u8bc4\u4f30\u80a1\u7968\u89c4\u6a21\u7684\u4e00\u4e2a\u5173\u952e\u6307\u6807\u3002\u8ba1\u7b97\u65b9\u6cd5\u662f\u7528\u80a1\u4ef7\u4e58\u4ee5\u5df2\u53d1\u884c\u80a1\u7968\u603b\u6570\u3002\u5728\u76d8\u4e2d\u6570\u636e\u6846\u67b6\u4e2d\uff0c\u6211\u5c06\u4f7f\u7528\u5f53\u5929\u7684\u5f00\u76d8\u4ef7\u4f5c\u4e3a\u80a1\u4ef7\u6765\u8ba1\u7b97\u5e02\u503c\u3002\u7136\u540e\u9664\u4ee5 100 \u4e07\uff0c\u5c31\u5f97\u5230\u4e86\u4ee5\u767e\u4e07\u4e3a\u5355\u4f4d\u7684\u5e02\u503c\u3002<\/p>\n\n\n\n
df[\"market_cap\"] = df.groupby([\"symbol\", \"date\"])[\"open\"].transform(\"first\") * df[\"symbol\"].apply(lambda x: fundamental_data[x][\"total_shares\"]).astype(float) \/ 1_000_000\n# Top 20 stocks by market capitalization\ntop_20_stocks = df.groupby(\"symbol\")[\"market_cap\"].last().nlargest(20)\n\n# Step 3: Plot the top 20 stocks\nplt.figure(figsize=(12, 8))\ntop_20_stocks.sort_values().plot(kind='barh', color='skyblue')\nplt.title(\"Top 20 Stocks by Market Capitalization\")\nplt.xlabel(\"Market Capitalization (in millions)\")\nplt.ylabel(\"Stock Symbol\")\nplt.show()<\/code><\/pre>\n\n\n\n
\u4e0b\u9762\u662f\u8bc4\u4f30\u51fa\u7684\u5e02\u503c\u6392\u540d\u524d 20 \u4f4d\u7684\u80a1\u7968\uff1a<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u8ba9\u6211\u4eec\u6df1\u5165\u63a2\u8ba8\u65e5\u5185\u4ea4\u6613\u8005\u6781\u4e3a\u5173\u6ce8\u7684\u6307\u6807\u2014\u2014\u6ce2\u52a8\u6027\u3002\u6ce2\u52a8\u6027\u53cd\u6620\u4e86\u80a1\u7968\u4ef7\u683c\u7684\u201c\u8109\u52a8\u201d\uff0c\u5373\u4ef7\u683c\u968f\u65f6\u95f4\u53d8\u5316\u7684\u6ce2\u52a8\u5e45\u5ea6\u3002\u8fd9\u79cd\u6ce2\u52a8\u901a\u8fc7\u4e24\u79cd\u4e3b\u8981\u65b9\u5f0f\u88ab\u6355\u6349\uff1a\u5b83\u91cf\u5316\u4e86\u80a1\u7968\u4ef7\u683c\u5728\u4efb\u4e00\u65b9\u5411\u4e0a\u7684\u5267\u70c8\u53d8\u52a8\u3002\u53ef\u4ee5\u5c06\u6ce2\u52a8\u6027\u89c6\u4e3a\u80a1\u7968\u7684\u201c\u6027\u683c\u201d\uff0c\u6709\u7684\u80a1\u7968\u8868\u73b0\u5e73\u7a33\uff0c\u800c\u6709\u7684\u5219\u53d8\u52a8\u65e0\u5e38\u3002\u9ad8\u6ce2\u52a8\u6027\u7684\u80a1\u7968\u5f80\u5f80\u4f34\u968f\u7740\u8f83\u9ad8\u7684\u98ce\u9669\uff0c\u4f46\u5b83\u4eec\u5bf9\u65e5\u5185\u4ea4\u6613\u8005\u6765\u8bf4\u9887\u5177\u5438\u5f15\u529b\uff0c\u56e0\u4e3a\u5927\u5e45\u7684\u4ef7\u683c\u6ce2\u52a8\u63d0\u4f9b\u4e86\u66f4\u591a\u7684\u76c8\u5229\u6f5c\u529b\uff08\u540c\u65f6\u4e5f\u4f34\u968f\u7740\u66f4\u9ad8\u7684\u4e8f\u635f\u98ce\u9669\uff09\u3002\u4e3a\u4e86\u5c06\u8fd9\u79cd\u6ce2\u52a8\u6027\u91cf\u5316\uff0c\u6211\u4eec\u91c7\u7528\u5386\u53f2\u6ce2\u52a8\u7387\u6765\u8fdb\u884c\u8ba1\u7b97\uff0c\u5177\u4f53\u65b9\u6cd5\u5982\u4e0b\uff1a<\/p>\n\n\n\n
    \n
  • \u53d6\u6bcf\u65e5\u56de\u62a5\u7684\u6807\u51c6\u504f\u5dee<\/li>\n\n\n\n
  • \u6309\u5f00\u76d8\u4ef7\u5f52\u4e00\u5316<\/li>\n\n\n\n
  • \u5c06\u7ed3\u679c\u6309\u5e74\u8ba1\u7b97\uff08\u4e58\u4ee5 \u221a252\uff0c\u5373\u4e00\u5e74\u7684\u4ea4\u6613\u5929\u6570\uff09<\/li>\n<\/ul>\n\n\n\n
    \u8fd9\u4e3a\u6211\u4eec\u63d0\u4f9b\u4e86\u4e00\u79cd\u6807\u51c6\u5316\u7684\u65b9\u6cd5\u6765\u6bd4\u8f83\u4e0d\u540c\u80a1\u7968\u7684\u201c\u7d27\u5f20\u5ea6\u201d\uff0c\u800c\u65e0\u9700\u8003\u8651\u5b83\u4eec\u7684\u4ef7\u683c\u6c34\u5e73\u3002\u5e74\u5316\u6ce2\u52a8\u7387\u4e3a30%\u7684\u80a1\u7968\uff0c\u5176\u4ef7\u683c\u53d8\u52a8\u901a\u5e38\u8981\u6bd4\u6ce2\u52a8\u7387\u4e3a15%\u7684\u80a1\u7968\u66f4\u4e3a\u5267\u70c8\u3002<\/p>\n\n\n\n
    df_daily = df.groupby('symbol').resample('D').agg({\n    'close': 'last'\n}).dropna() # dropna to remove Saturdays and Sundays\n# Calculate the daily log returns\ndf_daily['return'] = df_daily.groupby('symbol')['close'].transform(lambda x: np.log(x \/ x.shift(1)))\n# Calculate the rolling standard deviation of the daily returns\nwindow_size = 30\ndf_daily['rolling_std'] = df_daily.groupby('symbol')['return'].rolling(window=window_size, min_periods=1).std().reset_index(0, drop=True)\n# Annualize the volatility\ndf_daily['hist_vol'] = (df_daily['rolling_std'] * np.sqrt(252))\ndf_daily['hist_vol'] = df_daily['hist_vol'].fillna(method='bfill')\n# Merge the daily historical volatility back into the original DataFrame\ndf_daily[\"symbol\"] = df_daily.index.get_level_values(0)\ndf_daily[\"datetime\"] = df_daily.index.get_level_values(1)\ndf_daily = df_daily.reset_index(drop=True)\ndf_daily[\"date\"] = df_daily[\"datetime\"].dt.date\nminute_index = df.index\ndf = df.merge(df_daily[['symbol', 'date', 'hist_vol']], on=['symbol', 'date'], how='left').set_index(minute_index)<\/code><\/pre>\n\n\n\n
    \u4e8c\u3001\u5206\u6790\u4ea4\u6613\u52a8\u6001<\/strong><\/h2>\n\n\n\n
    \u6d41\u901a\u80a1\u4ee3\u8868\u5b9e\u9645\u53ef\u4f9b\u516c\u5f00\u4ea4\u6613\u7684\u80a1\u7968\u6570\u91cf\u3002\u867d\u7136\u4e00\u5bb6\u516c\u53f8\u53ef\u80fd\u6709\u6570\u767e\u4e07\u6d41\u901a\u80a1\uff0c\u4f46\u6d41\u901a\u80a1\u901a\u5e38\u8f83\u5c11\uff0c\u5176\u8ba1\u7b97\u65b9\u6cd5\u662f\u4ece\u6d41\u901a\u80a1\u603b\u6570\u4e2d\u51cf\u53bb\u9650\u5236\u6027\u80a1\u7968\uff08\u7531\u5185\u90e8\u4eba\u5458\u3001\u5458\u5de5\u548c\u4e3b\u8981\u80a1\u4e1c\u6301\u6709\uff09\u3002\u6d6e\u52a8\u89c4\u6a21\u5bf9\u4ea4\u6613\u52a8\u6001\u6709\u91cd\u5927\u5f71\u54cd\uff1a<\/p>\n\n\n\n
      \n
    • \u4f4e\u6d41\u901a\u91cf\u80a1\u7968<\/strong>\uff08\u53ef\u81ea\u7531\u4ea4\u6613\u7684\u80a1\u7968\u6570\u91cf\u8f83\u5c11\uff09\u7684\u4ef7\u683c\u6ce2\u52a8\u5f80\u5f80\u66f4\u5927\uff0c\u56e0\u4e3a\u5373\u4f7f\u4ea4\u6613\u91cf\u9002\u4e2d\uff0c\u4e5f\u4f1a\u9020\u6210\u5de8\u5927\u7684\u4ef7\u683c\u538b\u529b.<\/li>\n\n\n\n
    • \u9ad8\u6d41\u901a\u91cf\u80a1\u7968<\/strong>\u7684\u4ef7\u683c\u8d70\u52bf\u901a\u5e38\u66f4\u4e3a\u7a33\u5b9a\uff0c\u56e0\u4e3a\u9700\u8981\u66f4\u591a\u7684\u6210\u4ea4\u91cf\u624d\u80fd\u63a8\u52a8\u4ef7\u683c\u53d8\u52a8\u3002<\/li>\n<\/ul>\n\n\n\n
      \u4ea4\u6613\u8005\u4f1a\u5bc6\u5207\u5173\u6ce8\u6d6e\u52a8\u6c34\u5e73<\/strong>\uff0c\u56e0\u4e3a\uff1a<\/p>\n\n\n\n
        \n
      • \u4ef7\u683c\u5f71\u54cd\uff1a<\/strong>\u6d6e\u52a8\u6bd4\u7387\u8d8a\u5c0f\uff0c\u610f\u5473\u7740\u6bcf\u7b14\u4ea4\u6613\u5bf9\u4ef7\u683c\u7684\u6f5c\u5728\u5f71\u54cd\u8d8a\u5927\u3002<\/li>\n\n\n\n
      • \u6d41\u52a8\u6027\u98ce\u9669\uff1a<\/strong>\u4f4e\u6d41\u901a\u91cf\u80a1\u7968\u53ef\u80fd\u4f1a\u51fa\u73b0\u8f83\u5927\u7684\u4e70\u5356\u5dee\u4ef7\u548c\u5feb\u901f\u7684\u4ef7\u683c\u53d8\u5316\u3002<\/li>\n\n\n\n
      • \u4ea4\u6613\u91cf\uff1a<\/strong>\u6d6e\u52a8\u4ea4\u6613\u91cf\u6bd4\u6709\u52a9\u4e8e\u8861\u91cf\u6709\u591a\u5c11\u53ef\u7528\u4f9b\u5e94\u91cf\u5728\u79ef\u6781\u4ea4\u6613\u3002<\/li>\n<\/ul>\n\n\n\n
        \u5728\u4ea4\u6613\u4f4e\u6d41\u901a\u91cf\u80a1\u7968\u65f6<\/strong>\uff0c\u8981\u6ce8\u610f\u4ee5\u4e0b\u51e0\u70b9\uff1a<\/p>\n\n\n\n
          \n
        • \u4ef7\u683c\u6ce2\u52a8\u53ef\u80fd\u66f4\u5267\u70c8\u3001\u66f4\u96be\u4ee5\u9884\u6d4b\u3002<\/li>\n\n\n\n
        • \u7531\u4e8e\u6d41\u52a8\u6027\u9650\u5236\uff0c\u5934\u5bf8\u89c4\u6a21\u53d8\u5f97\u81f3\u5173\u91cd\u8981\u3002<\/li>\n\n\n\n
        • \u7531\u4e8e\u4ef7\u5dee\u6269\u5927\uff0c\u9000\u51fa\u4ef7\u683c\u53ef\u80fd\u5927\u5927\u504f\u79bb\u9884\u671f\u6c34\u5e73\u3002<\/li>\n<\/ul>\n\n\n\n
          df[\"shares_float\"] = df[\"symbol\"].apply(lambda x: fundamental_data[x][\"float_shares\"])\nhist_and_float_df = df.groupby('symbol').agg({\n    'hist_vol': 'mean',\n    'shares_float': 'last',\n    'market_cap': 'mean'\n}).dropna()\n\n# Create a 1x2 grid of plots\nfig, axes = plt.subplots(1, 2, figsize=(18, 8))\n\n# Plot 1: Shares Float vs. Historical Volatility\nsns.scatterplot(data=hist_and_float_df, x='shares_float', y='hist_vol', alpha=0.5, ax=axes[0])\naxes[0].set_xscale('log')\naxes[0].set_yscale('log')\naxes[0].set_title('Shares Float vs. Historical Volatility')\naxes[0].set_xlabel('Shares Float')\naxes[0].set_ylabel('Historical Volatility')\n\n# Plot 2: Market Cap vs. Shares Float\nsns.scatterplot(data=hist_and_float_df, x='market_cap', y='hist_vol', alpha=0.5, ax=axes[1])\naxes[1].set_xscale('log')\naxes[1].set_yscale('log')\naxes[1].set_title('Market Cap vs. Historical Volatility')\naxes[1].set_xlabel('Market Cap (in millions)')\naxes[1].set_ylabel('Historical Volatility')\n\n# Show the plots\nplt.tight_layout()\nplt.show()<\/pre>\n\n\n\n
          \u80a1\u7968\u6d6e\u52a8\u4e0e\u5386\u53f2\u6ce2\u52a8\u4e4b\u95f4\u7684\u76f8\u5173\u6027\u5206\u6790\uff1a<\/p>\n\n\n\n
          \"\"<\/figure>\n\n\n\n
          \u6563\u70b9\u56fe\u76f4\u89c2\u5730\u8bc1\u5b9e\u4e86\u8fd9\u4e9b\u5173\u7cfb–\u80a1\u7968\u6d41\u901a\u91cf\u548c\u5e02\u503c\u90fd\u4e0e\u6ce2\u52a8\u7387\u5448\u660e\u663e\u7684\u8d1f\u76f8\u5173\u3002\u6d41\u901a\u76d8\u8f83\u5c0f\u3001\u5e02\u503c\u8f83\u4f4e\u7684\u80a1\u7968\uff08\u6bcf\u5f20\u56fe\u7684\u5de6\u4fa7\uff09\u59cb\u7ec8\u8868\u73b0\u51fa\u8f83\u9ad8\u7684\u6ce2\u52a8\u6027\uff08\u5728 y \u8f74\u4e0a\u8f83\u9ad8\uff09\uff0c\u800c\u6d41\u901a\u76d8\u8f83\u5927\u3001\u5e02\u503c\u8f83\u9ad8\u7684\u80a1\u7968\u5219\u96c6\u4e2d\u5728\u4e24\u5f20\u56fe\u7684\u53f3\u4e0b\u65b9\uff0c\u8868\u73b0\u51fa\u66f4\u7a33\u5b9a\u7684\u4ef7\u683c\u8d70\u52bf\u3002<\/p>\n\n\n\n
          \u4e09\u3001ATR\uff1a\u8861\u91cf\u65e5\u5185\u6ce2\u52a8\u6027<\/strong><\/h2>\n\n\n\n
          \u5e73\u5747\u771f\u5b9e\u6ce2\u52a8\u8303\u56f4\uff08ATR\uff09\u662f\u8861\u91cf\u80a1\u7968\u65e5\u5185\u6ce2\u52a8\u7684\u6709\u529b\u5de5\u5177\u3002\u4e0e\u53ea\u770b\u6536\u76d8\u4ef7\u7684\u6807\u51c6\u6ce2\u52a8\u7387\u8861\u91cf\u65b9\u6cd5\u4e0d\u540c\uff0cATR \u53ef\u4ee5\u6355\u6349\u6574\u4e2a\u4ea4\u6613\u65e5\u7684\u5168\u90e8\u4ef7\u683c\u6ce2\u52a8\u8303\u56f4\uff0c\u56e0\u6b64\u5bf9\u65e5\u5185\u4ea4\u6613\u8005\u7279\u522b\u6709\u4ef7\u503c\u3002ATR\u7684\u5de5\u4f5c\u6a21\u5f0f\u5206\u4e3a\u4e24\u9879\uff1a<\/p>\n\n\n\n
           1.\u9996\u5148\uff0c\u6211\u4eec\u8ba1\u7b97\u6bcf\u5929\u7684\u771f\u5b9e\u6ce2\u52a8\u8303\u56f4 (TR)<\/strong>\uff0c\u5b83\u662f\u4ee5\u4e0b\u4e24\u9879\u4e2d\u6700\u5927\u7684\u4e00\u9879\uff1a<\/p>\n\n\n\n
            \n
          • \u5f53\u524d\u9ad8\u7535\u5e73\u51cf\u53bb\u5f53\u524d\u4f4e\u7535\u5e73\u3002<\/li>\n\n\n\n
          • \u5f53\u524d\u6700\u9ad8\u4ef7\u51cf\u53bb\u524d\u6536\u76d8\u4ef7\u3002<\/li>\n\n\n\n
          • \u524d\u6536\u76d8\u4ef7\u51cf\u5f53\u524d\u6700\u4f4e\u4ef7\u3002<\/li>\n<\/ul>\n\n\n\n
            2.\u7136\u540e\uff0c\u5c06 ATR \u8ba1\u7b97\u4e3a\u8fd9\u4e9b\u771f\u5b9e\u8303\u56f4\u503c\u7684 14 \u5929\u79fb\u52a8\u5e73\u5747\u503c\u3002\u8be5\u516c\u5f0f\u8003\u8651\u4e86\u6240\u6709\u7684\u4ef7\u683c\u7f3a\u53e3\u548c\u9650\u4ef7\u79fb\u52a8\uff0c\u63d0\u4f9b\u4e86\u4e00\u4e2a\u5168\u9762\u7684\u4ef7\u683c\u6ce2\u52a8\u89c6\u56fe\u3002<\/p>\n\n\n\n
            \u5f53\u5206\u6790 ATR \u65f6\uff1a<\/strong><\/p>\n\n\n\n
              \n
            • ATR \u503c\u8d8a\u9ad8\uff0c\u8868\u793a\u80a1\u7968\u6ce2\u52a8\u8d8a\u5927\uff0c\u4ef7\u683c\u6ce2\u52a8\u5e45\u5ea6\u8d8a\u5927\u3002<\/li>\n\n\n\n
            • \u8f83\u4f4e\u7684 ATR \u503c\u8868\u660e\u4ef7\u683c\u8d70\u52bf\u66f4\u7a33\u5b9a\u3001\u66f4\u53ef\u9884\u6d4b\u3002<\/li>\n\n\n\n
            • ATR \u901a\u5e38\u901a\u8fc7\u9664\u4ee5\u80a1\u7968\u4ef7\u683c\u8fdb\u884c\u5f52\u4e00\u5316\u5904\u7406\uff0c\u4ee5\u4fbf\u5728\u4e0d\u540c\u4ef7\u683c\u6c34\u5e73\u4e4b\u95f4\u8fdb\u884c\u6bd4\u8f83\u3002<\/li>\n<\/ul>\n\n\n\n
              \u4e0b\u9762\u662f\u6211\u4eec\u7684\u8ba1\u7b97\u65b9\u6cd5\uff1a<\/p>\n\n\n\n
              df_daily = get_daily_df(df,{\n    'close': 'last',\n    'high': 'max',\n    'low': 'min'\n})\nminute_index = df.index\ndf_daily[\"TR\"] = ta.true_range(high=df_daily['high'], low=df_daily['low'], close=df_daily['close'])\ndf_daily[\"ATR\"] = ta.atr(high=df_daily['high'], low=df_daily['low'], close=df_daily['close'], length=14)\ndf = df.merge(df_daily[['symbol', 'date', 'ATR', 'TR']], on=['symbol', 'date'], how='left')\ndf.set_index(minute_index, inplace=True)\ndf[\"ATR\"] = df[\"ATR\"].fillna(method='bfill')\ndf[\"TR\"] = df[\"TR\"].fillna(method='bfill')\ndf[\"ATR\"] = df[\"ATR\"] \/ df[\"close\"] # Normalize ATR by the stock price\ndf[\"TR\"] = df[\"TR\"] \/ df[\"close\"] # Normalize TR by the stock price<\/code><\/pre>\n\n\n\n
              \u867d\u7136 ATR \u548c\u5386\u53f2\u6ce2\u52a8\u7387\u90fd\u80fd\u8861\u91cf\u4ef7\u683c\u8d70\u52bf\uff0c\u4f46\u5b83\u4eec\u6355\u6349\u7684\u662f\u5e02\u573a\u884c\u4e3a\u7684\u4e0d\u540c\u65b9\u9762\u3002\u8ba9\u6211\u4eec\u5c06\u5b83\u4eec\u7684\u5173\u7cfb\u5f62\u8c61\u5316\uff0c\u4ee5\u4e86\u89e3\u5b83\u4eec\u662f\u5982\u4f55\u4e92\u8865\u7684\uff1a<\/p>\n\n\n\n
                \n
              • ATR \u4e13\u6ce8\u4e8e\u65e5\u5185\u4ef7\u683c\u8303\u56f4\uff0c\u5305\u62ec\u4ea4\u6613\u65f6\u6bb5\u4e4b\u95f4\u7684\u95f4\u9699\u3002<\/li>\n\n\n\n
              • \u5386\u53f2\u6ce2\u52a8\u7387\u8003\u5bdf\u7684\u662f\u65e5\u6536\u76ca\u7387\u5728\u4e00\u6bb5\u65f6\u95f4\u5185\u7684\u79bb\u6563\u7a0b\u5ea6\u3002<\/li>\n\n\n\n
              • \u901a\u8fc7\u5c06\u8fd9\u4e9b\u6307\u6807\u76f8\u4e92\u5bf9\u7167\uff0c\u6211\u4eec\u53ef\u4ee5\u8bc6\u522b\u51fa\u54ea\u4e9b\u80a1\u7968\uff1a<\/li>\n\n\n\n
              • \u5728\u4e24\u79cd\u8861\u91cf\u6807\u51c6\u4e2d\u90fd\u663e\u793a\u51fa\u8f83\u9ad8\u7684\u6ce2\u52a8\u6027\uff08\u53ef\u80fd\u662f\u65e5\u95f4\u4ea4\u6613\u7684\u7406\u60f3\u5019\u9009\u8005\uff09\u3002<\/li>\n\n\n\n
              • \u65e5\u6ce2\u52a8\u7387\u8f83\u9ad8\uff0c\u4f46\u76d8\u4e2d\u6ce2\u52a8\u5e45\u5ea6\u8f83\u5c0f\uff08\u5bb9\u6613\u51fa\u73b0\u7f3a\u53e3\u7684\u80a1\u7968\uff09\u3002<\/li>\n\n\n\n
              • \u663e\u793a\u76d8\u4e2d\u5927\u5e45\u6ce2\u52a8\uff0c\u4f46\u6bcf\u65e5\u6536\u76d8\u66f4\u52a0\u4e00\u81f4\u3002<\/li>\n<\/ul>\n\n\n\n
                # plot correlation between ATR and hist_vol\ndata = df.groupby(\"symbol\").agg({\"hist_vol\": \"mean\", \"ATR\": \"mean\"}).dropna()\nplt.figure(figsize=(12, 8))\nsns.scatterplot(data=data, x=\"ATR\", y=\"hist_vol\")\nplt.title(\"ATR vs Historical Volatility\")\nplt.xscale(\"log\")\nplt.yscale(\"log\")\nplt.xlabel(\"ATR\")\nplt.ylabel(\"Historical Volatility\")\nplt.show()<\/code><\/pre>\n\n\n\n
                \"\"<\/figure>\n\n\n\n
                \u5bf9\u6570\u6563\u70b9\u56fe\u663e\u793a\uff0cATR \u4e0e\u5386\u53f2\u6ce2\u52a8\u7387\u4e4b\u95f4\u5b58\u5728\u660e\u663e\u7684\u6b63\u76f8\u5173\u5173\u7cfb\uff0c\u5404\u70b9\u5f62\u6210\u660e\u663e\u7684\u5bf9\u89d2\u7ebf\u5f62\u6001\u3002\u8fd9\u79cd\u7d27\u5bc6\u7684\u5173\u7cfb\u662f\u6709\u9053\u7406\u7684\uff1a\u65e5\u5185\u4ef7\u683c\u5927\u5e45\u6ce2\u52a8\uff08\u9ad8 ATR\uff09\u7684\u80a1\u7968\u5f80\u5f80\u4e5f\u4f1a\u8868\u73b0\u51fa\u663e\u8457\u7684\u65e5\u95f4\u6536\u76ca\u53d8\u5316\uff08\u9ad8\u5386\u53f2\u6ce2\u52a8\u7387\uff09\u3002\u8fdc\u79bb\u4e3b\u8981\u8d8b\u52bf\u7ebf\u7684\u5c11\u6570\u79bb\u7fa4\u70b9\u53ef\u80fd\u4ee3\u8868\u4ea4\u6613\u6a21\u5f0f\u5f02\u5e38\u7684\u80a1\u7968\uff0c\u5982\u53d7\u91cd\u5927\u65b0\u95fb\u4e8b\u4ef6\u6216\u516c\u53f8\u884c\u52a8\u5f71\u54cd\u7684\u80a1\u7968\u3002<\/p>\n\n\n\n
                \u56db\u3001\u4e86\u89e3\u76c8\u5229\u4e8b\u4ef6<\/strong><\/h2>\n\n\n\n
                \u76c8\u5229\u516c\u544a\u662f\u80a1\u7968\u4ea4\u6613\u5468\u671f\u4e2d\u7684\u5173\u952e\u65f6\u523b\u3002\u5f53\u5e02\u573a\u6d88\u5316\u6709\u5173\u516c\u53f8\u4e1a\u7ee9\u7684\u65b0\u4fe1\u606f\u65f6\uff0c\u8fd9\u4e9b\u5b63\u5ea6\u4e8b\u4ef6\u53ef\u80fd\u4f1a\u5f15\u53d1\u5267\u70c8\u7684\u4ef7\u683c\u6ce2\u52a8\u3002\u8ba9\u6211\u4eec\u6765\u63a2\u8ba8\u4e00\u4e0b\u8fd9\u4e9b\u65e5\u671f\u4e3a\u4f55\u91cd\u8981\uff0c\u4ee5\u53ca\u5982\u4f55\u8861\u91cf\u5b83\u4eec\u7684\u5f71\u54cd\uff1a<\/p>\n\n\n\n
                  \n
                • \u516c\u53f8\u5728\u5f00\u5e02\u524d\uff08”BMO”\uff09\u6216\u6536\u5e02\u540e\uff08”AMC”\uff09\u62a5\u544a\u8d22\u52a1\u4e1a\u7ee9\u3002<\/li>\n\n\n\n
                • \u80a1\u4ef7\u53cd\u5e94\u53d6\u51b3\u4e8e\u5b9e\u9645\u7ed3\u679c\u4e0e\u5e02\u573a\u9884\u671f\u7684\u5bf9\u6bd4\u60c5\u51b5\u3002<\/li>\n\n\n\n
                • \u8fd9\u4e9b\u4e8b\u4ef6\u5f80\u5f80\u4f1a\u5bfc\u81f4\u80a1\u4ef7\u51fa\u73b0\u6700\u5927\u7684\u5355\u65e5\u6ce2\u52a8\u3002<\/li>\n\n\n\n
                • \u516c\u544a\u53d1\u5e03\u540e\uff0c\u5267\u70c8\u6ce2\u52a8\u901a\u5e38\u4f1a\u6301\u7eed\u6570\u5929\u3002<\/li>\n<\/ul>\n\n\n\n
                  \u4e3a\u4e86\u91cf\u5316\u6536\u76ca\u7684\u5f71\u54cd\uff0c\u6211\u4eec\u5c06\u6bd4\u8f83\u6536\u76ca\u540e\u4ea4\u6613\u65e5\u4e0e\u6b63\u5e38\u4ea4\u6613\u65e5\u7684\u4ef7\u683c\u6ce2\u52a8\u6027\u3002\u6211\u4eec\u4f7f\u7528\u533a\u95f4-ATR \u6bd4\u7387\u6765\u8ba1\u7b97\uff0c\u8be5\u6bd4\u7387\u8861\u91cf\u7684\u662f\u4e0e\u5178\u578b\u7684\u6bcf\u65e5\u6ce2\u52a8\u76f8\u6bd4\uff0c\u4e00\u53ea\u80a1\u7968\u7684\u6ce2\u52a8\u5e45\u5ea6\u6709\u591a\u5927\uff1a<\/p>\n\n\n\n
                  from pandas._libs.tslibs.offsets import BDay\n\ndef get_next_trading_day(datetime):\n    bdays = BDay()\n    the_day_after = datetime + timedelta(days=1)\n    is_business_day = bdays.is_on_offset(the_day_after)\n    while not is_business_day:\n        the_day_after = the_day_after + timedelta(days=1)\n        is_business_day = bdays.is_on_offset(the_day_after)\n    return the_day_after\n\nfundamental_data_df = pd.DataFrame.from_dict(fundamental_data, orient='index').explode(\"earning_dates\").dropna()\n\ndef get_date_following_earning(date_str):\n    datetime_ = datetime.strptime(date_str, \"%Y-%m-%d %H:%M:%S\")\n    if datetime_.hour < 16:\n        return datetime_.date()\n    else:\n        return get_next_trading_day(datetime_).date()\n    \nfundamental_data_df[\"date_after_earnings\"] = fundamental_data_df[\"earning_dates\"].apply(get_date_following_earning)\nfundamental_data_df[\"date\"] = fundamental_data_df[\"earning_dates\"].apply(lambda x: datetime.strptime(x, \"%Y-%m-%d %H:%M:%S\").date())\ndf_index = df.index\ndf = df.merge(fundamental_data_df.reset_index()[[\"index\", \"date\", \"date_after_earnings\"]], left_on=[\"symbol\", \"date\"], right_on=[\"index\", \"date\"], how=\"left\").drop(columns=[\"index\"]).rename(columns={\"date_after_earnings\": \"is_earnings_day\"}).fillna(False).set_index(df_index)\ndf[\"is_earnings_day\"] = df[\"is_earnings_day\"].apply(lambda date: True if date else False)\n\n# range is defined as the intra-day price range of a stock\nrange = df.groupby([\"symbol\", \"date\"])[\"high\"].transform(\"max\") - df.groupby([\"symbol\", \"date\"])[\"low\"].transform(\"min\")\ndf[\"range_to_ATR_ratio\"] = range \/ df[\"close\"] \/ df[\"ATR\"]\nplot_agg = df.groupby([\"symbol\", \"is_earnings_day\"])[\"range_to_ATR_ratio\"].mean().reset_index()\n# plot boxplot of TR to ATR ratio on earnings days\nplt.figure(figsize=(12, 8))\nsns.boxplot(data=plot_agg, x=\"is_earnings_day\", y=\"range_to_ATR_ratio\")\nplt.title(\"Range to ATR ratio on earnings days\")\nplt.show()<\/code><\/pre>\n\n\n\n
                  \u4e0b\u56fe\u6e05\u695a\u5730\u663e\u793a\u4e86\u76c8\u5229\u516c\u544a\u5bf9\u4ef7\u683c\u6ce2\u52a8\u7684\u91cd\u5927\u5f71\u54cd\u3002\u8ba9\u6211\u4eec\u6765\u5206\u6790\u4e00\u4e0b\u6211\u4eec\u6240\u770b\u5230\u7684\uff1a<\/p>\n\n\n\n
                  \"\"<\/figure>\n\n\n\n
                  \u6b63\u5e38\u4ea4\u6613\u65e5\uff08\u5047\uff09\uff1a<\/strong><\/p>\n\n\n\n
                    \n
                  • \u4e2d\u503c\u8303\u56f4\/ATR \u6bd4\u7387\u7ea6\u4e3a 0.7-0.8<\/li>\n\n\n\n
                  • \u7d27\u5bc6\u7684\u56db\u5206\u4f4d\u6570\u533a\u95f4\u663e\u793a\u51fa\u4e00\u81f4\u7684\u884c\u4e3a<\/li>\n\n\n\n
                  • 0.5 \u4ee5\u4e0b\u7684\u5f02\u5e38\u503c\u5f88\u5c11<\/li>\n<\/ul>\n\n\n\n
                    \u76c8\u5229\u540e\u7684\u65e5\u5b50\uff08\u771f\u5b9e\uff09\uff1a<\/strong><\/p>\n\n\n\n
                      \n
                    • \u4e2d\u503c\u8303\u56f4\/ATR \u6bd4\u7387\u7ea6\u4e3a 1.0<\/li>\n\n\n\n
                    • \u56db\u5206\u4f4d\u6570\u8303\u56f4\u66f4\u5e7f\uff080.8 \u81f3 1.4\uff09<\/li>\n\n\n\n
                    • \u51e0\u4e2a\u6781\u7aef\u5f02\u5e38\u503c\u8d85\u8fc7\u6b63\u5e38\u8303\u56f4\u7684 2.5 \u500d<\/li>\n\n\n\n
                    • \u5e95\u90e8\u6676\u987b\u4ecd\u4e0e\u6b63\u5e38\u5929\u6570\u76f8\u4f3c\uff0c\u8868\u660e\u5e76\u975e\u6240\u6709\u6536\u76ca\u90fd\u4f1a\u5f15\u8d77\u6ce2\u52a8<\/li>\n<\/ul>\n\n\n\n
                      \u8fd9\u79cd\u53ef\u89c6\u5316\u6548\u679c\u8bc1\u5b9e\uff0c\u6536\u76ca\u65e5\u540e\u7684\u4ef7\u683c\u6ce2\u52a8\u901a\u5e38\u6bd4\u6b63\u5e38\u4ea4\u6613\u65e5\u9ad8\u51fa 40-50% \u5de6\u53f3\uff0c\u6709\u4e9b\u80a1\u7968\u7684\u6ce2\u52a8\u5e45\u5ea6\u662f\u5e73\u65f6\u7684\u4e24\u500d\u591a\u3002\u8fd9\u79cd\u660e\u663e\u7684\u5dee\u5f02\u4f7f\u6536\u76ca\u65e5\u6210\u4e3a\u6211\u4eec TFT \u6a21\u578b\u7684\u4e00\u4e2a\u57fa\u672c\u7279\u5f81\uff0c\u56e0\u4e3a\u5b83\u53ef\u4ee5\u5e2e\u52a9\u6211\u4eec\uff1a<\/p>\n\n\n\n
                        \n
                      • \u8c03\u6574\u9884\u6d4b\u7f6e\u4fe1\u533a\u95f4<\/li>\n\n\n\n
                      • \u8003\u8651\u6f5c\u5728\u7684\u66f4\u5927\u4ef7\u683c\u6ce2\u52a8<\/li>\n\n\n\n
                      • \u66f4\u597d\u5730\u6821\u51c6\u98ce\u9669\u7ba1\u7406\u53c2\u6570<\/li>\n<\/ul>\n\n\n\n
                        \u4e94\u3001\u4e86\u89e3\u6210\u4ea4\u91cf\u52a8\u6001<\/strong><\/h2>\n\n\n\n
                        \u6210\u4ea4\u91cf\u662f\u5e02\u573a\u6d3b\u52a8\u7684\u91cd\u8981\u6307\u6807\uff0c\u6709\u52a9\u4e8e\u9a8c\u8bc1\u4ef7\u683c\u8d70\u52bf\u548c\u8bc6\u522b\u91cd\u8981\u7684\u4ea4\u6613\u673a\u4f1a\u3002\u8ba9\u6211\u4eec\u6765\u5206\u6790\u4e24\u4e2a\u5173\u952e\u7684\u6210\u4ea4\u91cf\u6307\u6807\uff1a<\/p>\n\n\n\n
                        \u4ea4\u6613\u91cf\u57fa\u672c\u60c5\u51b5\uff1a<\/strong><\/p>\n\n\n\n
                          \n
                        • \u6307\u5728\u4e00\u4e2a\u65f6\u671f\u5185\u4ea4\u6362\u7684\u80a1\u4efd\u603b\u6570\u3002<\/li>\n\n\n\n
                        • \u9ad8\u6210\u4ea4\u91cf\u901a\u5e38\u9884\u793a\u7740 – \u91cd\u8981\u4ef7\u683c\u6c34\u5e73\u6b63\u5728\u786e\u7acb – \u5e02\u573a\u52a8\u6001\u65b0\u95fb\u6216\u4e8b\u4ef6 – \u8d8b\u52bf\u5ef6\u7eed\u7684\u53ef\u80fd\u6027\u589e\u52a0\u3002<\/li>\n<\/ul>\n\n\n\n
                          \u65e5\u5747\u4ea4\u6613\u91cf (ADV)\uff1a<\/strong><\/p>\n\n\n\n
                            \n
                          • \u6309\u6bcf\u65e5\u4ea4\u6613\u91cf\u7684 30 \u5929\u79fb\u52a8\u5e73\u5747\u503c\u8ba1\u7b97\u3002<\/li>\n\n\n\n
                          • \u4e3a\u6b63\u5e38\u8d38\u6613\u6d3b\u52a8\u63d0\u4f9b\u57fa\u7ebf\u3002<\/li>\n\n\n\n
                          • \u8ba1\u7b97\u516c\u5f0f\uff1aADV = \u5e73\u5747\u503c\uff08\u8fc7\u53bb 30 \u5929\u7684\u65e5\u4ea4\u6613\u91cf\uff09\u3002<\/li>\n<\/ul>\n\n\n\n
                            \u76f8\u5bf9\u5bb9\u91cf\uff08Rel Vol\uff09\uff1a<\/strong><\/p>\n\n\n\n
                              \n
                            • \u4e0e\u5178\u578b\u6c34\u5e73\u76f8\u6bd4\uff0c\u8861\u91cf\u5f53\u524d\u7684\u4ea4\u6613\u6d3b\u52a8\u3002<\/li>\n\n\n\n
                            • \u8ba1\u7b97\u516c\u5f0f\u4e3a\uff1a\u7d2f\u8ba1\u6210\u4ea4\u91cf\/\u65e5\u5747\u6210\u4ea4\u91cf\u3002<\/li>\n\n\n\n
                            • \u89e3\u91ca\uff1a – Rel Vol = 1.0\uff1a\u4ea4\u6613\u8282\u594f\u6b63\u5e38 – Rel Vol > 1.0\uff1a\u6bd4\u5e73\u65f6\u66f4\u6d3b\u8dc3 – Rel Vol < 1.0\uff1a\u6bd4\u5e73\u65f6\u66f4\u4e0d\u6d3b\u8dc3\u3002<\/li>\n\n\n\n
                            • \u65f6\u95f4\u654f\u611f\u6027\uff1a\u4e2d\u5348\u7684\u76f8\u5bf9\u6ce2\u52a8\u7387\u8d85\u8fc7 0.7 \u53ef\u80fd\u8868\u660e\u4ea4\u6613\u6d3b\u52a8\u9891\u7e41\uff08\u56e0\u4e3a\u4ea4\u6613\u65e5\u53ea\u8fc7\u53bb\u4e86\u4e00\u534a\uff09\u3002<\/li>\n<\/ul>\n\n\n\n
                              \u4f8b\u5982\uff0c\u5982\u679c\u4e00\u53ea\u80a1\u7968\u901a\u5e38\u6bcf\u5929\u4ea4\u6613 100 \u4e07\u80a1\uff1a<\/strong><\/p>\n\n\n\n
                                \n
                              • \u4e0a\u5348 10:00 \u524d 400,000 \u80a1 \u2192 \u76f8\u5bf9\u6210\u4ea4\u91cf = 0.4\uff08\u4e0a\u5348\u975e\u5e38\u6d3b\u8dc3\uff09<\/li>\n\n\n\n
                              • \u6536\u76d8\u65f6\u4e3a 150 \u4e07\u80a1 \u2192 \u76f8\u5bf9\u6210\u4ea4\u91cf = 1.5\uff08\u9ad8\u6210\u4ea4\u91cf\u65e5\uff09<\/li>\n<\/ul>\n\n\n\n
                                df_daily = get_daily_df(df,{\n    'volume': 'sum'\n})\ndf_daily = df_daily[df_daily[\"volume\"] > 0]\ndf_daily[\"average_volume\"] = df_daily.groupby(\"symbol\")[\"volume\"].rolling(window=30, min_periods=1).mean().reset_index(0, drop=True)\ndf_index = df.index\ndf = df.merge(df_daily[[\"symbol\", \"date\", \"average_volume\"]], on=[\"symbol\", \"date\"], how=\"left\").set_index(df_index)\ndf[\"rel_volume\"] = df.groupby([\"symbol\", \"date\"])[\"volume\"].cumsum() \/ df[\"average_volume\"]<\/code><\/pre>\n\n\n\n
                                \u8ba9\u6211\u4eec\u6765\u9a8c\u8bc1\u4e00\u4e2a\u91cd\u8981\u7684\u4ea4\u6613\u5047\u8bbe\uff1a\u6e05\u6668\u5f02\u5e38\u9ad8\u7684\u4ea4\u6613\u91cf\u662f\u5426\u9884\u793a\u7740\u5f53\u5929\u5269\u4f59\u65f6\u95f4\u7684\u6ce2\u52a8\u6027\u4f1a\u589e\u52a0\uff1f\u6211\u4eec\u5c06\u901a\u8fc7\u89c2\u5bdf\u4e00\u79cd\u7279\u5b9a\u7684\u60c5\u51b5\u6765\u7814\u7a76\u8fd9\u4e2a\u95ee\u9898\uff1a<\/p>\n\n\n\n
                                \u6211\u4eec\u7684\u6d4b\u8bd5\u6761\u4ef6\uff1a<\/strong><\/p>\n\n\n\n
                                  \n
                                • \u65f6\u95f4\uff1a\u4e0a\u5348 10:00\uff08\u4ea4\u6613\u7684\u524d\u534a\u5c0f\u65f6\uff09<\/li>\n\n\n\n
                                • \u4f53\u79ef\u9608\u503c\uff1a\u76f8\u5bf9\u4f53\u79ef > 0.5<\/li>\n\n\n\n
                                • \u8fd9\u610f\u5473\u7740\u4ec0\u4e48\uff1f\u53ea\u9700 30 \u5206\u949f\u5c31\u80fd\u5b8c\u6210\u534a\u5929\u7684\u5178\u578b\u4ea4\u6613\u91cf<\/li>\n\n\n\n
                                • \u6d4b\u91cf\uff1a\u6bd4\u8f83\u9ad8\u9500\u91cf\u65e5\u548c\u6b63\u5e38\u65e5\u7684\u4ef7\u683c\u8303\u56f4<\/li>\n<\/ul>\n\n\n\n
                                  high_rel_vol_at_10 = 0.5 # 50% of average daily volume traded in the first half hour\ndf[\"is_high_rel_early_candle\"] = (df[\"rel_volume\"] > high_rel_vol_at_10) & (df[\"hour\"] == 9)\ndf[\"high_rel_volume_day\"] = df.groupby([\"symbol\", \"date\"])[\"is_high_rel_early_candle\"].transform(\"max\")\n# plot boxplot of TR to ATR ratio on high rel volume days compared to regular days\nplot_agg = df.groupby([\"symbol\", \"high_rel_volume_day\"])[\"range_to_ATR_ratio\"].mean().reset_index()\nplt.figure(figsize=(12, 8))\nsns.boxplot(data=plot_agg, x=\"high_rel_volume_day\", y=\"range_to_ATR_ratio\")\nplt.title(\"Range to ATR ratio on high rel volume days\")\nplt.show()<\/code><\/pre>\n\n\n\n
                                  \"\"<\/figure>\n\n\n\n
                                  \u8fd9\u79cd\u53ef\u89c6\u5316\u65b9\u5f0f\u8bc1\u5b9e\u4e86\u6211\u4eec\u7684\u5047\u8bbe\uff1a\u5982\u679c\u5728\u524d 30 \u5206\u949f\u5185\u4ea4\u6613\u4e86\u4e00\u5929\u7684\u4e00\u534a\u4ea4\u6613\u91cf\uff0c\u6211\u4eec\u5c31\u53ef\u4ee5\u9884\u671f\uff1a<\/p>\n\n\n\n
                                    \n
                                  • \u6bd4\u4e00\u822c\u4ef7\u683c\u9ad8\u51fa\u8fd1\u4e00\u500d<\/li>\n\n\n\n
                                  • \u4ef7\u683c\u8d70\u52bf\u66f4\u52a0\u96be\u4ee5\u9884\u6d4b<\/li>\n\n\n\n
                                  • \u51fa\u73b0\u6781\u7aef\u52a8\u4f5c\u7684\u6982\u7387\u66f4\u9ad8<\/li>\n<\/ul>\n\n\n\n
                                    \u8ba9\u6211\u4eec\u6765\u68c0\u9a8c\u4e00\u4e0b\u4e00\u4e2a\u5e38\u89c1\u7684\u4ea4\u6613\u89c2\u5ff5\uff1a\u9ad8\u6210\u4ea4\u91cf\u4f1a\u9a8c\u8bc1\u4ef7\u683c\u8d70\u52bf\u3002\u8fd9\u79cd\u89c2\u70b9\u8ba4\u4e3a\uff0c\u6709\u5f3a\u52b2\u6210\u4ea4\u91cf\u652f\u6491\u7684\u8d70\u52bf\u66f4\u6709\u53ef\u80fd\u6301\u7eed\uff0c\u800c\u4f4e\u6210\u4ea4\u91cf\u7684\u8d70\u52bf\u5219\u53ef\u80fd\u9006\u8f6c\u3002\u6211\u4eec\u5c06\u901a\u8fc7\u4ee5\u4e0b\u65b9\u5f0f\u8fdb\u884c\u5206\u6790\uff1a<\/p>\n\n\n\n
                                    \u6211\u4eec\u7684\u8861\u91cf\u6807\u51c6\uff1a<\/strong><\/p>\n\n\n\n
                                      \n
                                    • \u521d\u59cb\u8d70\u52bf\uff1a\u4e0a\u5348 10:00 \u80a1\u4ef7\u4e0e\u5f00\u76d8\u4ef7\u76f8\u6bd4\u7684\u53d8\u5316<\/li>\n\n\n\n
                                    • \u7ee7\u7eed\uff1a\u4e0a\u5348 10:00 \u81f3\u4e2d\u5348 12:00 \u671f\u95f4\u7684\u5e73\u5747\u4ef7\u683c\u53d8\u5316<\/li>\n\n\n\n
                                    • \u6210\u4ea4\u91cf\u80cc\u666f\uff1a\u4e0a\u5348 10:00 \u7684\u76f8\u5bf9\u6210\u4ea4\u91cf<\/li>\n<\/ul>\n\n\n\n
                                      change_from_open_at_10 = df[(df[\"hour\"] == 10) & (df[\"minute\"] == 0)].groupby([\"symbol\", \"date\"])[\"change_from_open\"].first().reset_index()\nchange_from_open_10_12 = df[(df[\"hour\"] < 12) & (df[\"hour\"] >= 10)].groupby([\"symbol\", \"date\"])[\"change_from_open\"].mean().reset_index()\nrel_volume_at_10 = df[(df[\"hour\"] == 10) & (df[\"minute\"] == 0)][[\"symbol\", \"rel_volume\", \"date\", \"ATR\"]]<\/code><\/pre>\n\n\n\n
                                      joined = change_from_open_at_10.merge(change_from_open_10_12, on=[\"symbol\", \"date\"], suffixes=(\"_10\", \"_10_12_mean\"))\njoined = joined.merge(rel_volume_at_10, on=[\"symbol\", \"date\"]).rename(columns={\"rel_volume\": \"rel_volume_10\"})\n# we want to focus on days where the move is large in the first half hour of the day, like days where the change from open is greater than the ATR\njoined = joined[joined[\"change_from_open_10\"].abs() > joined[\"ATR\"]]\nchange_bins = [-np.inf, -0.10, -0.075, -0.05, -0.025, 0.025, 0.05, 0.075, 0.10, np.inf]\nchange_labels = ['<-10', '-10 to -7.5', '-7.5 to -5', '-5 to -2.5', '-2.5 to 2.5', '2.5 to 5', '5 to 7.5', '7.5 to 10', '>10']\n# Bin the change_from_open_10 values\njoined['change_bin'] = pd.cut(joined['change_from_open_10'], bins=change_bins, labels=change_labels, include_lowest=True)\njoined[\"is_positive_change\"] = joined[\"change_from_open_10\"] > 0\njoined[\"continuation_move\"] = joined[\"change_from_open_10_12_mean\"] - joined[\"change_from_open_10\"]\nprint(f\"Average rel volume at 10:00: {joined['rel_volume_10'].mean()}\")\n\n# Scatter plot for negative days\nplt.figure(figsize=(10, 6))\nsns.scatterplot(x='rel_volume_10', y='continuation_move', hue='change_bin', data=joined, palette='coolwarm', legend='full')\nplt.title(\"Scatter Plot of Continuation vs Relative Volume with Change Bins\")\nplt.xlabel(\"Relative Volume at 10 AM\")\nplt.ylabel(\"Continuation (10 AM - 12 PM)\")\nplt.legend(title='Change from Open at 10:00')\nplt.vlines(0.65, min(joined['continuation_move']), max(joined['continuation_move']), colors='r', linestyles='dashed')\nplt.show()<\/code><\/pre>\n\n\n\n
                                      \u6301\u7eed\u65f6\u95f4\u4e0e\u76f8\u5bf9\u4f53\u79ef\u7684\u6563\u70b9\u56fe\uff0c\u5e26\u53d8\u5316\u5206\u6bb5\uff1a<\/p>\n\n\n\n
                                      \"\"<\/figure>\n\n\n\n
                                      \u8ba9\u6211\u4eec\u6765\u5206\u6790\u4e00\u4e0b\u8fd9\u4e2a\u590d\u6742\u4f46\u5374\u80fd\u63ed\u793a\u65e9\u671f\u4ea4\u6613\u91cf\u4e0e\u4ef7\u683c\u53d8\u52a8\u6a21\u5f0f\u4e4b\u95f4\u5173\u7cfb\u7684\u6563\u70b9\u56fe\uff1a<\/p>\n\n\n\n