{"id":4855,"date":"2026-09-11T07:33:00","date_gmt":"2026-09-10T23:33:00","guid":{"rendered":"https:\/\/laoyulaoyu.com\/?p=4855"},"modified":"2026-09-04T16:43:38","modified_gmt":"2026-09-04T08:43:38","slug":"%e7%ad%96%e7%95%a5%e5%b9%b4%e5%8c%961-66%ef%bc%8c%e8%b7%91%e8%be%93spy%e4%b9%9d%e4%b8%aa%e7%82%b9%ef%bc%8c%e6%88%91%e5%8d%b4%e6%b2%a1%e5%88%a0%e5%ae%83%ef%bc%88%e9%99%84%e5%ae%8c%e6%95%b4%e6%ba%90","status":"publish","type":"post","link":"https:\/\/laoyulaoyu.com\/index.php\/2026\/09\/11\/%e7%ad%96%e7%95%a5%e5%b9%b4%e5%8c%961-66%ef%bc%8c%e8%b7%91%e8%be%93spy%e4%b9%9d%e4%b8%aa%e7%82%b9%ef%bc%8c%e6%88%91%e5%8d%b4%e6%b2%a1%e5%88%a0%e5%ae%83%ef%bc%88%e9%99%84%e5%ae%8c%e6%95%b4%e6%ba%90\/","title":{"rendered":"\u7b56\u7565\u5e74\u53161.66%\uff0c\u8dd1\u8f93SPY\u4e5d\u4e2a\u70b9\uff0c\u6211\u5374\u6ca1\u5220\u5b83\uff08\u9644\u5b8c\u6574\u6e90\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

\u6587\u4e2d\u793a\u4f8b\u4ec5\u7528\u4e8e\u6280\u672f\u8ba8\u8bba\uff0c\u4e0d\u6784\u6210\u4efb\u4f55\u64cd\u4f5c\u5efa\u8bae\u3002
\u91cf\u5316\u7b56\u7565\u5f00\u53d1\u5e94\u4ee5\u5b66\u4e60\u548c\u6280\u672f\u4ea4\u6d41\u4e3a\u76ee\u7684\u3002
\u672c\u53f7\u4e0d\u8350\u80a1\u3001\u4e0d\u5356\u8bfe\u3001\u4e0d\u627f\u8bfa\u6536\u76ca\u3002
\u5e02\u573a\u6709\u98ce\u9669\uff0c\u8bf7\u5408\u6cd5\u5408\u89c4\u6295\u8d44\u3002<\/p>\n<\/blockquote>\n\n\n\n

\u672c\u6587\u7ea6 7000 \u5b57 | \u9884\u8ba1\u9605\u8bfb 15-18 \u5206\u949f\uff0c\u6587\u540e\u6709\u5b8c\u6574\u4ee3\u7801\uff0c\u5efa\u8bae\u5148\u6536\u85cf\u3002<\/p>\n\n\n\n

\u00b7 \u00b7 \u00b7<\/p>\n\n\n\n

\u6211\u5728\u67d0\u89c6\u9891\u5e73\u53f0\u5237\u5230\u4e00\u4e2a\u53eb Rubber Band \u7684\u5747\u503c\u56de\u5f52\u7cfb\u7edf\uff0c\u4e3b\u8bb2\u4eba\u62a5\u4e86\u4e09\u4e2a\u6570\uff1a\u7b56\u7565\u5728SPY \u4e0a\u8dd1\uff0c\u6bcf\u7b14\u5e73\u5747\u8d5a 0.66%\uff0c\u5e74\u5316 6.4%\uff0c\u5728\u5e02\u65f6\u95f4\u53ea\u6709 14%\uff0c\u7136\u540e\u8bdd\u950b\u4e00\u8f6c\uff0c\u6b7b\u6d3b\u4e0d\u80af\u7ed9\u89c4\u5219\uff0c\u53ea\u8bf4”\u6bd4\u7b2c\u4e00\u4e2a\u7cfb\u7edf\u7a0d\u5fae\u590d\u6742\u4e00\u70b9”\u3002<\/p>\n\n\n\n

\u6211\u611f\u89c9\u8fd9\u4e2a\u91cc\u9762\u6709\u95ee\u9898\uff0c\u7167\u7740\u628a\u5b83\u91cd\u5efa\u4e86\u51fa\u6765\uff0c\u8dd1\u5b8c SPY \u4e09\u5341\u4e09\u5e74\u534a\u7684\u65e5\u7ebf\uff0c\u5e74\u5316\u53ea\u6709 1.66%\uff0c\u800c\u540c\u671f\u4e70\u5165\u6301\u6709\u662f 10.84%\uff0c\u6309\u6700\u5e38\u89c1\u7684\u6807\u51c6\uff0c\u8fd9\u4e2a\u7b56\u7565\u662f\u8be5\u8fdb\u56de\u6536\u7ad9\u7684\u5931\u8d25\u54c1\u3002<\/p>\n\n\n\n

\u4f46\u6211\u5728\u6d4b\u8bd5\u4e2d\u53d1\u73b0\uff0c\u8fd9\u4e2a\u7b56\u7565\u7684\u5e02\u573a\u655e\u53e3\u53ea\u6709 2%\uff0c33 \u5e74\u91cc\u53ea\u505a\u4e86 79 \u7b14\u4ea4\u6613\u3001\u80dc\u7387\u5374\u6709 83.5%\uff0c\u6309\u5b9e\u9645\u90e8\u7f72\u7684\u8d44\u672c\u5e74\u5316\u7ea6 128%\uff0c\u5b83\u6bcf\u5e74\u8f93\u7ed9\u4e70\u5165\u6301\u6709 9 \u4e2a\u70b9\uff0c\u4ec5\u4ec5\u662f\u56e0\u4e3a 98% \u7684\u65f6\u95f4\u6839\u672c\u4e0d\u5728\u573a\u3002<\/p>\n\n\n\n

\u8fd9\u7bc7\u6211\u60f3\u8bb2\u4e09\u4ef6\u4e8b\uff1a\u600e\u4e48\u4ece\u4e00\u7ec4\u6c47\u603b\u7edf\u8ba1\u53cd\u63a8\u51fa\u4e00\u4e2a\u7b56\u7565\uff0c\u53c8\u4e3a\u4ec0\u4e48\u53cd\u63a8\u51fa\u6765\u7684\u53ea\u662f”\u4e00\u4e2a”\u7b56\u7565\uff1b\u600e\u4e48\u53bb\u8bfb CAGR \u4e4b\u5916\u7684\u6570\u5b57\uff1b\u4ee5\u53ca\u4e00\u4e2a\u6bcf\u5e74\u53ea\u51fa\u624b 2.3 \u6b21\u7684\u7cfb\u7edf\uff0c\u5230\u5e95\u8be5\u600e\u4e48\u7528\u3002<\/p>\n\n\n\n

\n

\u4ee3\u7801\u5168\u5728\u4e0b\u9762\uff0c\u6309\u7ae0\u8282\u62c6\u6210\u4e94\u6bb5\uff0c\u6587\u672b\u6709\u62fc\u597d\u7684\u5b8c\u6574\u811a\u672c\u3002\u6570\u636e SPY \u65e5\u7ebf\u53d6\u81ea Yahoo Finance\uff0c\u56de\u6d4b\u7528 vectorbt\uff0c\u6210\u672c 1bp \u624b\u7eed\u8d39\u52a0 2bp \u6ed1\u70b9\u3002<\/p>\n<\/blockquote>\n\n\n\n

\u4e8b\u60c5\u7684\u8d77\u5934\uff1a\u4e00\u4e2a\u4e0d\u80af\u7ed9\u89c4\u5219\u7684\u7cfb\u7edf<\/strong><\/h2>\n\n\n\n

\u89c6\u9891\u8bb2\u5f97\u5934\u5934\u662f\u9053\uff0c\u7edf\u8ba1\u4e5f\u6f02\u4eae\uff0c\u53ef\u4e3b\u8bb2\u4eba\u4ece\u5934\u5230\u5c3e\u6ca1\u7ed9\u5165\u573a\u6761\u4ef6\uff0c\u53ea\u6709\u4e00\u4e2a\u540d\u5b57\u548c\u4e09\u4e2a\u6570\u5b57\u3002\u540d\u5b57\u8d77\u5f97\u633a\u5999\uff0c\u6838\u5fc3\u601d\u60f3\u5168\u85cf\u5728\u91cc\u5934\uff1a\u62c9\u4f38\u6a61\u76ae\u7b4b\uff0c\u7b49\u5b83\u56de\u5f39\uff0c\u8dcc\u5f97\u8d8a\u72e0\uff0c\u5f39\u5f97\u8d8a\u9ad8\uff0c\u5747\u503c\u56de\u5f52\u8fd9\u4e00\u5957\u7684\u8bf4\u670d\u529b\u4ece\u6765\u4e0d\u9760\u903b\u8f91\uff0c\u9760\u7684\u5c31\u662f”\u8dcc\u591a\u4e86\u603b\u4f1a\u56de\u6765”\u8fd9\u4e2a\u76f4\u89c9\u3002<\/p>\n\n\n\n

\u53ef\u76f4\u89c9\u6bd5\u7adf\u4e0d\u80fd\u5f53\u7b56\u7565\u7528\uff0c\u6211\u624b\u5934\u53ea\u6709\u4e09\u6837\u4e1c\u897f\uff1a\u6807\u7684 SPY\u3001\u5355\u7b14 0.66%\u3001\u5728\u5e02 14%\uff0c\u5176\u4f59\u89c4\u5219\uff0c\u5168\u5f97\u81ea\u5df1\u642d\u3002<\/p>\n\n\n\n

\u89c4\u5219\uff1a\u56db\u9053\u95f8\u95e8<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u91cd\u5efa\u7248\u7684\u6838\u5fc3\u7ed3\u6784\uff1a\u4e09\u9053\u5165\u573a\u95f8\u95e8\u4e32\u8054\uff0c\u51fa\u573a\u4e09\u9009\u4e00\u5148\u5230\u5148\u8d70\u3002\u6a61\u76ae\u7b4b\u6302\u5728\u4e0a\u5347\u8d8b\u52bf\u7684\u8f68\u9053\u4e0a\uff0c\u62c9\u5f97\u591f\u5f00\u624d\u52a8\u624b\u3002<\/p>\n\n\n\n

\u5165\u573a\u662f\u4e09\u4e2a\u6761\u4ef6\u4e32\u8054\uff0c\u5168\u90e8\u6ee1\u8db3\u624d\u8fdb\u3002<\/p>\n\n\n\n

    \n
  • \u8d8b\u52bf\u8fc7\u6ee4\u3002\u6536\u76d8\u5728 200 \u65e5\u5747\u7ebf\u4e0a\u65b9\u3002\u53ea\u5728\u4e0a\u5347\u8d8b\u52bf\u91cc\u63a5\u56de\u8c03\uff0c\u4e0d\u505a\u4e0b\u8dcc\u8d8b\u52bf\u91cc\u7684\u4fbf\u5b9c\u8d27\u3002<\/strong><\/li>\n\n\n\n
  • \u62c9\u4f38\u3002\u6700\u8fd1 5 \u5929\u7d2f\u8ba1\u8dcc\u5e45 \u2264 -3%\u3002\u6a61\u76ae\u7b4b\u5f97\u62c9\u5f97\u591f\u5f00\uff0c\u5c0f\u6253\u5c0f\u95f9\u4e0d\u78b0\u3002<\/strong><\/li>\n\n\n\n
  • \u786e\u8ba4\u3002IBS < 0.25\uff0c\u4e14\u6536\u76d8\u662f\u6700\u8fd1 5 \u5929\u7684\u6700\u4f4e\u6536\u76d8\u3002<\/strong><\/li>\n<\/ul>\n\n\n\n

    IBS \u662f Internal Bar Strength \u7684\u7f29\u5199\uff0c\u7b97\u6cd5\u662f\uff08\u6536\u76d8\u2212\u6700\u4f4e\uff09\u00f7\uff08\u6700\u9ad8\u2212\u6700\u4f4e\uff09\u3002\u5b83\u628a\u4e00\u5929\u7684\u5f3a\u5f31\u538b\u5230 0 \u5230 1 \u4e4b\u95f4\uff1a0.5 \u662f\u4e2d\u95f4\uff0c\u5c0f\u4e8e 0.25 \u610f\u5473\u7740\u6536\u5728\u5f53\u65e5\u533a\u95f4\u7684\u4e0b\u56db\u5206\u4e4b\u4e00\uff0c\u5168\u5929\u5356\u538b\u4e3b\u5bfc\u3001\u6536\u5f97\u975e\u5e38\u96be\u770b\u3002\u8fd9\u79cd”\u4e11\u6536”\u662f\u7ecf\u5178\u7684\u77ed\u671f\u53cd\u8f6c\u4fe1\u53f7\u3002<\/p>\n\n\n\n

    \u51fa\u573a\u4e09\u9009\u4e00\uff0c\u8c01\u5148\u5230\u5c31\u5148\u8d70\uff1a\u6536\u76d8\u7ad9\u4e0a\u6628\u5929\u6700\u9ad8\u70b9\uff1bIBS > 0.75\uff0c\u6536\u5f97\u6f02\u4eae\u5c31\u843d\u888b\uff1b\u6216\u8005\u62ff\u6ee1 10 \u6839 K \u7ebf\uff0c\u5f3a\u5e73\u3002<\/p>\n\n\n\n

    \u2460 \u53c2\u6570\u4e0e\u6570\u636e<\/p>\n\n\n\n

    import pandas as pd\nimport numpy as np\nimport vectorbt as vbt\nimport yfinance as yf\n\nSYMBOL \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 = \"SPY\"\nSTART_DATE \u00a0 \u00a0 \u00a0 = \"1993-01-01\"\nEND_DATE \u00a0 \u00a0 \u00a0 \u00a0 = \"2030-01-01\"\nTREND_MA \u00a0 \u00a0 \u00a0 \u00a0 = 200 \u00a0 \u00a0 \u00a0# \u8d8b\u52bf\u8fc7\u6ee4\u5747\u7ebf\uff0c\u8bbe 0 \u53ef\u5173\u95ed\nSTRETCH_LOOKBACK = 5 \u00a0 \u00a0 \u00a0 \u00a0# \u6a61\u76ae\u7b4b\u56de\u770b\u7a97\u53e3\uff08\u5929\uff09\nSTRETCH_PCT \u00a0 \u00a0 \u00a0= -0.03 \u00a0 \u00a0# 5 \u65e5\u7d2f\u8ba1\u6536\u76ca \u2264 -3%\nIBS_ENTRY \u00a0 \u00a0 \u00a0 \u00a0= 0.25 \u00a0 \u00a0 # \u5165\u573a\uff1a\u6536\u76d8\u8d34\u7740\u5f53\u65e5\u4f4e\u70b9\nIBS_EXIT \u00a0 \u00a0 \u00a0 \u00a0 = 0.75 \u00a0 \u00a0 # \u79bb\u573a\uff1a\u6536\u76d8\u8d34\u7740\u5f53\u65e5\u9ad8\u70b9\nLOWEST_N \u00a0 \u00a0 \u00a0 \u00a0 = 5 \u00a0 \u00a0 \u00a0 \u00a0# \u6536\u76d8\u987b\u4e3a\u6700\u8fd1 5 \u6839\u7684\u6700\u4f4e\u6536\u76d8\nMAX_HOLD \u00a0 \u00a0 \u00a0 \u00a0 = 10 \u00a0 \u00a0 \u00a0 # \u65f6\u95f4\u6b62\u635f\uff1a\u6700\u591a\u62ff 10 \u6839 K \u7ebf\nINIT_CASH \u00a0 \u00a0 \u00a0 \u00a0= 100_000\nFEES \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 = 0.0001\nSLIPPAGE \u00a0 \u00a0 \u00a0 \u00a0 = 0.0002\n\ndf = yf.download(SYMBOL, start=START_DATE, end=END_DATE,\n\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0interval=\"1d\", multi_level_index=False)\ndf = df.dropna(subset=[\"Open\", \"High\", \"Low\", \"Close\"])<\/code><\/pre>\n\n\n\n
    \u628a\u4fe1\u53f7\u4ea4\u7ed9\u56de\u6d4b\u5f15\u64ce<\/strong><\/h2>\n\n\n\n
    \u2461 \u4fe1\u53f7\u6784\u9020<\/p>\n\n\n\n
    def build_signals(df, trend_ma=TREND_MA, lookback=STRETCH_LOOKBACK,\n                  stretch=STRETCH_PCT, ibs_in=IBS_ENTRY,\n                  ibs_out=IBS_EXIT, lowest_n=LOWEST_N):\n    out = df.copy()\n\n    # \u9664\u96f6\u4fdd\u62a4\uff1aHigh == Low \u7684\u6781\u7aef K \u7ebf\n    rng = (out[\"High\"] - out[\"Low\"]).replace(0, np.nan)\n    out[\"IBS\"] = (out[\"Close\"] - out[\"Low\"]) \/ rng\n\n    # \u95f8\u95e8\u4e00\uff1a\u53ea\u5728\u4e0a\u5347\u8d8b\u52bf\u91cc\u63a5\u56de\u8c03\uff08trend_ma=0 \u53ef\u5173\u95ed\uff09\n    if trend_ma and trend_ma > 0:\n        out[\"Trend\"] = out[\"Close\"] > out[\"Close\"].rolling(trend_ma).mean()\n    else:\n        out[\"Trend\"] = True\n\n    # \u95f8\u95e8\u4e8c\uff1a\u6a61\u76ae\u7b4b\u88ab\u62c9\u5f00\n    out[\"Stretch\"] = out[\"Close\"].pct_change(lookback) <= stretch\n\n    # \u95f8\u95e8\u4e09\uff1a\u4e11\u6536\u786e\u8ba4\uff0c\u6536\u76d8\u8d34\u7740\u4f4e\u70b9\u4e14\u662f 5 \u65e5\u6700\u4f4e\n    out[\"Confirm\"] = (out[\"IBS\"] < ibs_in) & \\\n                     (out[\"Close\"] <= out[\"Close\"].rolling(lowest_n).min())\n\n    out[\"Entry\"] = out[\"Trend\"] & out[\"Stretch\"] & out[\"Confirm\"]\n    out[\"Exit\"] = (out[\"Close\"] > out[\"High\"].shift(1)) | (out[\"IBS\"] > ibs_out)\n\n    return out<\/code><\/pre>\n\n\n\n
    \u786e\u8ba4\u6761\u4ef6\u91cc\u6709\u4e2a\u5bb9\u6613\u88ab\u5ffd\u7565\u7684\u7ec6\u8282\uff1arolling(5).min()<\/code> \u5305\u542b\u5f53\u5929\u3002\u6240\u4ee5”\u6536\u76d8 \u2264 5 \u65e5\u6700\u4f4e\u6536\u76d8”\u7ffb\u8bd1\u8fc7\u6765\u662f”\u4eca\u5929\u7684\u6536\u76d8\u5c31\u662f\u6700\u8fd1\u4e94\u5929\u7684\u6700\u4f4e”\u2014\u2014\u4e0b\u8dcc\u672b\u7aef\u3001\u6050\u614c\u6700\u91cd\u7684\u90a3\u4e00\u5929\u624d\u89e6\u53d1\uff0c\u8fd9\u6b63\u662f\u5747\u503c\u56de\u5f52\u8981\u63a5\u7684\u4f4d\u7f6e\u3002<\/p>\n\n\n\n
    \u2462 \u65f6\u95f4\u6b62\u635f\u4e0e\u4e0b\u5355<\/p>\n\n\n\n
    def apply_time_stop(entries, exits, max_hold=MAX_HOLD):\n    \"\"\"\u5165\u573a\u540e\u6570 K \u7ebf\uff0c\u62ff\u6ee1 max_hold \u6839\u5f3a\u5236\u79bb\u573a\"\"\"\n    n = len(entries)\n    bars_held = np.full(n, 0)\n    out_exits = exits.copy()\n    in_pos = False\n\n    for i in range(n):\n        if not in_pos and entries[i]:\n            in_pos = True\n            bars_held[i] = 0\n        elif in_pos:\n            bars_held[i] = bars_held[i - 1] + 1\n            if bars_held[i] >= max_hold:\n                out_exits[i] = True\n\n        if in_pos and out_exits[i]:\n            in_pos = False\n            bars_held[i] = 0\n\n    return out_exits\n\n\n# ----- \u4e3b\u6d41\u7a0b -----\nsig = build_signals(df)\n\n# \u4fe1\u53f7\u5ef6\u540e\u4e00\u6839 K \u7ebf\uff0c\u7528\u5f00\u76d8\u4ef7\u6210\u4ea4\uff0c\u907f\u514d\u672a\u6765\u51fd\u6570\u3002\n# \u6ce8\u610f\u987a\u5e8f\uff1a\u5148 fillna \u518d astype\uff0cshift \u51fa\u6765\u7684 NaN \u4f1a\u88ab bool() \u5224\u6210 True\nshift_entries = sig[\"Entry\"].shift(1).fillna(False).astype(bool).to_numpy()\nshift_exits = sig[\"Exit\"].shift(1).fillna(False).astype(bool).to_numpy()\nshift_exits = apply_time_stop(shift_entries, shift_exits)\n\npf = vbt.Portfolio.from_signals(\n    close=df[\"Open\"],\n    entries=shift_entries,\n    exits=shift_exits,\n    init_cash=INIT_CASH,\n    fees=FEES,\n    slippage=SLIPPAGE,\n    direction=\"longonly\",\n    freq=\"1D\",\n)<\/code><\/pre>\n\n\n\n
    vectorbt \u7684 from_signals<\/code> \u6ca1\u6709\u73b0\u6210\u7684”N \u6839\u5f3a\u5e73”\u53c2\u6570\uff0c\u65f6\u95f4\u6b62\u635f\u5f97\u81ea\u5df1\u5199\u3002\u9010\u6839\u626b\u63cf\u662f\u7b28\u529e\u6cd5\uff0c\u4f46\u903b\u8f91\u6e05\u695a\uff1a\u5165\u573a\u65e5\u8d77\u6570 K \u7ebf\uff0c\u6570\u6ee1 10 \u6839\u5c31\u5728\u5f53\u6839\u7f6e\u4e00\u4e2a\u51fa\u573a\u6807\u8bb0\uff0c\u5148\u4e8e\u4ef7\u683c\u51fa\u573a\u6761\u4ef6\u4e5f\u6ca1\u5173\u7cfb\uff0c\u53cd\u6b63\u5148\u5230\u5148\u8d70\u3002<\/p>\n\n\n\n
    \u8868\u9762\u7ed3\u679c\uff1a\u4e00\u4e2a\u6807\u51c6\u7684\u5931\u8d25\u54c1<\/strong><\/h2>\n\n\n\n
    \"\"<\/figure>\n\n\n\n
    SPY \u4e09\u5341\u4e09\u5e74\u534a\u7684\u5b8c\u6574\u56de\u6d4b\u3002Underwater \u5b50\u56fe\u91cc\u90a3\u6761\u6700\u6df1\u7684\u5751\uff0c\u5c31\u662f\u540e\u9762\u8981\u8bb2\u7684 3,046 \u5929\u3002<\/p>\n\n\n\n
    \u6307\u6807<\/th>\u6a61\u76ae\u7b4b\u7b56\u7565<\/th>\u4e70\u5165\u6301\u6709 SPY<\/th><\/tr><\/thead>
    \u5e74\u5316\u6536\u76ca CAGR<\/td>1.66%<\/td>10.84%<\/td><\/tr>
    \u603b\u56de\u62a5<\/td>73.5%<\/td>3,062%<\/td><\/tr>
    \u6700\u5927\u56de\u64a4<\/td>10.21%<\/td>\u7ea6 55%<\/td><\/tr>
    \u590f\u666e\u6bd4\u7387<\/td>0.596<\/strong><\/td>\u7ea6 0.50<\/td><\/tr>
    \u80dc\u7387<\/td>83.5%<\/td>\u2014<\/td><\/tr>
    \u76c8\u4e8f\u6bd4<\/td>2.51<\/td>\u2014<\/td><\/tr>
    \u4ea4\u6613\u6b21\u6570<\/td>79<\/td>1<\/td><\/tr>
    \u6700\u957f\u56de\u64a4\u6301\u7eed<\/td>3,046 \u5929<\/td>\u2014<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
    \u533a\u95f4 1993-01 \u81f3 2026-08\uff0c\u5171 33.6 \u5e74\u3002\u4e70\u5165\u6301\u6709\u7684\u80dc\u7387\u548c\u76c8\u4e8f\u6bd4\u6ca1\u6709\u610f\u4e49\uff0c\u5b83\u53ea\u6709\u4e00\u7b14\u4ea4\u6613\u3002<\/p>\n\n\n\n
    \u6bcf\u5e74\u843d\u540e 9 \u4e2a\u767e\u5206\u70b9\uff0c\u590d\u5229\u4e09\u5341\u4e09\u5e74\uff0c\u5c31\u662f 73.5% \u5bf9 3,062% \u7684\u5dee\u8ddd\u3002<\/p>\n\n\n\n
    \u66f4\u96be\u53d7\u7684\u662f\uff1a\u6700\u957f\u56de\u64a4\u6301\u7eed 3,046 \u5929\uff0c\u516b\u5e74\u56db\u4e2a\u6708\u6ca1\u521b\u65b0\u9ad8\u3002\u5bf9 standalone \u7b56\u7565\u6765\u8bf4\u8fd9\u662f\u6b7b\u5211\u5224\u51b3\uff0c\u6ca1\u4eba\u625b\u5f97\u4f4f\u516b\u5e74\u4e0d\u521b\u65b0\u9ad8\u8fd8\u9762\u4e0d\u6539\u8272\u3002<\/p>\n\n\n\n
    \u5199\u5230\u8fd9\u91cc\uff0c\u5927\u591a\u6570\u5206\u6790\u7684\u5267\u60c5\u662f\u5220\u6389\u7b56\u7565\u3001\u6539\u540d failed_ideas<\/code>\u3001\u53e6\u8d77\u7089\u7076\u3002\u6211\u5dee\u4e00\u70b9\u4e5f\u8fd9\u4e48\u5e72\u4e86\u3002<\/p>\n\n\n\n
    \u6539\u53d8\u4e00\u5207\u7684\u6570\u5b57\uff1a2.0%<\/strong><\/h2>\n\n\n\n
    \u5e02\u573a\u655e\u53e3\uff1a\u7ea6 8,447 \u4e2a\u4ea4\u6613\u65e5\u91cc\uff0c\u53ea\u6301\u4ed3\u7ea6 169 \u5929<\/p>\n\n\n\n
    \"\"<\/figure>\n\n\n\n
    \u8fd9\u5f20\u6c47\u603b\u56fe\u628a79\u4e2a\u5165\u573a\u70b9\u6492\u5728\u4e09\u5341\u4e09\u5e74\u7684\u65f6\u95f4\u8f74\u4e0a\uff0c\u5bc6\u5bc6\u9ebb\u9ebb\u5374\u53ea\u53602%\u3002\u4e94\u4e2a\u5173\u952e\u6570\u90fd\u5217\u5728\u4e0b\u9762\u3002<\/p>\n\n\n\n
    1.66% \u7684\u5e74\u5316\uff0c\u4e0d\u662f”\u5929\u5929\u5728\u573a\u4f46\u8868\u73b0\u5dee”\uff0c\u800c\u662f”98% \u7684\u65f6\u95f4\u62ff\u7740\u73b0\u91d1\uff0c\u51fa\u624b\u4e00\u6b21\u8d5a\u4e24\u5929”\u300279 \u7b14\u4ea4\u6613\u5e73\u5747\u6301\u4ed3 2.1 \u5929\uff0c\u4e00\u5e74 2.3 \u7b14\u3002<\/p>\n\n\n\n
    \u628a\u5206\u6bcd\u6362\u6210\u5b9e\u9645\u90e8\u7f72\u7684\u8d44\u672c\uff0c\u8d26\u5b8c\u5168\u662f\u53e6\u4e00\u672c\uff1a73.5% \u7684\u603b\u56de\u62a5\u662f\u5728\u90a3 169 \u5929\u91cc\u8d5a\u5230\u7684\uff0c\u6309\u90e8\u7f72\u671f\u95f4\u5e74\u5316\u7ea6 128%\u3002<\/p>\n\n\n\n
    \u800c128% \u662f\u6bd4\u7387\uff0c\u4e0d\u662f\u9884\u6d4b\u3002\u4f60\u6ca1\u6cd5\u6309\u9700\u5236\u9020\u66f4\u591a\u4fe1\u53f7\u53bb\u6536\u5272\uff0cSPY \u4e00\u5e74\u5c31\u7ed9\u4f60\u4e24\u4e09\u6b21\u673a\u4f1a\u3002\u4f46\u5b83\u544a\u8bc9\u4f60\u4fe1\u53f7\u7684\u5355\u4f4d\u98ce\u9669\u4ef7\u503c\uff0c\u8fd9\u662f\u6bd4\u8f83\u4e00\u4e2a 2% \u655e\u53e3\u7cfb\u7edf\u548c\u4e00\u4e2a 100% \u655e\u53e3\u7cfb\u7edf\u65f6\uff0c\u552f\u4e00\u8bda\u5b9e\u7684\u5c3a\u5b50\u3002<\/p>\n\n\n\n
    79 \u7b14\uff0c\u5747\u503c 0.717%\uff0c\u5355\u7b14\u6807\u51c6\u5dee\u7ea6 1.40%\uff0ct \u7edf\u8ba1\u91cf\u7ea6 4.6\u3002\u5c31\u7b97\u6309\u626b\u8fc7 36 \u7ec4\u53c2\u6570\u505a Bonferroni \u5f0f\u6821\u6b63\uff0c\u4e5f\u7a33\u7a33\u8d85\u8fc7 3.2 \u7684\u9608\u503c\u3002\u6837\u672c\u4e0d\u5927\uff0c\u4f46\u6548\u5e94\u91cf\u5927\u3001\u65b9\u5411\u4e00\u81f4\uff0c\u4e0d\u592a\u50cf\u566a\u97f3\u3002<\/p>\n\n\n\n
    \"\"<\/figure>\n\n\n\n
    \u4e09\u5341\u4e09\u5e74\u7684\u65e5\u5386\u6a2a\u5e26\uff0c\u4eae\u8d77\u7684\u683c\u5b50\u4e0d\u5230\u5341\u5206\u4e4b\u4e00\u3002\u73b0\u91d1\u4ed3\u5728\u7edd\u5927\u90e8\u5206\u65f6\u95f4\u662f\u95f2\u7f6e\u7684\u3002<\/p>\n\n\n\n
    \u2463 \u655e\u53e3\u4e0e\u9010\u7b14\u7edf\u8ba1<\/p>\n\n\n\n
    in_position = pf.position_mask().to_numpy()\n\nexposure = in_position.mean() * 100\nprint(f\"Market exposure: {exposure:.1f}%  \"\n      f\"({int(in_position.sum())} of {len(df)} bars)\")\n\ntrades = pf.trades.records_readable\nprint(f\"Trades: {len(trades)}  \"\n      f\"Win rate: {(trades['PnL'] > 0).mean() * 100:.1f}%  \"\n      f\"Avg per trade: {trades['Return'].mean() * 100:.3f}%\")\n\n# \u6309\u90e8\u7f72\u8d44\u672c\u5e74\u5316\uff1a\u603b\u56de\u62a5\u53d1\u751f\u5728\u6301\u4ed3\u65e5\u91cc\uff0c\u5206\u6bcd\u8981\u7528\u6301\u4ed3\u5e74\u6570\ndays_held = ((pd.to_datetime(trades[\"Exit Timestamp\"]) -\n              pd.to_datetime(trades[\"Entry Timestamp\"])).dt.days + 1)\nheld_years = days_held.sum() \/ 252\ntotal_ret = (1 + trades[\"Return\"]).prod() - 1\nprint(f\"Return on deployed capital (annualized): \"\n      f\"{(1 + total_ret) ** (1 \/ held_years) - 1:.0%}\")\n\n# \u9010\u7b14\u4f18\u52bf\u7684\u663e\u8457\u6027\uff1a\u5747\u503c\u662f\u4e0d\u662f\u566a\u97f3\nt_stat = trades[\"Return\"].mean() \/ (trades[\"Return\"].std() \/ np.sqrt(len(trades)))\nprint(f\"Trade t-stat: {t_stat:.2f}\")<\/code><\/pre>\n\n\n\n
    \u6211\u91cd\u5efa\u7684\u4e0d\u662f\u90a3\u4e2a\u7b56\u7565<\/strong><\/h2>\n\n\n\n
    \u6309\u90e8\u7f72\u8d44\u672c\u5e74\u5316 128% \u542c\u7740\u5f88\u723d\uff0c\u4f46\u6709\u4e00\u7b14\u8d26\u6211\u5fc5\u987b\u81ea\u5df1\u5148\u6233\u7834\u3002<\/p>\n\n\n\n
    \u6211\u7684\u7248\u672c\u655e\u53e3 2.0%\uff0c\u89c6\u9891\u5ba3\u79f0 14%\u3002\u5dee\u4e86 7 \u500d\uff0c\u800c\u4e14\u8fd9\u4e0d\u662f\u8c03\u53c2\u80fd\u586b\u5e73\u7684\u2014\u201436 \u7ec4\u53c2\u6570\u626b\u4e0b\u6765\uff0c\u6700\u6d3b\u8dc3\u7684\u7ec4\u5408\u4e5f\u53ea\u6709 5.4% \u655e\u53e3\uff0c\u6ca1\u6709\u4e00\u4e2a\u63a5\u8fd1 14%\u3002<\/p>\n\n\n\n
    \u6240\u4ee5\u76f4\u8bf4\uff1a\u8fd9\u4e0d\u662f\u89c6\u9891\u91cc\u90a3\u4e2a\u7b56\u7565\u3002\u5b83\u53ea\u662f\u78b0\u5de7\u5728\u53e6\u4e00\u79cd\u4ea4\u6613\u8282\u594f\u4e0a\uff0c\u957f\u51fa\u4e86\u76f8\u4f3c\u7684\u5355\u7b14\u5e73\u5747\u6536\u76ca\uff08\u6211\u7684 0.717% \u5bf9\u4ed6\u7684 0.66%\uff09\u3002\u539f\u59cb\u7b56\u7565\u53ef\u80fd\u66f4\u677e\uff1a\u66f4\u6d45\u7684\u56de\u8c03\u9608\u503c\u3001\u6ca1\u6709\u786e\u8ba4\u6761\u4ef6\u3001\u4e5f\u8bb8\u538b\u6839\u6ca1\u6709\u8d8b\u52bf\u8fc7\u6ee4\uff0c\u4e5f\u8bb8\u4e0d\u6b62\u8dd1 SPY\u3002<\/p>\n\n\n\n
    \n
    \u5339\u914d\u4e00\u4e2a\u7edf\u8ba1\u91cf\u3001\u5374\u5728\u53e6\u4e00\u4e2a\u4e0a\u5dee 7 \u500d\uff0c\u8fd9\u4e0d\u53eb\u590d\u73b0\u6210\u529f\u3002\u4ece\u6c47\u603b\u7edf\u8ba1\u53cd\u63a8\uff0c\u4f60\u5f97\u5230\u7684\u662f”\u4e00\u4e2a”\u7b56\u7565\uff0c\u4e0d\u662f”\u90a3\u4e2a”\u7b56\u7565\u3002<\/p>\n<\/blockquote>\n\n\n\n
    \u8fd9\u4e00\u6bb5\u662f\u6211\u5199\u7ed9\u81ea\u5df1\u770b\u7684\u3002\u53cd\u63a8\u662f\u597d\u7684\u7ec3\u4e60\uff0c\u4f46\u53cd\u63a8\u7684\u4ea7\u7269\u8981\u6309”\u65b0\u7b56\u7565”\u8d70\u5168\u5957\u9a8c\u8bc1\u6d41\u7a0b\uff0c\u4e0d\u80fd\u56e0\u4e3a”\u5bf9\u4e0a\u4e86\u89c6\u9891\u7684\u6570\u5b57”\u5c31\u8df3\u8fc7\u3002<\/p>\n\n\n\n
    \u53c2\u6570\u626b\u63cf\uff1a\u7f51\u683c\u4fee\u6b63\u4e86\u6211\u7684\u5ba1\u7f8e<\/strong><\/h2>\n\n\n\n
    \u9009\u53c2\u6570\u6700\u6015\u62cd\u8111\u888b\u3002\u6211\u628a\u62c9\u4f38\u9608\u503c\u3001IBS \u5165\u573a\u7ebf\u3001\u65f6\u95f4\u6b62\u635f\u5404\u53d6\u4e09\u6863\uff0c\u626b\u4e86 36 \u7ec4\u7ec4\u5408\uff0c\u8ba9\u6570\u636e\u66ff\u6211\u505a\u8fd9\u4e2a\u51b3\u5b9a\u3002<\/p>\n\n\n\n
    \"\"<\/figure>\n\n\n\n
    36 \u683c\u626b\u51fa\u6765\u7684\u5730\u5f62\u662f\u9ad8\u539f\u4e0d\u662f\u5c16\u5cf0\uff1a-2% \u90a3\u683c\u5728\u9ad8\u539f\u9876\u4e0a\uff0c\u5f80\u56db\u5468\u8d70\u6570\u5b57\u5e73\u7f13\u4e0b\u6ed1\uff0c\u4e0d\u584c\u65b9\u3002<\/p>\n\n\n\n
    \u56de\u8c03\u9608\u503c<\/th>\u4ea4\u6613\u6b21\u6570<\/th>\u5355\u7b14\u5747\u503c<\/th>\u590f\u666e<\/th>\u655e\u53e3<\/th><\/tr><\/thead>
    -2%<\/td>187<\/strong><\/td>0.528%<\/td>0.769<\/strong><\/td>5.4%<\/td><\/tr>
    -3%\uff08\u57fa\u51c6\uff09<\/td>79<\/td>0.717%<\/td>0.596<\/td>2.0%<\/td><\/tr>
    -4%<\/td>44<\/td>1.032%<\/td>0.578<\/td>0.9%<\/td><\/tr>
    -5%<\/td>13<\/td>1.831%<\/td>0.445<\/td>0.2%<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
    \u5355\u8c03\u7684\u6743\u8861\u7ebf\uff1a\u9608\u503c\u8d8a\u6df1\u5355\u7b14\u8d8a\u80a5\uff0c\u8bc1\u636e\u8d8a\u8584\u3002-5% \u90a3\u884c 13 \u7b14\uff0c\u662f\u6545\u4e8b\u4e0d\u662f\u6837\u672c\u3002<\/p>\n\n\n\n
    \u6574\u4e2a\u7f51\u683c\u7684\u4e2d\u4f4d\u590f\u666e\u662f 0.548\uff0c64% \u7684\u683c\u5b50\u590f\u666e > 0.5\u3002\u8fd9\u662f\u9ad8\u539f\uff0c\u4e0d\u662f\u5c16\u5cf0\uff0c\u5f62\u6001\u5065\u5eb7\uff1a\u53c2\u6570\u632a\u4e00\u632a\u6570\u5b57\u4e0d\u4f1a\u5d29\u3002<\/p>\n\n\n\n
    \u597d\u6d88\u606f\u5728 -2% \u90a3\u4e00\u884c\uff1a\u590f\u666e 0.769\uff0c187 \u7b14\uff0c\u655e\u53e3 5.4%\uff0c\u6700\u5927\u56de\u64a4\u548c\u57fa\u51c6\u57fa\u672c\u6301\u5e73\u3002\u540c\u6837\u7684\u56de\u64a4\uff0c\u66f4\u9ad8\u7684\u98ce\u9669\u8c03\u6574\u6536\u76ca\uff0c2.4 \u500d\u7684\u8bc1\u636e\u91cf\u3002\u5f80\u4e0b\u63a8\u8fdb\u7684\u7248\u672c\u5e94\u8be5\u662f -2%\uff0c\u4e0d\u662f\u6211\u7684 -3%\u3002<\/p>\n\n\n\n
    \u2464 36 \u683c\u53c2\u6570\u626b\u63cf<\/p>\n\n\n\n
    grid = [(s, i, t)\n        for s in [-0.02, -0.03, -0.04, -0.05]\n        for i in [0.15, 0.25, 0.35]\n        for t in [5, 10, 15]]\n\nrows = []\nfor s, i, t in grid:\n    g_sig = build_signals(df, stretch=s, ibs_in=i)\n    e = g_sig[\"Entry\"].shift(1).fillna(False).astype(bool).to_numpy()\n    x = g_sig[\"Exit\"].shift(1).fillna(False).astype(bool).to_numpy()\n    x = apply_time_stop(e, x, max_hold=t)\n    p = vbt.Portfolio.from_signals(close=df[\"Open\"], entries=e, exits=x,\n                                   init_cash=INIT_CASH, fees=FEES,\n                                   slippage=SLIPPAGE, freq=\"1D\")\n    tr = p.trades.records_readable\n    rows.append({\n        \"stretch\": s, \"ibs_in\": i, \"time_stop\": t,\n        \"trades\": len(tr),\n        \"avg_pct\": tr[\"Return\"].mean() * 100 if len(tr) else np.nan,\n        \"sharpe\": p.sharpe_ratio(),\n        \"exposure\": p.position_mask().to_numpy().mean() * 100,\n    })\n\nres = pd.DataFrame(rows).sort_values(\"sharpe\", ascending=False)\nprint(res.to_string(index=False))\n\nprint(f\"\\n\u7f51\u683c\u4e2d\u4f4d\u590f\u666e: {res['sharpe'].median():.3f}  \"\n      f\"\u590f\u666e>0.5 \u5360\u6bd4: {(res['sharpe'] > 0.5).mean() * 100:.0f}%\")<\/code><\/pre>\n\n\n\n
    \u8bfb\u626b\u63cf\u7ed3\u679c\u65f6\u522b\u53ea\u627e\u6700\u9ad8\u90a3\u683c\u3002\u8981\u770b\u7684\u662f\uff1a\u8c03\u53c2\u4e4b\u540e\u8868\u73b0\u662f\u5e73\u6ed1\u8870\u51cf\uff0c\u8fd8\u662f\u53ea\u5728\u7cbe\u786e\u5750\u6807\u4e0a\u7075\u5149\u3002\u524d\u8005\u662f\u6548\u5e94\uff0c\u540e\u8005\u662f\u8fc7\u62df\u5408\u7684\u7b7e\u540d\u3002<\/p>\n\n\n\n
    \u7ec4\u5408\u8bba\u7684\u5751\uff1a\u4e24\u4e2a\u7cfb\u7edf\u90fd\u60f3\u5728\u5468\u4e8c\u505a\u591a<\/strong><\/h2>\n\n\n\n
    \"\"<\/figure>\n\n\n\n
    \u6700\u987a\u624b\u7684\u914d\u5bf9\u65b9\u6848\uff0c\u6070\u6070\u649e\u8f66\uff1a\u4e24\u8fb9\u90fd\u76ef\u7740”SPY 200 \u65e5\u7ebf\u4e0a\u65b9”\u8fd9\u4e2a\u524d\u63d0\u8981\u94b1\u3002<\/p>\n\n\n\n
    \u4f4e\u655e\u53e3\u7b56\u7565\u6700\u5e38\u89c1\u7684\u5356\u6cd5\u662f”\u514d\u8d39\u53e0\u52a0”\uff1a\u5b83 98% \u7684\u65f6\u95f4\u62ff\u7740\u73b0\u91d1\uff0c\u53e0\u5230\u4efb\u4f55\u6838\u5fc3\u7b56\u7565\u4e0a\uff0c\u7b49\u4e8e\u8ba9\u95f2\u7f6e\u8d44\u91d1\u591a\u6536\u4e00\u6761\u6d41\u3002\u542c\u8d77\u6765\u65e0\u672c\u4e07\u5229\u3002<\/p>\n\n\n\n
    \u4f46\u8fd9\u4e2a\u8bba\u70b9\u6709\u4e2a\u6d1e\u3002\u8d8b\u52bf\u8fc7\u6ee4\u8981\u6c42\u6536\u76d8\u7ad9\u5728 200 \u65e5\u7ebf\u4e0a\u65b9\uff0c\u800c\u5b83\u9700\u8981\u8d44\u91d1\u7684\u65e5\u5b50\uff0c\u6070\u597d\u662f\u4efb\u4f55\u591a\u5934\u8d8b\u52bf\u6216\u52a8\u91cf\u7cfb\u7edf\u5df2\u7ecf\u6ee1\u4ed3\u7684\u65e5\u5b50\u3002\u6ca1\u6709\u95f2\u7f6e\u8d44\u91d1\u53ef\u817e\u3002\u4e24\u4e2a\u7cfb\u7edf\u4e0d\u662f\u4e92\u8865\uff0c\u662f\u62a2\u94b1\uff1a\u8981\u4e48\u4e0a\u6760\u6746\uff0c\u8981\u4e48\u4e3a\u4e86\u4e00\u7b14\u4e24\u5929\u7684\u4ea4\u6613\u6324\u6389\u6838\u5fc3\u4ed3\u4f4d\u3002<\/p>\n\n\n\n
    \u8fd9\u6740\u6b7b\u4e86\u6700\u987a\u624b\u7684\u914d\u5bf9\uff0c\u4f46\u6ca1\u6740\u6b7b\u60f3\u6cd5\uff0c\u53ea\u662f\u7ed9\u5408\u4f5c\u65b9\u5b9a\u4e86\u6761\u4ef6\uff1a\u8d44\u672c\u9700\u6c42\u8981\u7ed3\u6784\u6027\u9519\u5f00\uff0c\u800c\u4e0d\u53ea\u662f\u6536\u76ca\u76f8\u5173\u6027\u4f4e\u3002<\/p>\n\n\n\n
      \n
    • \u4e0d\u540c\u8d44\u4ea7\u7c7b\u522b\u3002\u503a\u5238\u3001\u9ec4\u91d1\u3001\u5916\u6c47\u8d8b\u52bf\u7cfb\u7edf\u6709\u81ea\u5df1\u7684\u65f6\u949f\uff0cTLT \u4e0d\u4f1a\u5728 SPY \u56de\u8c03 3% \u7684\u540c\u4e00\u5929\u8981\u94b1\u3002<\/strong><\/li>\n\n\n\n
    • \u5e02\u573a\u4e2d\u6027\u6216\u7a7a\u5934\u65b9\u5411\u3002\u5bf9\u51b2\u8d26\u7c3f\u7684\u8d44\u91d1\u4e0d\u7ed1\u5b9a SPY \u7684 200 \u65e5\u7ebf\u72b6\u6001\u3002<\/strong><\/li>\n\n\n\n
    • \u53cd\u5411 regime \u7cfb\u7edf\u3002\u5728 200 \u65e5\u7ebf\u4e0b\u65b9\u90e8\u7f72\u7684\u7cfb\u7edf\u662f\u5929\u7136\u5bf9\u624b\u65b9\uff0c\u4e24\u5957\u7cfb\u7edf\u62fc\u63a5\u65e5\u5386\u800c\u4e0d\u662f\u91cd\u53e0\u3002<\/strong><\/li>\n\n\n\n
    • \u6709\u4efb\u52a1\u7684\u73b0\u91d1\u4ed3\u3002\u4f60\u672c\u6765\u5c31\u7559\u7740\u6284\u5e95\u7684\u73b0\u91d1\uff0c\u6b63\u597d\u8ba9\u5b83\u5728 2% \u7684\u65f6\u95f4\u91cc\u89c4\u5219\u5316\u5e72\u6d3b\u3002<\/strong><\/li>\n<\/ul>\n\n\n\n
      \n
      \u53e0\u52a0\u7b56\u7565\u7684\u6b63\u786e\u987a\u5e8f\uff1a\u5148\u68c0\u67e5\u655e\u53e3\u7a97\u53e3\u662f\u5426\u91cd\u53e0\uff0c\u518d\u770b\u6536\u76ca\u76f8\u5173\u6027\u3002\u4e24\u4e2a\u7cfb\u7edf\u6536\u76ca\u76f8\u5173\u6027 -0.1 \u4e5f\u53ef\u4ee5\u6beb\u65e0\u7528\u5904\uff0c\u56e0\u4e3a\u5b83\u4eec\u90fd\u60f3\u5728\u540c\u4e00\u4e2a\u5468\u4e8c\u505a\u591a\u3002<\/p>\n<\/blockquote>\n\n\n\n
      \u89c2\u70b9\u603b\u7ed3<\/strong><\/h2>\n\n\n\n
      \u6bcf\u5e74 2.3 \u7b14\u3001\u6bcf\u7b14 0.717%\u3001\u65e0\u6760\u6746\uff0c\u5bf9\u7ec4\u5408\u7684\u5e74\u8d21\u732e\u7ea6 1.7%\uff0c\u8fd9\u6839\u672c\u7b97\u4e0d\u4e0a\u4e00\u95e8\u751f\u610f\uff1a\u4e00\u5e74\u53ea\u6709\u4e24\u7b14\u4ea4\u6613\uff0c\u9047\u5230\u5176\u4e2d\u4e00\u7b14\u8fd8\u4e8f\u7740\u7684\u5e74\u4efd\uff0c\u4f60\u51e0\u4e4e\u5bdf\u89c9\u4e0d\u5230\u5b83\u7684\u5b58\u5728\u3002\u4f46\u6709\u4e09\u4ef6\u4e8b\u662f\u5b9e\u7684\uff0c\u503c\u5f97\u62ce\u51fa\u6765\u8bf4\u3002<\/p>\n\n\n\n
      \u7b2c\u4e00\uff0c<\/strong>\u5b83\u662f\u8d44\u672c\u6548\u7387\u7ec4\u4ef6\uff0c\u4e0d\u662f\u72ec\u7acb\u7b56\u7565\uff0c\u653e\u8fdb\u7b56\u7565\u6808\u3001\u7ed9\u5c0f\u6bd4\u4f8b\u914d\u7f6e\uff0c\u7528\u8fb9\u9645\u590f\u666e\u8d21\u732e\u53bb\u8bc4\u4f30\uff0c\u522b\u62ff\u72ec\u7acb CAGR \u8bf4\u4e8b\u3002<\/p>\n\n\n\n
      \u7b2c\u4e8c\uff0c<\/strong>IBS \u52a0\u62c9\u4f38\u52a0 regime \u7684\u7ec4\u5408\u901a\u8fc7\u4e86\u663e\u8457\u6027\u68c0\u9a8c\uff0c\u5728 36 \u683c\u7f51\u683c\u91cc\u4e5f\u6d3b\u5f97\u597d\u597d\u7684\uff0c\u4f18\u52bf\u662f\u771f\u7684\uff0c\u53ea\u662f\u5c0f\u3002<\/p>\n\n\n\n
      \u7b2c\u4e09\uff0c<\/strong>\u8c03\u53c2\u95ee\u9898\u5df2\u7ecf\u89e3\u51b3\u3002-2% \u7248\u672c\uff0c187 \u7b14\uff0c\u590f\u666e 0.769\uff0c\u56de\u64a4\u4e0d\u6076\u5316\uff0c\u8fd9\u624d\u662f\u503c\u5f97\u5f80\u4e0b\u5e26\u8d70\u7684\u7248\u672c\u3002<\/p>\n\n\n\n
      \u56de\u770b\u6574\u4e2a\u8fc7\u7a0b\uff0c\u5f53\u628a 1.66% \u548c 10.84% \u6446\u5728\u4e00\u8d77\u7684\u65f6\u5019\uff0c\u6211\u4eec\u53ef\u80fd\u4f1a\u5220\u6389\u6587\u4ef6\u3002\u4e00\u4e2a\u5728\u5e02 2% \u7684\u7b56\u7565\uff0c\u4e0d\u662f\u5728\u548c\u4e70\u5165\u6301\u6709\u7ade\u4e89\u3002\u8bc4\u4ef7\u5b83\u7684\u5c3a\u5b50\u4e0d\u662f CAGR\uff0c\u662f\u4fe1\u53f7\u7684\u5355\u4f4d\u98ce\u9669\u4ef7\u503c\u3001\u655e\u53e3\u7a97\u53e3\u7684\u5f62\u72b6\uff0c\u4ee5\u53ca\u5b83\u8ddf\u7ec4\u5408\u91cc\u5176\u4ed6\u7cfb\u7edf\u62a2\u4e0d\u62a2\u94b1\u3002<\/p>\n\n\n\n
      \u4e0b\u6b21\u770b\u5230\u4e00\u4e2a\u96be\u770b\u7684\u5e74\u5316\uff0c\u5148\u95ee\u4e00\u53e5\uff1a\u5b83\u7684\u94b1\u5728\u573a\u5916\u5f85\u4e86\u591a\u4e45\uff1f<\/strong><\/p>\n\n\n\n
      \u5b8c\u6574\u4ee3\u7801<\/strong><\/h2>\n\n\n\n
      \u4e0a\u9762\u4e94\u6bb5\u662f\u62c6\u5f00\u8bb2\u7684\uff0c\u8fd9\u4efd\u662f\u62fc\u597d\u7684\u5b8c\u6574\u811a\u672c\uff0c\u53ef\u76f4\u63a5\u590d\u5236\u8fd0\u884c\u3002\u88c5\u4f9d\u8d56\uff1a<\/p>\n\n\n\n
      pip install vectorbt yfinance pandas numpy<\/code><\/pre>\n\n\n\n
      import pandas as pd\nimport numpy as np\nimport vectorbt as vbt\nimport yfinance as yf\n\n# -------------------------\n# \u53c2\u6570\n# -------------------------\nSYMBOL = \"SPY\"\nSTART_DATE = \"1993-01-01\"\nEND_DATE = \"2030-01-01\"\nTREND_MA = 200          # \u8d8b\u52bf\u8fc7\u6ee4\u5747\u7ebf\uff0c\u8bbe 0 \u53ef\u5173\u95ed\nSTRETCH_LOOKBACK = 5    # \u6a61\u76ae\u7b4b\u56de\u770b\u7a97\u53e3\uff08\u5929\uff09\nSTRETCH_PCT = -0.03     # 5 \u65e5\u7d2f\u8ba1\u6536\u76ca\u9608\u503c\nIBS_ENTRY = 0.25        # \u5165\u573a\uff1a\u6536\u76d8\u8d34\u7740\u5f53\u65e5\u4f4e\u70b9\nIBS_EXIT = 0.75         # \u79bb\u573a\uff1a\u6536\u76d8\u8d34\u7740\u5f53\u65e5\u9ad8\u70b9\nLOWEST_N = 5            # \u6536\u76d8\u987b\u4e3a\u6700\u8fd1 N \u6839\u7684\u6700\u4f4e\u6536\u76d8\nMAX_HOLD = 10           # \u65f6\u95f4\u6b62\u635f\uff08\u6839\uff09\nINIT_CASH = 100_000\nFEES = 0.0001\nSLIPPAGE = 0.0002\n\n# -------------------------\n# \u6570\u636e\n# -------------------------\ndf = yf.download(SYMBOL, start=START_DATE, end=END_DATE,\n                 interval=\"1d\", multi_level_index=False)\ndf = df.dropna(subset=[\"Open\", \"High\", \"Low\", \"Close\"])\n\n# -------------------------\n# \u4fe1\u53f7\u6784\u9020\n# -------------------------\ndef build_signals(df, trend_ma=TREND_MA, lookback=STRETCH_LOOKBACK,\n                  stretch=STRETCH_PCT, ibs_in=IBS_ENTRY,\n                  ibs_out=IBS_EXIT, lowest_n=LOWEST_N):\n    out = df.copy()\n    rng = (out[\"High\"] - out[\"Low\"]).replace(0, np.nan)\n    out[\"IBS\"] = (out[\"Close\"] - out[\"Low\"]) \/ rng\n\n    if trend_ma and trend_ma > 0:\n        out[\"Trend\"] = out[\"Close\"] > out[\"Close\"].rolling(trend_ma).mean()\n    else:\n        out[\"Trend\"] = True\n\n    out[\"Stretch\"] = out[\"Close\"].pct_change(lookback) <= stretch\n    out[\"Confirm\"] = (out[\"IBS\"] < ibs_in) & \\\n                     (out[\"Close\"] <= out[\"Close\"].rolling(lowest_n).min())\n\n    out[\"Entry\"] = out[\"Trend\"] & out[\"Stretch\"] & out[\"Confirm\"]\n    out[\"Exit\"] = (out[\"Close\"] > out[\"High\"].shift(1)) | (out[\"IBS\"] > ibs_out)\n    return out\n\n# -------------------------\n# \u65f6\u95f4\u6b62\u635f\n# -------------------------\ndef apply_time_stop(entries, exits, max_hold=MAX_HOLD):\n    n = len(entries)\n    bars_held = np.full(n, 0)\n    out_exits = exits.copy()\n    in_pos = False\n\n    for i in range(n):\n        if not in_pos and entries[i]:\n            in_pos = True\n            bars_held[i] = 0\n        elif in_pos:\n            bars_held[i] = bars_held[i - 1] + 1\n            if bars_held[i] >= max_hold:\n                out_exits[i] = True\n\n        if in_pos and out_exits[i]:\n            in_pos = False\n            bars_held[i] = 0\n\n    return out_exits\n\n# -------------------------\n# \u56de\u6d4b\n# -------------------------\nsig = build_signals(df)\n\nshift_entries = sig[\"Entry\"].shift(1).fillna(False).astype(bool).to_numpy()\nshift_exits = sig[\"Exit\"].shift(1).fillna(False).astype(bool).to_numpy()\nshift_exits = apply_time_stop(shift_entries, shift_exits)\n\npf = vbt.Portfolio.from_signals(\n    close=df[\"Open\"],\n    entries=shift_entries,\n    exits=shift_exits,\n    init_cash=INIT_CASH,\n    fees=FEES,\n    slippage=SLIPPAGE,\n    direction=\"longonly\",\n    freq=\"1D\",\n)\n\n# -------------------------\n# \u655e\u53e3\u4e0e\u7ee9\u6548\n# -------------------------\nin_position = pf.position_mask().to_numpy()\n\nprint(f\"\\nRubber Band Mean Reversion on {SYMBOL}\")\nprint(f\"Market exposure: {in_position.mean() * 100:.1f}%  \"\n      f\"({int(in_position.sum())} of {len(df)} bars)\")\n\nbh = vbt.Portfolio.from_holding(close=df[\"Open\"], init_cash=INIT_CASH, freq=\"1D\")\nprint(f\"Strategy CAGR : {pf.annualized_return() * 100:.2f}%\")\nprint(f\"Buy&Hold CAGR : {bh.annualized_return() * 100:.2f}%\")\nprint(f\"Strategy MaxDD: {pf.max_drawdown() * 100:.2f}%\")\nprint(f\"Buy&Hold MaxDD: {bh.max_drawdown() * 100:.2f}%\")\n\ntrades = pf.trades.records_readable\nprint(f\"Trades: {len(trades)}  \"\n      f\"Win rate: {(trades['PnL'] > 0).mean() * 100:.1f}%  \"\n      f\"Avg per trade: {trades['Return'].mean() * 100:.3f}%\")\n\ndays_held = ((pd.to_datetime(trades[\"Exit Timestamp\"]) -\n              pd.to_datetime(trades[\"Entry Timestamp\"])).dt.days + 1)\nheld_years = days_held.sum() \/ 252\ntotal_ret = (1 + trades[\"Return\"]).prod() - 1\nprint(f\"Return on deployed capital (annualized): \"\n      f\"{(1 + total_ret) ** (1 \/ held_years) - 1:.0%}\")\n\nt_stat = trades[\"Return\"].mean() \/ (trades[\"Return\"].std() \/ np.sqrt(len(trades)))\nprint(f\"Trade t-stat: {t_stat:.2f}\")\n\n# -------------------------\n# \u7ed8\u56fe\n# -------------------------\nprint(pf.stats())\nfig = pf.plot(\n    subplots=[\"value\", \"underwater\", \"orders\", \"trade_pnl\"],\n    title=f\"{SYMBOL} \u2014 Rubber Band\",\n    make_subplots_kwargs=dict(\n        row_heights=[0.30, 0.20, 0.20, 0.30],\n        vertical_spacing=0.03,\n    )\n)\nfig.update_layout(height=1200)\nfig.show()\n\n# -------------------------\n# 36 \u683c\u53c2\u6570\u626b\u63cf\uff08\u6162\uff0c\u9ed8\u8ba4\u5173\u95ed\uff09\n# -------------------------\nRUN_GRID_SWEEP = False\n\nif RUN_GRID_SWEEP:\n    grid = [(s, i, t)\n            for s in [-0.02, -0.03, -0.04, -0.05]\n            for i in [0.15, 0.25, 0.35]\n            for t in [5, 10, 15]]\n\n    rows = []\n    for s, i, t in grid:\n        g_sig = build_signals(df, stretch=s, ibs_in=i)\n        e = g_sig[\"Entry\"].shift(1).fillna(False).astype(bool).to_numpy()\n        x = g_sig[\"Exit\"].shift(1).fillna(False).astype(bool).to_numpy()\n        x = apply_time_stop(e, x, max_hold=t)\n        g_pf = vbt.Portfolio.from_signals(close=df[\"Open\"], entries=e, exits=x,\n                                          init_cash=INIT_CASH, fees=FEES,\n                                          slippage=SLIPPAGE, freq=\"1D\")\n        g_tr = g_pf.trades.records_readable\n        rows.append({\n            \"stretch\": s, \"ibs_in\": i, \"time_stop\": t,\n            \"trades\": len(g_tr),\n            \"avg_pct\": g_tr[\"Return\"].mean() * 100 if len(g_tr) else np.nan,\n            \"sharpe\": g_pf.sharpe_ratio(),\n            \"exposure\": g_pf.position_mask().to_numpy().mean() * 100,\n        })\n\n    g_res = pd.DataFrame(rows).sort_values(\"sharpe\", ascending=False)\n    print(g_res.to_string(index=False))\n    print(f\"\\n\u7f51\u683c\u4e2d\u4f4d\u590f\u666e: {g_res['sharpe'].median():.3f}  \"\n          f\"\u590f\u666e>0.5 \u5360\u6bd4: {(g_res['sharpe'] > 0.5).mean() * 100:.0f}%\")<\/code><\/pre>\n\n\n\n
      \u6570\u636e\uff1aSPY \u65e5\u7ebf\uff0c\u6765\u6e90 Yahoo Finance\uff0c\u533a\u95f4 1993-01 \u81f3 2026-08\u3002\u56de\u6d4b\u6846\u67b6 vectorbt\uff0c\u6210\u672c 1bp \u624b\u7eed\u8d39 + 2bp \u6ed1\u70b9\u3002\u6587\u4e2d\u53cd\u63a8\u7b56\u7565\u4e0e\u89c6\u9891\u4e3b\u8bb2\u4eba\u5ba3\u79f0\u7684\u539f\u59cb\u7cfb\u7edf\u5b58\u5728\u5dee\u5f02\uff0c\u8be6\u89c1\u6b63\u6587\u3002<\/p>\n\n\n\n
      \n\n\n\n
      #\u5747\u503c\u56de\u5f52<\/a> #\u6a61\u76ae\u7b4b\u7b56\u7565<\/a> #IBS<\/a> #SPY<\/a> #\u8d8b\u52bf\u8fc7\u6ee4<\/a> #\u53c2\u6570\u626b\u63cf<\/a> #\u5e02\u573a\u655e\u53e3<\/a> #\u590f\u666e\u6bd4\u7387<\/a> #\u7b56\u7565\u53e0\u52a0<\/a> #Python\u91cf\u5316<\/a> #vectorbt<\/a> #\u65f6\u95f4\u6b62\u635f<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"
      \u4f5c \u8005\uff1a\u8001\u4f59\u635e\u9c7c \u539f\u521b\u4e0d\u6613\uff0c\u8f6c\u8f7d\u8bf7\u6807\u660e\u51fa\u5904\u53ca\u539f\u4f5c\u8005…<\/p>\n
      Continue reading\u7b56\u7565\u5e74\u53161.66%\uff0c\u8dd1\u8f93SPY\u4e5d\u4e2a\u70b9\uff0c\u6211\u5374\u6ca1\u5220\u5b83\uff08\u9644\u5b8c\u6574\u6e90\u7801\uff09<\/span><\/a><\/div>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[17,4,5,6],"class_list":["post-4855","post","type-post","status-publish","format-standard","hentry","category-aiinvest","tag-ai","tag-6","entry"],"_links":{"self":[{"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/posts\/4855","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/comments?post=4855"}],"version-history":[{"count":1,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/posts\/4855\/revisions"}],"predecessor-version":[{"id":4863,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/posts\/4855\/revisions\/4863"}],"wp:attachment":[{"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/media?parent=4855"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/categories?post=4855"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/laoyulaoyu.com\/index.php\/wp-json\/wp\/v2\/tags?post=4855"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}