Backtest Expert\u30fb\u591a\u5e02\u573a\u91cf\u5316\u7b56\u7565\u56de\u6d4b\u4e13\u5bb6 | \u5168\u56e0\u5b50\u4eff\u771f + \u5b9e\u76d8\u62df\u5408 + AI \u8c03\u53c2 + \u6279\u91cf\u7b56\u7565\u8bc4\u6d4b + \u4e00\u4f53\u5316\u91cf\u5316\u5de5\u4f5c\u53f0<\/strong><\/p>\n\n\n\n \u9762\u5411 A \u80a1\u3001\u671f\u8d27\u3001ETF\u3001\u6307\u6570\u7684\u4e13\u4e1a\u7ea7\u4ea4\u6613\u7b56\u7565\u56de\u6d4b\u5f15\u64ce<\/strong>\uff0c\u533a\u522b\u4e8e\u666e\u901a\u56de\u6d4b\u6846\u67b6\uff0c\u4e3b\u6253\u6781\u81f4\u5b9e\u76d8\u62df\u5408\u3001\u591a\u7ef4\u5ea6\u98ce\u9669\u6307\u6807\u3001\u53ef\u89c6\u5316\u5f52\u56e0\u3001\u7b56\u7565\u6279\u91cf\u8bc4\u6d4b\u3001AI \u667a\u80fd\u4f18\u5316\u3001\u6ed1\u70b9 \/ \u624b\u7eed\u8d39 \/ \u6da8\u8dcc\u505c\u5168\u771f\u6a21\u62df<\/strong>\uff0c\u652f\u6301\u81ea\u5b9a\u4e49\u6307\u6807\u3001\u591a\u5468\u671f\u5171\u632f\u3001\u591a\u54c1\u79cd\u7ec4\u5408\u7b56\u7565\uff0c\u6df1\u5ea6\u5bf9\u63a5\u524d\u6587 OpenClaw \u5168\u7cfb\u5de5\u5177\u94fe\uff0c\u5b9e\u73b0\u300c\u884c\u60c5\u76d1\u63a7\u2192\u7b56\u7565\u56de\u6d4b\u2192\u5b9e\u76d8\u4fe1\u53f7\u2192\u98ce\u63a7\u544a\u8b66\u300d\u5168\u94fe\u8def\u81ea\u52a8\u5316\u3002<\/p>\n\n\n\n plaintext<\/p>\n\n\n\n txt<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n python<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n bash<\/p>\n\n\n\n \u8fd0\u884c<\/p>\n\n\n\n plaintext<\/p>\n\n\n\n plaintext<\/p>\n\n\n\n \u4e00\u3001\u4ea7\u54c1\u72ec\u7acb\u5b9a\u4f4d & \u5dee\u5f02\u5316\u7279\u8272\uff08\u72ec\u6709\u6539\u9020\u70b9\uff09 \u4e13\u5c5e\u6807\u9898 Backtest Expert\u30fb\u591a\u5e02\u573a\u91cf\u5316\u7b56 […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[],"class_list":["post-299","post","type-post","status-publish","format-standard","hentry","category-mianfeiercikaifalongxiaskill"],"_links":{"self":[{"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/posts\/299","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/comments?post=299"}],"version-history":[{"count":1,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/posts\/299\/revisions"}],"predecessor-version":[{"id":300,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/posts\/299\/revisions\/300"}],"wp:attachment":[{"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/media?parent=299"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/categories?post=299"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.longxiapro.com\/index.php\/wp-json\/wp\/v2\/tags?post=299"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\u6838\u5fc3\u5b9a\u4f4d<\/h3>\n\n\n\n
\u72ec\u5bb6\u6539\u9020\u7279\u8272\uff08\u533a\u522b\u4e8e\u4f20\u7edf\u56de\u6d4b\u6846\u67b6\uff09<\/h3>\n\n\n\n
\n
Memory Tiering<\/code>\uff0c\u5386\u53f2 K \u7ebf \/ \u56de\u6d4b\u62a5\u544a\u51b7\u6570\u636e\u4e0b\u6c89\u5b58\u50a8\uff0c\u9ad8\u9891\u56de\u6d4b\u8ba1\u7b97\u70ed\u6570\u636e\u5e38\u9a7b\u5185\u5b58\uff0c\u5927\u5e45\u63d0\u5347\u5927\u6837\u672c\u56de\u6d4b\u901f\u5ea6\u3002<\/li>\n\n\n\nStock Monitor<\/code> \u5b9e\u65f6\u884c\u60c5\u3001Baidu Web Search<\/code> \u8206\u60c5\u56e0\u5b50\u3001Social Media Scheduler<\/code> \u7b56\u7565\u62a5\u544a\u5206\u53d1\u3001Remotion<\/code> \u56de\u6d4b\u52a8\u6001\u89c6\u9891\u590d\u76d8\u3002<\/li>\n\n\n\n\u4e8c\u3001\u4f20\u7edf\u56de\u6d4b\u5de5\u5177\u75db\u70b9<\/h3>\n\n\n\n
\n
\u4e09\u3001\u6838\u5fc3\u529f\u80fd\u6a21\u5757<\/h3>\n\n\n\n
1. \u6570\u636e\u6e90\u6a21\u5757<\/h3>\n\n\n\n
\n
2. \u5168\u771f\u4ea4\u6613\u6a21\u62df\u6a21\u5757<\/h3>\n\n\n\n
\n
3. \u7b56\u7565\u5f15\u64ce\u6a21\u5757<\/h3>\n\n\n\n
\n
4. \u7ee9\u6548\u5206\u6790 & \u5f52\u56e0\u6a21\u5757<\/h3>\n\n\n\n
\n
5. AI \u53c2\u6570\u4f18\u5316\u6a21\u5757<\/h3>\n\n\n\n
\n
6. \u6279\u91cf\u56de\u6d4b & \u4efb\u52a1\u8c03\u5ea6\u6a21\u5757<\/h3>\n\n\n\n
\n
7. \u53ef\u89c6\u5316 & \u5bfc\u51fa\u6a21\u5757<\/h3>\n\n\n\n
\n
\u56db\u3001\u9879\u76ee\u76ee\u5f55\u7ed3\u6784<\/h3>\n\n\n\n
backtest-expert\/\n\u251c\u2500\u2500 config\/ # \u5168\u5c40\u914d\u7f6e\u3001\u4ea4\u6613\u89c4\u5219\u3001\u6570\u636e\u6e90\u914d\u7f6e\n\u251c\u2500\u2500 core\/ # \u56de\u6d4b\u5185\u6838\u3001\u4ea4\u6613\u6a21\u62df\u3001\u64ae\u5408\u5f15\u64ce\n\u251c\u2500\u2500 data\/ # \u5386\u53f2\u884c\u60c5\u6570\u636e\u3001\u6570\u636e\u6e05\u6d17\u5de5\u5177\n\u251c\u2500\u2500 strategy\/ # \u7b56\u7565\u76ee\u5f55\u3001\u5185\u7f6e\u7b56\u7565\u3001\u81ea\u5b9a\u4e49\u7b56\u7565\u63d2\u4ef6\n\u251c\u2500\u2500 indicator\/ # \u6280\u672f\u6307\u6807\u5e93\u3001\u81ea\u5b9a\u4e49\u6307\u6807\u6269\u5c55\n\u251c\u2500\u2500 analysis\/ # \u7ee9\u6548\u8ba1\u7b97\u3001\u5f52\u56e0\u5206\u6790\u3001\u62a5\u8868\u751f\u6210\n\u251c\u2500\u2500 ai_optimize\/ # AI\u53c2\u6570\u4f18\u5316\u7b97\u6cd5\n\u251c\u2500\u2500 task\/ # \u4efb\u52a1\u961f\u5217\u3001\u6279\u91cf\u8c03\u5ea6\n\u251c\u2500\u2500 api\/ # FastAPI\u5bf9\u5916\u63a5\u53e3\u3001OpenClaw\u5bf9\u63a5\u63a5\u53e3\n\u251c\u2500\u2500 utils\/ # \u5de5\u5177\u51fd\u6570\u3001\u65e5\u5fd7\u3001\u7f13\u5b58\u3001\u5185\u5b58\u5206\u5c42\u9002\u914d\n\u251c\u2500\u2500 web\/ # \u524d\u7aef\u53ef\u89c6\u5316\u770b\u677f\n\u251c\u2500\u2500 requirements.txt # \u4f9d\u8d56\u5305\n\u2514\u2500\u2500 main.py # \u9879\u76ee\u5165\u53e3\n<\/code><\/pre>\n\n\n\n\u4e94\u3001\u5b8c\u6574\u4e8c\u6b21\u5f00\u53d1\u4ee3\u7801<\/h3>\n\n\n\n
1. \u4f9d\u8d56\u6587\u4ef6 requirements.txt<\/h3>\n\n\n\n
# \u57fa\u7840\u8ba1\u7b97&\u6570\u636e\nnumpy>=1.24.0\npandas>=2.0.0\ntushare>=1.7.9\nakshare>=1.11.0\n\n# \u53ef\u89c6\u5316\nmatplotlib>=3.7.0\nplotly>=5.15.0\n\n# \u63a5\u53e3&\u8c03\u5ea6\nfastapi>=0.104.0\nuvicorn>=0.24.0\napscheduler>=3.10.0\n\n# AI\u53c2\u6570\u4f18\u5316\nscikit-optimize>=0.9.0\ndeap>=1.3.3\n\n# \u5de5\u5177\nloguru>=0.7.2\npython-dotenv>=1.0.0\nredis>=5.0.0\nopenpyxl>=3.1.0\n<\/code><\/pre>\n\n\n\n2. \u5168\u5c40\u914d\u7f6e config\/settings.py<\/h3>\n\n\n\n
import os\nfrom dotenv import load_dotenv\n\nload_dotenv()\n\n# \u6570\u636e\u6e90\u914d\u7f6e\nTUSHARE_TOKEN = os.getenv(\"TUSHARE_TOKEN\", \"\")\nDATA_CACHE_DIR = \".\/data\/cache\"\nHIST_DATA_DIR = \".\/data\/history\"\n\n# \u4ea4\u6613\u6a21\u62df\u5168\u5c40\u914d\u7f6e\uff08\u5b9e\u76d8\u62df\u5408\u6838\u5fc3\uff09\nTRADING_RULE = {\n \"trade_type\": \"T1\", # T1 A\u80a1 \/ T0 \u671f\u8d27\n \"commission_rate\": 0.0003, # \u4f63\u91d1\u8d39\u7387\n \"stamp_tax\": 0.001, # \u5370\u82b1\u7a0e\uff08\u5356\u51fa\u6536\u53d6\uff09\n \"transfer_fee\": 0.00001, # \u8fc7\u6237\u8d39\n \"slippage\": 0.0005, # \u56fa\u5b9a\u6ed1\u70b9\n \"limit_up\": 0.1, # \u6da8\u505c\u5e45\u5ea6 10%\n \"limit_down\": -0.1, # \u8dcc\u505c\u5e45\u5ea6 -10%\n \"max_position_ratio\": 0.8, # \u6700\u5927\u4ed3\u4f4d 80%\n \"single_max_order\": 10000 # \u5355\u7b14\u6700\u5927\u4e0b\u5355\u6570\u91cf\n}\n\n# \u56de\u6d4b\u57fa\u7840\u914d\u7f6e\nDEFAULT_START_DATE = \"2023-01-01\"\nDEFAULT_END_DATE = \"2026-06-01\"\nINIT_CAPITAL = 1000000 # \u521d\u59cb\u8d44\u91d1\n\n# \u5185\u5b58\u5206\u5c42\uff08\u5bf9\u63a5 Memory Tiering\uff09\nHOT_DATA_TTL = 3600\nCOLD_DATA_COMPRESS = True\n\n# API\u914d\u7f6e\nAPI_HOST = \"0.0.0.0\"\nAPI_PORT = 8090\n<\/code><\/pre>\n\n\n\n3. \u65e5\u5fd7\u5de5\u5177 utils\/logger.py<\/h3>\n\n\n\n
from loguru import logger\nimport os\nfrom config.settings import DATA_CACHE_DIR\n\nlog_path = os.path.join(DATA_CACHE_DIR, \"backtest_log.log\")\n\n# \u65e5\u5fd7\u521d\u59cb\u5316\nlogger.add(\n log_path,\n rotation=\"500 MB\",\n retention=\"30 days\",\n level=\"INFO\",\n encoding=\"utf-8\"\n)\n\ndef get_logger():\n return logger\n<\/code><\/pre>\n\n\n\n4. \u6570\u636e\u5f15\u64ce core\/data_engine.py\uff08\u884c\u60c5\u83b7\u53d6 & \u6e05\u6d17\uff09<\/h3>\n\n\n\n
import tushare\nimport pandas as pd\nimport os\nfrom datetime import datetime\nfrom config.settings import TUSHARE_TOKEN, HIST_DATA_DIR\nfrom utils.logger import get_logger\n\nlogger = get_logger()\ntushare.set_token(TUSHARE_TOKEN)\npro = tushare.pro_api()\n\nclass DataEngine:\n def __init__(self):\n self.cache = {}\n\n def get_stock_daily(self, ts_code, start_date, end_date):\n \"\"\"\u83b7\u53d6\u4e2a\u80a1\u65e5\u7ebf\u6570\u636e + \u9664\u6743\u9664\u606f\u6e05\u6d17\"\"\"\n cache_key = f\"{ts_code}_{start_date}_{end_date}\"\n if cache_key in self.cache:\n return self.cache[cache_key]\n\n try:\n # \u62c9\u53d6\u65e5\u7ebf\u6570\u636e\n df = pro.daily(ts_code=ts_code, start_date=start_date.replace(\"-\", \"\"),\n end_date=end_date.replace(\"-\", \"\"))\n if df.empty:\n logger.warning(f\"\u65e0\u884c\u60c5\u6570\u636e: {ts_code}\")\n return None\n\n # \u6570\u636e\u683c\u5f0f\u5316\n df[\"trade_date\"] = pd.to_datetime(df[\"trade_date\"])\n df = df.sort_values(\"trade_date\").reset_index(drop=True)\n df.rename(\n columns={\n \"trade_date\": \"date\",\n \"open\": \"open\",\n \"high\": \"high\",\n \"low\": \"low\",\n \"close\": \"close\",\n \"vol\": \"volume\",\n \"amount\": \"amount\"\n }, inplace=True\n )\n # \u5254\u9664\u505c\u724c\u6570\u636e\n df = df[df[\"volume\"] > 0]\n self.cache[cache_key] = df\n logger.info(f\"\u884c\u60c5\u6570\u636e\u52a0\u8f7d\u5b8c\u6210: {ts_code} \u5171{len(df)}\u6761\")\n return df\n except Exception as e:\n logger.error(f\"\u884c\u60c5\u83b7\u53d6\u5931\u8d25 {ts_code}: {str(e)}\")\n return None\n\n# \u5168\u5c40\u6570\u636e\u5b9e\u4f8b\ndata_engine = DataEngine()\n<\/code><\/pre>\n\n\n\n5. \u6280\u672f\u6307\u6807\u5e93 indicator\/base_indicator.py<\/h3>\n\n\n\n
import numpy as np\nimport pandas as pd\n\nclass BaseIndicator:\n @staticmethod\n def calc_ma(df, period=5, col=\"close\"):\n \"\"\"\u79fb\u52a8\u5e73\u5747\u7ebf MA\"\"\"\n df[f\"ma{period}\"] = df[col].rolling(window=period).mean()\n return df\n\n @staticmethod\n def calc_macd(df, fast=12, slow=26, signal=9):\n \"\"\"MACD\u6307\u6807\"\"\"\n ema_fast = df[\"close\"].ewm(span=fast, adjust=False).mean()\n ema_slow = df[\"close\"].ewm(span=slow, adjust=False).mean()\n df[\"diff\"] = ema_fast - ema_slow\n df[\"dea\"] = df[\"diff\"].ewm(span=signal, adjust=False).mean()\n df[\"macd\"] = (df[\"diff\"] - df[\"dea\"]) * 2\n return df\n\n @staticmethod\n def calc_kdj(df, n=9, m1=3, m2=3):\n \"\"\"KDJ\u6307\u6807\"\"\"\n low_list = df[\"low\"].rolling(n).min()\n high_list = df[\"high\"].rolling(n).max()\n rsv = (df[\"close\"] - low_list) \/ (high_list - low_list) * 100\n df[\"k\"] = rsv.ewm(span=m1, adjust=False).mean()\n df[\"d\"] = df[\"k\"].ewm(span=m2, adjust=False).mean()\n df[\"j\"] = 3 * df[\"k\"] - 2 * df[\"d\"]\n return df\n\n @staticmethod\n def add_all_indicators(df):\n \"\"\"\u6279\u91cf\u52a0\u8f7d\u5e38\u7528\u6307\u6807\"\"\"\n df = BaseIndicator.calc_ma(df, 5)\n df = BaseIndicator.calc_ma(df, 10)\n df = BaseIndicator.calc_ma(df, 20)\n df = BaseIndicator.calc_macd(df)\n df = BaseIndicator.calc_kdj(df)\n return df\n<\/code><\/pre>\n\n\n\n6. \u4ea4\u6613\u6a21\u62df\u5185\u6838 core\/trading_sim.py\uff08\u5b9e\u76d8\u62df\u5408\u6838\u5fc3\uff09<\/h3>\n\n\n\n
import pandas as pd\nfrom config.settings import TRADING_RULE, INIT_CAPITAL\nfrom utils.logger import get_logger\n\nlogger = get_logger()\n\nclass TradingSimulator:\n def __init__(self):\n self.init_capital = INIT_CAPITAL\n self.cash = INIT_CAPITAL # \u53ef\u7528\u8d44\u91d1\n self.hold_shares = 0 # \u6301\u4ed3\u6570\u91cf\n self.hold_cost = 0 # \u6301\u4ed3\u6210\u672c\n self.trade_records = [] # \u4ea4\u6613\u8bb0\u5f55\n self.net_value_list = [] # \u6bcf\u65e5\u51c0\u503c\n\n def calc_trade_cost(self, price, volume, is_buy=True):\n \"\"\"\u8ba1\u7b97\u4ea4\u6613\u6210\u672c\uff1a\u4f63\u91d1\u3001\u5370\u82b1\u7a0e\u3001\u6ed1\u70b9\"\"\"\n # \u6ed1\u70b9\u5904\u7406\n if is_buy:\n real_price = price * (1 + TRADING_RULE[\"slippage\"])\n else:\n real_price = price * (1 - TRADING_RULE[\"slippage\"])\n\n total_money = real_price * volume\n commission = total_money * TRADING_RULE[\"commission_rate\"]\n transfer_fee = total_money * TRADING_RULE[\"transfer_fee\"]\n stamp_tax = total_money * TRADING_RULE[\"stamp_tax\"] if not is_buy else 0\n\n total_cost = commission + transfer_fee + stamp_tax\n return real_price, total_cost\n\n def check_limit_price(self, pre_close, price):\n \"\"\"\u68c0\u67e5\u662f\u5426\u89e6\u78b0\u6da8\u8dcc\u505c\uff0c\u65e0\u6cd5\u6210\u4ea4\"\"\"\n limit_up = pre_close * (1 + TRADING_RULE[\"limit_up\"])\n limit_down = pre_close * (1 + TRADING_RULE[\"limit_down\"])\n if price >= limit_up or price <= limit_down:\n return False\n return True\n\n def buy(self, date, price, pre_close, volume):\n \"\"\"\u4e70\u5165\u903b\u8f91\"\"\"\n # \u6da8\u8dcc\u505c\u6821\u9a8c\n if not self.check_limit_price(pre_close, price):\n logger.warning(f\"{date} \u4ef7\u683c\u89e6\u78b0\u6da8\u8dcc\u505c\uff0c\u4e70\u5165\u5931\u8d25\")\n return False\n\n real_price, cost = self.calc_trade_cost(price, volume, is_buy=True)\n total_pay = real_price * volume + cost\n\n # \u8d44\u91d1\u6821\u9a8c\n if total_pay > self.cash:\n logger.warning(f\"{date} \u8d44\u91d1\u4e0d\u8db3\uff0c\u4e70\u5165\u5931\u8d25\")\n return False\n\n self.cash -= total_pay\n self.hold_shares += volume\n self.hold_cost = real_price\n self.trade_records.append({\n \"date\": date,\n \"direction\": \"BUY\",\n \"price\": real_price,\n \"volume\": volume,\n \"cash_after\": self.cash,\n \"hold_shares\": self.hold_shares\n })\n logger.info(f\"{date} \u4e70\u5165\u6210\u529f \u4ef7\u683c:{real_price:.2f} \u6570\u91cf:{volume}\")\n return True\n\n def sell(self, date, price, pre_close, volume):\n \"\"\"\u5356\u51fa\u903b\u8f91\"\"\"\n if self.hold_shares < volume:\n logger.warning(f\"{date} \u6301\u4ed3\u4e0d\u8db3\uff0c\u5356\u51fa\u5931\u8d25\")\n return False\n if not self.check_limit_price(pre_close, price):\n logger.warning(f\"{date} \u4ef7\u683c\u89e6\u78b0\u6da8\u8dcc\u505c\uff0c\u5356\u51fa\u5931\u8d25\")\n return False\n\n real_price, cost = self.calc_trade_cost(price, volume, is_buy=False)\n total_get = real_price * volume - cost\n\n self.cash += total_get\n self.hold_shares -= volume\n self.trade_records.append({\n \"date\": date,\n \"direction\": \"SELL\",\n \"price\": real_price,\n \"volume\": volume,\n \"cash_after\": self.cash,\n \"hold_shares\": self.hold_shares\n })\n logger.info(f\"{date} \u5356\u51fa\u6210\u529f \u4ef7\u683c:{real_price:.2f} \u6570\u91cf:{volume}\")\n return True\n\n def update_net_value(self, date, now_price):\n \"\"\"\u66f4\u65b0\u5f53\u65e5\u8d26\u6237\u51c0\u503c\"\"\"\n net_value = self.cash + self.hold_shares * now_price\n self.net_value_list.append({\n \"date\": date,\n \"net_value\": net_value,\n \"cash\": self.cash,\n \"hold_shares\": self.hold_shares\n })\n return net_value\n<\/code><\/pre>\n\n\n\n7. \u7b56\u7565\u6838\u5fc3 & \u56de\u6d4b\u4e3b\u5f15\u64ce core\/backtest_engine.py<\/h3>\n\n\n\n
from core.data_engine import data_engine\nfrom core.trading_sim import TradingSimulator\nfrom indicator.base_indicator import BaseIndicator\nfrom analysis.performance import PerformanceAnalyzer\nfrom utils.logger import get_logger\n\nlogger = get_logger()\n\nclass BacktestEngine:\n def __init__(self, ts_code, start_date, end_date, strategy_func):\n self.ts_code = ts_code\n self.start_date = start_date\n self.end_date = end_date\n self.strategy = strategy_func\n self.sim = TradingSimulator()\n self.df = None\n\n def run(self):\n \"\"\"\u6267\u884c\u5b8c\u6574\u56de\u6d4b\u6d41\u7a0b\"\"\"\n # 1. \u52a0\u8f7d\u884c\u60c5\n self.df = data_engine.get_stock_daily(self.ts_code, self.start_date, self.end_date)\n if self.df is None or len(self.df) < 30:\n logger.error(\"\u884c\u60c5\u6570\u636e\u4e0d\u8db3\uff0c\u56de\u6d4b\u7ec8\u6b62\")\n return None\n\n # 2. \u8ba1\u7b97\u5168\u90e8\u6280\u672f\u6307\u6807\n self.df = BaseIndicator.add_all_indicators(self.df)\n\n # 3. \u9010K\u7ebf\u6267\u884c\u7b56\u7565\n for i in range(1, len(self.df)):\n row = self.df.iloc[i]\n pre_row = self.df.iloc[i-1]\n date = row[\"date\"]\n pre_close = pre_row[\"close\"]\n now_close = row[\"close\"]\n\n # \u6267\u884c\u81ea\u5b9a\u4e49\u7b56\u7565\u4fe1\u53f7\n signal, volume = self.strategy(self.df, i)\n\n if signal == \"BUY\":\n self.sim.buy(date, now_close, pre_close, volume)\n elif signal == \"SELL\":\n self.sim.sell(date, now_close, pre_close, volume)\n\n # \u66f4\u65b0\u6bcf\u65e5\u51c0\u503c\n self.sim.update_net_value(date, now_close)\n\n # 4. \u7ee9\u6548\u5206\u6790\n analyzer = PerformanceAnalyzer(self.sim.net_value_list, self.sim.trade_records)\n report = analyzer.calc_all_metrics()\n logger.info(f\"\u56de\u6d4b\u5b8c\u6210 {self.ts_code} \u5e74\u5316\u6536\u76ca:{report['annual_return']:.2%} \u6700\u5927\u56de\u64a4:{report['max_drawdown']:.2%}\")\n return {\n \"code\": self.ts_code,\n \"start_date\": self.start_date,\n \"end_date\": self.end_date,\n \"trade_records\": self.sim.trade_records,\n \"net_value\": self.sim.net_value_list,\n \"performance\": report\n }\n<\/code><\/pre>\n\n\n\n8. \u7ee9\u6548\u5206\u6790\u6a21\u5757 analysis\/performance.py\uff08\u5168\u7ef4\u5ea6\u6307\u6807\uff09<\/h3>\n\n\n\n
import pandas as pd\nimport numpy as np\n\nclass PerformanceAnalyzer:\n def __init__(self, net_value_data, trade_data):\n self.net_df = pd.DataFrame(net_value_data)\n self.trade_df = pd.DataFrame(trade_data)\n\n def calc_return(self):\n \"\"\"\u8ba1\u7b97\u603b\u6536\u76ca\u3001\u5e74\u5316\u6536\u76ca\"\"\"\n start_val = self.net_df.iloc[0][\"net_value\"]\n end_val = self.net_df.iloc[-1][\"net_value\"]\n total_return = (end_val - start_val) \/ start_val\n\n days = (self.net_df[\"date\"].iloc[-1] - self.net_df[\"date\"].iloc[0]).days\n annual_return = (1 + total_return) ** (365 \/ days) - 1 if days > 0 else 0\n return total_return, annual_return\n\n def calc_max_drawdown(self):\n \"\"\"\u8ba1\u7b97\u6700\u5927\u56de\u64a4\"\"\"\n net_series = self.net_df[\"net_value\"]\n rolling_max = net_series.cummax()\n drawdown = (net_series - rolling_max) \/ rolling_max\n max_dd = drawdown.min()\n return max_dd\n\n def calc_sharpe(self):\n \"\"\"\u590f\u666e\u6bd4\u7387\uff08\u7b80\u6613\u7248\uff09\"\"\"\n self.net_df[\"pct\"] = self.net_df[\"net_value\"].pct_change()\n daily_return = self.net_df[\"pct\"].mean()\n daily_std = self.net_df[\"pct\"].std()\n if daily_std == 0:\n return 0\n sharpe = daily_return \/ daily_std * np.sqrt(250)\n return sharpe\n\n def calc_win_rate(self):\n \"\"\"\u80dc\u7387\u3001\u76c8\u4e8f\u6bd4\"\"\"\n if len(self.trade_df) < 2:\n return 0, 0\n profit_list = []\n for i in range(1, len(self.trade_df), 2):\n if i >= len(self.trade_df):\n break\n buy_price = self.trade_df.iloc[i-1][\"price\"]\n sell_price = self.trade_df.iloc[i][\"price\"]\n profit = (sell_price - buy_price) \/ buy_price\n profit_list.append(profit)\n\n win_count = len([p for p in profit_list if p > 0])\n win_rate = win_count \/ len(profit_list) if profit_list else 0\n\n profit_avg = np.mean([p for p in profit_list if p > 0]) if [p for p in profit_list if p > 0] else 0\n loss_avg = abs(np.mean([p for p in profit_list if p < 0])) if [p for p in profit_list if p < 0] else 0\n profit_loss_ratio = profit_avg \/ loss_avg if loss_avg != 0 else 0\n return win_rate, profit_loss_ratio\n\n def calc_all_metrics(self):\n \"\"\"\u8ba1\u7b97\u5168\u90e8\u7ee9\u6548\u6307\u6807\"\"\"\n total_ret, annual_ret = self.calc_return()\n max_dd = self.calc_max_drawdown()\n sharpe = self.calc_sharpe()\n win_rate, pl_ratio = self.calc_win_rate()\n calmar = annual_ret \/ abs(max_dd) if max_dd != 0 else 0\n\n return {\n \"total_return\": total_ret,\n \"annual_return\": annual_ret,\n \"max_drawdown\": max_dd,\n \"sharpe_ratio\": sharpe,\n \"calmar_ratio\": calmar,\n \"win_rate\": win_rate,\n \"profit_loss_ratio\": pl_ratio,\n \"trade_times\": len(self.trade_df)\n }\n<\/code><\/pre>\n\n\n\n9. \u793a\u4f8b\u7b56\u7565 strategy\/demo_strategy.py\uff08MACD \u91d1\u53c9\u6b7b\u53c9\u7b56\u7565\uff09<\/h3>\n\n\n\n
\"\"\"\n\u81ea\u5b9a\u4e49\u7b56\u7565\u6a21\u677f\uff1aMACD\u91d1\u53c9\u4e70\u5165\uff0c\u6b7b\u53c9\u5356\u51fa\n\u8fd4\u56de\u683c\u5f0f\uff1a(\u4fe1\u53f7: BUY\/SELL\/NONE, \u4ea4\u6613\u6570\u91cf)\n\"\"\"\ndef macd_strategy(df, index):\n row = df.iloc[index]\n pre_row = df.iloc[index-1]\n signal = \"NONE\"\n volume = 1000 # \u56fa\u5b9a\u4e0b\u5355\u6570\u91cf\n\n # MACD\u91d1\u53c9\uff1adiff\u4e0a\u7a7fdea\n if pre_row[\"diff\"] < pre_row[\"dea\"] and row[\"diff\"] > row[\"dea\"]:\n signal = \"BUY\"\n # MACD\u6b7b\u53c9\uff1adiff\u4e0b\u7a7fdea\n elif pre_row[\"diff\"] > pre_row[\"dea\"] and row[\"diff\"] < row[\"dea\"]:\n signal = \"SELL\"\n\n return signal, volume\n<\/code><\/pre>\n\n\n\n10. AI \u53c2\u6570\u4f18\u5316\u6a21\u5757 ai_optimize\/param_optim.py\uff08\u9057\u4f20\u7b97\u6cd5\u8c03\u53c2\uff09<\/h3>\n\n\n\n
from deap import base, creator, tools, algorithms\nimport random\nfrom core.backtest_engine import BacktestEngine\nfrom strategy.demo_strategy import macd_strategy\n\n# \u9002\u5e94\u5ea6\u51fd\u6570\uff1a\u6700\u5927\u5316\u5e74\u5316\u6536\u76ca\uff0c\u6700\u5c0f\u5316\u56de\u64a4\ncreator.create(\"FitnessMax\", base.Fitness, weights=(1.0, -1.0))\ncreator.create(\"Individual\", list, fitness=creator.FitnessMax)\n\ntoolbox = base.Toolbox()\ntoolbox.register(\"attr_int\", random.randint, 5, 30)\ntoolbox.register(\"individual\", tools.initRepeat, creator.Individual, toolbox.attr_int, n=2)\ntoolbox.register(\"population\", tools.initRepeat, list, toolbox.individual)\n\ndef evaluate(individual):\n fast, slow = individual\n def test_strategy(df, idx):\n df = df.copy()\n df = df.iloc[:idx+1]\n from indicator.base_indicator import BaseIndicator\n df = BaseIndicator.calc_macd(df, fast, slow, 9)\n now = df.iloc[-1]\n pre = df.iloc[-2]\n sig = \"NONE\"\n vol = 1000\n if pre[\"diff\"] < pre[\"dea\"] and now[\"diff\"] > now[\"dea\"]:\n sig = \"BUY\"\n elif pre[\"diff\"] > pre[\"dea\"] and now[\"diff\"] < now[\"dea\"]:\n sig = \"SELL\"\n return sig, vol\n\n bt = BacktestEngine(\"600519.SH\", \"2023-01-01\", \"2026-01-01\", test_strategy)\n res = bt.run()\n if not res:\n return (-999, 999)\n return (res[\"performance\"][\"annual_return\"], res[\"performance\"][\"max_drawdown\"])\n\ntoolbox.register(\"evaluate\", evaluate)\ntoolbox.register(\"mate\", tools.cxTwoPoint)\ntoolbox.register(\"mutate\", tools.mutUniformInt, low=5, up=30, indpb=0.2)\ntoolbox.register(\"select\", tools.selTournament, tournsize=3)\n\ndef run_ga_optimize():\n pop = toolbox.population(n=20)\n hof = tools.HallOfFame(1)\n stats = tools.Statistics(lambda ind: ind.fitness.values)\n stats.register(\"avg\", lambda x: sum(i[0] for i in x)\/len(x))\n stats.register(\"max\", max)\n\n pop, log = algorithms.eaSimple(pop, toolbox, cxpb=0.5, mutpb=0.2, ngen=10, stats=stats, halloffame=hof)\n return hof[0], log\n<\/code><\/pre>\n\n\n\n11. FastAPI \u63a5\u53e3 api\/route.py\uff08\u5bf9\u5916\u63a5\u53e3 + OpenClaw \u5bf9\u63a5\uff09<\/h3>\n\n\n\n
from fastapi import FastAPI, Query\nfrom core.backtest_engine import BacktestEngine\nfrom strategy.demo_strategy import macd_strategy\nfrom ai_optimize.param_optim import run_ga_optimize\n\napp = FastAPI(title=\"Backtest Expert \u4ea4\u6613\u7b56\u7565\u56de\u6d4b\u63a5\u53e3\")\n\n@app.get(\"\/run_backtest\")\ndef run_backtest(\n code: str = Query(..., description=\"\u80a1\u7968\u4ee3\u7801\"),\n start: str = Query(\"2023-01-01\"),\n end: str = Query(\"2026-06-01\")\n):\n \"\"\"\u6267\u884c\u5355\u7b56\u7565\u56de\u6d4b\"\"\"\n bt = BacktestEngine(code, start, end, macd_strategy)\n result = bt.run()\n if not result:\n return {\"code\": code, \"status\": \"fail\", \"msg\": \"\u56de\u6d4b\u5931\u8d25\"}\n return {\"code\": code, \"status\": \"success\", \"data\": result}\n\n@app.get(\"\/optim_param\")\ndef optimize_param():\n \"\"\"AI\u81ea\u52a8\u53c2\u6570\u4f18\u5316\"\"\"\n best_param, log = run_ga_optimize()\n return {\n \"best_macd_fast\": best_param[0],\n \"best_macd_slow\": best_param[1],\n \"optim_log\": log\n }\n\n@app.get(\"\/health\")\ndef health():\n return {\"service\": \"Backtest Expert\", \"status\": \"running\"}\n<\/code><\/pre>\n\n\n\n12. \u9879\u76ee\u5165\u53e3 main.py<\/h3>\n\n\n\n
import uvicorn\nfrom api.route import app\nfrom config.settings import API_HOST, API_PORT\nfrom utils.logger import get_logger\n\nlogger = get_logger()\n\nif __name__ == \"__main__\":\n logger.info(\"Backtest Expert \u4ea4\u6613\u7b56\u7565\u56de\u6d4b\u4e13\u5bb6 \u542f\u52a8\u6210\u529f\")\n uvicorn.run(app, host=API_HOST, port=API_PORT)\n<\/code><\/pre>\n\n\n\n\u516d\u3001\u7279\u8272\u6539\u9020\u843d\u5730\u8bf4\u660e<\/h3>\n\n\n\n
\n
Stock Monitor<\/code> \u5b9e\u65f6\u884c\u60c5\u3001WeChat MP CN<\/code> \u884c\u4e1a\u8206\u60c5\u56e0\u5b50\u3001Remotion<\/code> \u52a8\u6001\u590d\u76d8\u89c6\u9891\u3001Social Media Scheduler<\/code> \u62a5\u544a\u81ea\u52a8\u5206\u53d1\u3002<\/li>\n\n\n\n\u4e03\u3001\u4f7f\u7528\u793a\u4f8b<\/h3>\n\n\n\n
1. \u542f\u52a8\u670d\u52a1<\/h3>\n\n\n\n
python main.py\n<\/code><\/pre>\n\n\n\n2. \u8c03\u7528\u56de\u6d4b\u63a5\u53e3<\/h3>\n\n\n\n
# MACD\u7b56\u7565\u56de\u6d4b \u8d35\u5dde\u8305\u53f0\nhttp://127.0.0.1:8090\/run_backtest?code=600519.SH&start=2023-01-01&end=2026-06-01\n<\/code><\/pre>\n\n\n\n3. AI \u81ea\u52a8\u53c2\u6570\u4f18\u5316<\/h3>\n\n\n\n
http://127.0.0.1:8090\/optim_param\n<\/code><\/pre>\n\n\n\n\u516b\u3001\u62d3\u5c55\u4e8c\u6b21\u5f00\u53d1\u65b9\u5411<\/h3>\n\n\n\n
\n