{"id":1500,"date":"2025-02-02T10:20:23","date_gmt":"2025-02-02T02:20:23","guid":{"rendered":"https:\/\/crab.cool\/?p=1500"},"modified":"2025-02-02T10:20:23","modified_gmt":"2025-02-02T02:20:23","slug":"ds%e6%8f%90%e4%be%9b%e7%9a%84%e9%87%8f%e5%8c%96%e6%8a%95%e8%b5%84%e6%a1%86%e6%9e%b6","status":"publish","type":"post","link":"https:\/\/crab.cool\/index.php\/2025\/02\/02\/ds%e6%8f%90%e4%be%9b%e7%9a%84%e9%87%8f%e5%8c%96%e6%8a%95%e8%b5%84%e6%a1%86%e6%9e%b6\/","title":{"rendered":"DS\u63d0\u4f9b\u7684\u91cf\u5316\u6295\u8d44\u6846\u67b6"},"content":{"rendered":"

\u53ef\u4ee5\u57fa\u4e8e\u91cf\u5316\u6295\u8d44\u7684\u65b9\u6cd5\u5efa\u7acb\u4e00\u4e2a\u9488\u5bf9\u67d0\u53ea\u80a1\u7968\u7684\u6295\u8d44\u7b56\u7565\u6a21\u578b\u3002\u4ee5\u4e0b\u662f\u4e00\u4e2a\u5b8c\u6574\u6846\u67b6<\/strong>\uff08\u4ee5Python\u4e3a\u4f8b\uff09\uff0c\u5305\u542b\u6570\u636e\u83b7\u53d6\u3001\u7b56\u7565\u8bbe\u8ba1\u3001\u56de\u6d4b\u548c\u98ce\u9669\u7ba1\u7406\u6a21\u5757\uff1a<\/p>\n

\n

1. \u6570\u636e\u51c6\u5907<\/strong><\/h3>\n
import pandas as pd\nimport numpy as np\nimport yfinance as yf  # \u96c5\u864e\u8d22\u7ecf\u6570\u636e\u63a5\u53e3\n\n# \u83b7\u53d6\u80a1\u7968\u5386\u53f2\u6570\u636e\uff08\u793a\u4f8b\uff1a\u5b81\u5fb7\u65f6\u4ee3\uff0cA\u80a1\u4ee3\u7801300750.SZ\uff09\nstock_data = yf.download("300750.SZ", start="2018-01-01", end="2023-12-31")\n# \u6dfb\u52a0\u5b8f\u89c2\u7ecf\u6d4e\u6570\u636e\uff08\u4f8b\u5982\u9502\u7535\u6c60\u539f\u6750\u6599\u78b3\u9178\u9502\u4ef7\u683c\uff09\nlithium_price = yf.download("LIT", start="2018-01-01", end="2023-12-31")  # \u5168\u7403\u9502ETF<\/code><\/pre>\n
\n
2. \u7b56\u7565\u903b\u8f91\u8bbe\u8ba1<\/strong><\/h3>\n
(1) \u591a\u56e0\u5b50\u9009\u80a1\u6a21\u578b<\/strong><\/h4>\n
\u7ed3\u5408\u57fa\u672c\u9762\u3001\u6280\u672f\u9762\u548c\u884c\u4e1a\u5468\u671f\u56e0\u5b50\uff1a<\/p>\n
# \u56e0\u5b501\uff1a\u52a8\u6001\u5e02\u76c8\u7387\uff08TTM\uff09\u5206\u4f4d\u6570\uff08\u4f4e\u4f30\u503c\uff09\ndef calculate_pe_rank(data):\n    data['PE_rank'] = data['Close'].rolling(60).apply(lambda x: (x[-1] - x.min()) \/ (x.max() - x.min()))\n\n# \u56e0\u5b502\uff1a\u9502\u7535\u6c60\u884c\u4e1a\u666f\u6c14\u5ea6\uff08\u901a\u8fc7\u78b3\u9178\u9502\u4ef7\u683c\u8d8b\u52bf\u4ee3\u7406\uff09\ndef lithium_trend(stock_data, lithium_price):\n    stock_data['Lithium_3M_Return'] = lithium_price['Close'].pct_change(60)  # 3\u4e2a\u6708\u9502\u4ef7\u6da8\u5e45\n\n# \u56e0\u5b503\uff1a\u6280\u672f\u9762\u52a8\u91cf\uff0860\u65e5\u6da8\u5e45\uff09\nstock_data['Momentum_60D'] = stock_data['Close'].pct_change(60)\n\n# \u56e0\u5b504\uff1a\u653f\u7b56\u4e8b\u4ef6\uff08\u4eba\u5de5\u6807\u6ce8\u5173\u952e\u653f\u7b56\u65e5\u671f\uff0c\u5982\u8865\u8d34\u9000\u5761\u3001\u78b3\u4e2d\u548c\u76ee\u6807\u53d1\u5e03\uff09\npolicy_dates = {'2020-11-02': '\u4e2d\u56fd\u65b0\u80fd\u6e90\u89c4\u5212', '2021-07-01': '\u6b27\u6d32\u78b3\u5173\u7a0e\u63d0\u6848'}<\/code><\/pre>\n
(2) \u4ea4\u6613\u4fe1\u53f7\u751f\u6210<\/strong><\/h4>\n
# \u7efc\u5408\u4fe1\u53f7\uff08\u793a\u4f8b\uff1a\u7b80\u5355\u52a0\u6743\uff09\nstock_data['Signal'] = (\n    0.4 * (1 - stock_data['PE_rank']) +  # \u4f4e\u4f30\u503c\u6743\u91cd40%\n    0.3 * stock_data['Lithium_3M_Return'] +  # \u884c\u4e1a\u666f\u6c14\u5ea630%\n    0.3 * stock_data['Momentum_60D']  # \u52a8\u91cf30%\n)\n# \u751f\u6210\u4e70\u5356\u4fe1\u53f7\uff08\u9608\u503c\u89e6\u53d1\uff09\nstock_data['Position'] = np.where(stock_data['Signal'] > 0.6, 1,  # \u4e70\u5165\n                         np.where(stock_data['Signal'] < 0.3, -1, 0))  # \u5356\u51fa<\/code><\/pre>\n
\n
3. \u56de\u6d4b\u4e0e\u7ee9\u6548\u8bc4\u4f30<\/strong><\/h3>\n
# \u8ba1\u7b97\u7b56\u7565\u6536\u76ca\nstock_data['Strategy_Return'] = stock_data['Position'].shift(1) * stock_data['Close'].pct_change()\n\n# \u5bf9\u6bd4\u57fa\u51c6\uff08\u4e70\u5165\u6301\u6709\uff09\nstock_data['Benchmark_Return'] = stock_data['Close'].pct_change()\n\n# \u8ba1\u7b97\u7d2f\u8ba1\u6536\u76ca\nstock_data[['Cum_Strategy', 'Cum_Benchmark']] = (\n    (1 + stock_data[['Strategy_Return', 'Benchmark_Return']]).cumprod()\n)\n\n# \u7ee9\u6548\u6307\u6807\ndef calculate_performance(returns):\n    sharpe_ratio = returns.mean() \/ returns.std() * np.sqrt(252)  # \u5e74\u5316\u590f\u666e\u6bd4\u7387\n    max_drawdown = (returns.cumsum().expanding().max() - returns.cumsum()).max()  # \u6700\u5927\u56de\u64a4\n    return sharpe_ratio, max_drawdown\n\nsharpe, drawdown = calculate_performance(stock_data['Strategy_Return'].dropna())<\/code><\/pre>\n
\n
4. \u98ce\u9669\u7ba1\u7406\u6a21\u5757<\/strong><\/h3>\n
(1) \u52a8\u6001\u4ed3\u4f4d\u63a7\u5236<\/strong><\/h4>\n
# \u6ce2\u52a8\u7387\u8c03\u6574\u4ed3\u4f4d\uff08\u4f8b\u5982\uff1aATR\u6307\u6807\uff09\nstock_data['ATR_14D'] = talib.ATR(stock_data['High'], stock_data['Low'], stock_data['Close'], timeperiod=14)\nstock_data['Position_Size'] = 0.02 \/ (stock_data['ATR_14D'] \/ stock_data['Close'])  # \u56fa\u5b9a\u98ce\u9669\u6bd4\u4f8b<\/code><\/pre>\n
(2) \u6b62\u635f\u673a\u5236<\/strong><\/h4>\n
# \u79fb\u52a8\u6b62\u635f\uff08\u8ddf\u8e2a\u8fc7\u53bbN\u65e5\u6700\u4f4e\u4ef7\uff09\nstop_loss_window = 20\nstock_data['Trailing_Stop'] = stock_data['Close'].rolling(stop_loss_window).min() * 0.95  # 5%\u6b62\u635f\u7ebf<\/code><\/pre>\n
\n
5. \u6a21\u578b\u4f18\u5316\u4e0e\u9a8c\u8bc1<\/strong><\/h3>\n