{"id":1727,"date":"2024-12-20T07:08:00","date_gmt":"2024-12-19T23:08:00","guid":{"rendered":"https:\/\/blog.laoyulaoyu.top\/?p=1727"},"modified":"2024-11-26T16:08:39","modified_gmt":"2024-11-26T08:08:39","slug":"%e6%8f%ad%e7%a7%98%e5%a4%96%e6%b1%87%e5%b8%82%e5%9c%ba%e7%9a%84%e9%9a%90%e7%a7%98%e9%87%91%e7%9f%bf%ef%bc%9a%e7%94%a8%e7%bb%9f%e8%ae%a1%e5%a5%97%e5%88%a9%e7%ad%96%e7%95%a5%e5%ae%9e","status":"publish","type":"post","link":"https:\/\/laoyulaoyu.com\/index.php\/2024\/12\/20\/%e6%8f%ad%e7%a7%98%e5%a4%96%e6%b1%87%e5%b8%82%e5%9c%ba%e7%9a%84%e9%9a%90%e7%a7%98%e9%87%91%e7%9f%bf%ef%bc%9a%e7%94%a8%e7%bb%9f%e8%ae%a1%e5%a5%97%e5%88%a9%e7%ad%96%e7%95%a5%e5%ae%9e\/","title":{"rendered":"\u63ed\u79d8\u5916\u6c47\u5e02\u573a\u7684\u201c\u9690\u79d8\u91d1\u77ff\u201d\uff1a\u7528\u7edf\u8ba1\u5957\u5229\u7b56\u7565\u5b9e\u73b0\u7a33\u5b9a\u76c8\u5229\uff08\u9644\u5b8c\u6574\u53ef\u8fd0\u884c\u4ee3\u7801\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\u8fd9\u7bc7\u6587\u7ae0\u4e2d\uff0c\u6211\u5c06\u5e26\u4f60\u6df1\u5165\u63a2\u7d22\u5916\u6c47\u5e02\u573a\u7684\u7edf\u8ba1\u5957\u5229\u7b56\u7565\u3002\u901a\u8fc7\u5206\u6790\u9876\u7ea7\u5bf9\u51b2\u57fa\u91d1\u7684\u64cd\u4f5c\u548c\u5b9e\u7528\u7684Python\u4ee3\u7801\u793a\u4f8b\uff08\u9644\u5b8c\u6574\u53ef\u8fd0\u884c\u4ee3\u7801\uff09\uff0c\u6211\u5c06\u5c55\u793a\u5982\u4f55\u5728\u8d27\u5e01\u5bf9\u95f4\u5bfb\u627e\u76c8\u5229\u673a\u4f1a\uff0c\u5e76\u63a2\u8ba8\u7b56\u7565\u7684\u98ce\u9669\u7ba1\u7406\u3002\u8ba9\u6211\u4eec\u4e00\u8d77\u6765\u6316\u6398\u8fd9\u4e2a\u5916\u6c47\u5e02\u573a\u4e2d\u7684\u9690\u79d8\u91d1\u77ff\uff0c\u5f00\u542f\u7a33\u5b9a\u76c8\u5229\u4e4b\u8def\u3002<\/pre>\n<\/blockquote>\n\n\n\n
\u4e00\u3001\u53d1\u73b0\u5916\u6c47\u4ea4\u6613\u4e2d\u7684\u7edf\u8ba1\u5957\u5229<\/h2>\n\n\n\n
\u6709\u6ca1\u6709\u60f3\u8fc7\u5bf9\u51b2\u57fa\u91d1\u662f\u5982\u4f55\u4e0d\u53d7\u5e02\u573a\u8d70\u5411\u5f71\u54cd\u800c\u8d5a\u94b1\u7684\uff1f\u8ba9\u6211\u63ed\u793a\u4ed6\u4eec\u7684\u79d8\u5bc6\u6b66\u5668\u4e4b\u4e00\uff1a\u7edf\u8ba1\u5957\u5229\u3002\u8fd9\u4e0d\u662f\u7b80\u5355\u7684\u201c\u4f4e\u4e70\u9ad8\u5356\u201d\uff0c\u800c\u662f\u4e00\u79cd\u6316\u6398\u5927\u591a\u6570\u4ea4\u6613\u8005\u672a\u66fe\u5bdf\u89c9\u7684\u4e00\u79cd\u6570\u5b66\u7b56\u7565\u3002<\/p>\n\n\n\n
\u4ec0\u4e48\u662f\u914d\u5bf9\u4ea4\u6613\u7edf\u8ba1\u5957\u5229\u7b56\u7565\uff1f<\/h3>\n\n\n\n
\u914d\u5bf9\u4ea4\u6613\u7edf\u8ba1\u5957\u5229\u662f\u4e00\u79cd\u57fa\u4e8e\u7edf\u8ba1\u5b66\u539f\u7406\u7684\u4ea4\u6613\u7b56\u7565\uff0c\u65e8\u5728\u5229\u7528\u76f8\u5173\u8d44\u4ea7\u4e4b\u95f4\u7684\u4ef7\u683c\u5dee\u5f02\u8fdb\u884c\u83b7\u5229\u3002\u8be5\u7b56\u7565\u901a\u5e38\u6d89\u53ca\u9009\u62e9\u4e00\u5bf9\u9ad8\u5ea6\u76f8\u5173\u7684\u91d1\u878d\u5de5\u5177\uff08\u5982\u80a1\u7968\u3001ETF\u6216\u5916\u6c47\u7b49\u8d44\u4ea7\uff09\uff0c\u5e76\u5728\u5b83\u4eec\u7684\u4ef7\u683c\u5173\u7cfb\u504f\u79bb\u5386\u53f2\u5747\u8861\u65f6\u8fdb\u884c\u4ea4\u6613\u3002\u8fd9\u79cd\u65b9\u6cd5\u7684\u6838\u5fc3\u5728\u4e8e\uff0c\u968f\u7740\u65f6\u95f4\u7684\u63a8\u79fb\uff0c\u4ef7\u683c\u5173\u7cfb\u4f1a\u56de\u5f52\u5230\u5176\u5386\u53f2\u5e73\u5747\u6c34\u5e73\uff0c\u4ece\u800c\u4e3a\u4ea4\u6613\u8005\u63d0\u4f9b\u83b7\u5229\u673a\u4f1a\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u914d\u5bf9\u4ea4\u6613\u7684\u57fa\u672c\u539f\u7406\u662f\u5229\u7528\u4e24\u79cd\u8d44\u4ea7\u4e4b\u95f4\u7684\u4ef7\u683c\u5173\u7cfb\u3002\u4f8b\u5982\uff0c\u5f53A\u8d44\u4ea7\u7684\u4ef7\u683c\u4e0a\u6da8\u800cB\u8d44\u4ea7\u7684\u4ef7\u683c\u672a\u80fd\u540c\u6b65\u4e0a\u6da8\u65f6\uff0c\u4ea4\u6613\u8005\u53ef\u80fd\u4f1a\u5356\u7a7aA\u8d44\u4ea7\u5e76\u4e70\u5165B\u8d44\u4ea7\uff0c\u671f\u5f85\u4ef7\u683c\u5173\u7cfb\u6700\u7ec8\u56de\u5f52\u6b63\u5e38\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u5728\u5206\u6790\u5916\u6c47\u5e02\u573a\u4ea4\u6613\u6570\u636e\u65f6\uff0c\u6211\u5076\u7136\u53d1\u73b0\u4e86\u4e00\u4e2a\u73b0\u8c61\uff1a\u6709\u4e9b\u8d27\u5e01\u5bf9\u7684\u8d70\u52bf\u5c31\u50cf\u5929\u5e73\u4e24\u7aef\u7684\u781d\u7801\u4e00\u6837\u3002\u4ee5 DKK&JPY\uff08\u4e39\u9ea6\u514b\u6717\u5bf9\u65e5\u5143\uff09\u548c EUR&JPY\uff08\u6b27\u5143\u5bf9\u65e5\u5143\uff09\u4e3a\u4f8b\uff0c\u5b83\u4eec\u4e4b\u95f4\u7684\u5173\u7cfb\u6709\u4e25\u8c28\u7684\u6570\u5b66\u8868\u8fbe\u3002\u63a5\u4e0b\u6765\u8ba9\u6211\u4eec\u6765\u5206\u6790\u4e00\u4e0b\u53d1\u73b0\u8fd9\u4e9b\u6570\u5b57\uff1a<\/p>\n\n\n\n
# The magic numbers\nP-Value: 0.0001 (99.99% confidence!)\nHedge Ratio: 0.1325\nHalf-Life: 0.7 days<\/code><\/pre>\n\n\n\n
\u9996\u5148\uff0c\u8ba9\u6211\u4eec\u6765\u770b\u770b\u6211\u4eec\u662f\u5982\u4f55\u627e\u5230\u8fd9\u4e9b\u201d\u9ec4\u91d1\u201c\u914d\u5bf9\u7684\uff1a<\/p>\n\n\n\n
def test_cointegration(pair1, pair2, data, significance_level=0.05):\n    try:\n        # Calculate the optimal hedge ratio\n        Y = data[pair1].values\n        X = data[pair2].values\n        X = np.vstack([X, np.ones(len(X))]).T\n        beta = np.linalg.lstsq(X, Y, rcond=None)[0][0]\n        \n        # Find the spread\n        spread = data[pair1] - beta * data[pair2]\n        \n        # Test for cointegration\n        result = adfuller(spread.dropna())\n        p_value = result[1]\n        \n        # Calculate mean reversion speed\n        spread_lag = spread.shift(1)\n        spread_diff = spread - spread_lag\n        model = np.polyfit(spread_lag.dropna(), spread_diff.dropna(), 1)\n        half_life = -np.log(2) \/ model[0] if model[0] < 0 else np.inf\n        \n        return (p_value < significance_level, p_value, beta, half_life)\n    except Exception as e:\n        return (False, 1.0, 0, np.inf)<\/code><\/pre>\n\n\n\n
\u6b65\u9aa4 1\uff1a\u67e5\u627e\u5173\u7cfb<\/h3>\n\n\n\n
beta = np.linalg.lstsq(X, Y, rcond=None)[0][0]<\/code><\/pre>\n\n\n\n
\u5212\u51fa\u8fd9\u6761\u7ebf\u53ef\u4ee5\u627e\u5230\u8d27\u5e01\u5bf9\u4e4b\u95f4\u7684\u5b8c\u7f8e\u6bd4\u4f8b\u3002<\/p>\n\n\n\n
\u6b65\u9aa4 2\uff1a\u6d4b\u8bd5\u5f3a\u5ea6<\/h3>\n\n\n\n
result = adfuller(spread.dropna())\np_value = result[1]<\/code><\/pre>\n\n\n\n
\u8fd9\u5c31\u597d\u6bd4\u6d4b\u8bd5\u6865\u6881\u7684\u5f3a\u5ea6\u3002\u6211\u4eec\u7684 p \u503c\u4e3a 0.0001\uff0c\u8bf4\u660e\u8fd9\u5ea7\u6865\u662f\u7531\u949b\u5408\u91d1\u5236\u6210\u7684\u3002<\/p>\n\n\n\n
\u6b65\u9aa4 3\uff1a\u6d4b\u91cf\u901f\u5ea6 <\/h3>\n\n\n\n
 # Calculate half-life of mean reversion\n        spread_lag = spread.shift(1) #Calculate the Lagged Spread\n        spread_diff = spread - spread_lag #Calculate the Difference in Spread\n        spread_lag = spread_lag.dropna() #  ignoring any NaN values\n        spread_diff = spread_diff.dropna()\n  # This fits a linear model to the lagged spread and the spread difference\n        model = np.polyfit(spread_lag, spread_diff, 1)\n  # This calculates the half-life based on the slope of the linear model\n        half_life = -np.log(2) \/ model[0] if model[0] < 0 else np.inf<\/code><\/pre>\n\n\n\n
\u8fd9\u5c31\u544a\u8bc9\u6211\u4eec\uff0c\u6211\u4eec\u7684\u914d\u5bf9\u5728\u5206\u79bb\u540e\u91cd\u65b0\u805a\u5728\u4e00\u8d77\u7684\u901f\u5ea6\u6709\u591a\u5feb\u30020.7 \u5929\u610f\u5473\u7740\u5b83\u4eec\u5c31\u50cf\u78c1\u94c1\u4e00\u6837\uff0c\u4e0d\u4f1a\u5206\u5f00\u592a\u4e45\uff01<\/p>\n\n\n\n
\u4e8c\u3001\u4ea4\u6613\u7b56\u7565<\/h2>\n\n\n\n
<\/p>\n\n\n\n
2.1 \u7b56\u7565\u5b9e\u73b0<\/h3>\n\n\n\n
\u73b0\u5728\uff0c\u4ee4\u4eba\u5174\u594b\u7684\u5730\u65b9\u6765\u4e86……\u5f53 DKKJPY \u548c EURJPY \u51fa\u73b0\u5206\u6b67\u65f6\uff0c\u6211\u4eec\uff1a<\/p>\n\n\n\n