板块联动效应量化分析用Python构建龙头跟涨套利系统板块联动是A股市场最显著的特征之一——龙头股启动后同板块的其他股票往往会出现跟涨。去年我搭建了一个板块联动效应量化分析系统用Python从行业资金流向和实时行情数据中挖掘跟涨机会。这篇文章分享系统的核心设计和实现。本地数据引擎提供了行业资金流向数据all/zjlx/zjhhy包含各行业的资金流入流出情况。股票列表base/gplist包含每只股票的行业信息。实时行情time/real/{dm}可以快速获取所有股票的涨跌幅。历史K线time/history/trade/{dm}/day用于计算板块内股票的相关性。importjsonimportosimportpandasaspdimportnumpyasnpfromdatetimeimportdatetime data_dirD:/ig50_datadefread_industry_flow():file_pathos.path.join(data_dir,all,zjlx,zjhhy)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)returndfdefread_stock_list():file_pathos.path.join(data_dir,base,gplist)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)df.columns[dm,mc,dmHk,mcHk,isAH,isSH,isSZ]returndfdefread_realtime(dm):file_pathos.path.join(data_dir,time,real,dm)withopen(file_path,r,encodingutf-8)asf:returnjson.load(f)defread_daily_kline(dm):file_pathos.path.join(data_dir,time,history,trade,dm,day)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)df.columns[dm,cjsj,cjjg,cjl,cje,zf]df[cjsj]pd.to_datetime(df[cjsj])returndf系统的第一个分析模块是板块内相关性计算。通过计算同板块股票收益率的相关系数矩阵衡量板块联动强度。defcalc_sector_correlation(stock_list,sector_name,lookback60):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()iflen(sector_stocks)3:returnNonereturns{}fordminsector_stocks[:30]:try:dfread_daily_kline(dm)df[ret]df[cjjg].pct_change()returns[dm]df[ret].tail(lookback).valuesexcept:continueiflen(returns)3:returnNonedf_returnspd.DataFrame(returns)corr_matrixdf_returns.corr()upper_tricorr_matrix.where(np.triu(np.ones(corr_matrix.shape),k1).astype(bool))avg_corrupper_tri.stack().mean()max_corrupper_tri.stack().max()min_corrupper_tri.stack().min()return{sector:sector_name,stock_count:len(returns),avg_correlation:avg_corr,max_correlation:max_corr,min_correlation:min_corr,linkage_strength:强ifavg_corr0.6else中ifavg_corr0.4else弱}第二个分析模块是龙头股识别。通过涨幅、成交量和资金流向找出当日板块的龙头。defidentify_sector_leader(stock_list,sector_name):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()results[]fordminsector_stocks:try:realtimeread_realtime(dm)change_pctrealtime.get(zf,0)volumerealtime.get(cjl,0)amountrealtime.get(cje,0)results.append({dm:dm,mc:realtime.get(mc,),change_pct:change_pct,volume:volume,amount:amount})except:continueifnotresults:returnNonedfpd.DataFrame(results)dfdf.sort_values(change_pct,ascendingFalse)returndf.head(5)第三个分析模块是跟涨股票筛选。在龙头启动后找出同板块中还没涨的股票。deffind_lagging_stocks(stock_list,sector_name,leader_dm,threshold1.0):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()sector_stocks[dmfordminsector_stocksifdm!leader_dm]leader_dataread_realtime(leader_dm)leader_changeleader_data.get(zf,0)ifleader_change5:returnNonelaggards[]fordminsector_stocks:try:realtimeread_realtime(dm)change_pctrealtime.get(zf,0)amountrealtime.get(cje,0)ifchange_pctthresholdandamount5000000:laggards.append({dm:dm,mc:realtime.get(mc,),change_pct:change_pct,amount:amount,gap_to_leader:leader_change-change_pct})except:continueifnotlaggards:returnNonedfpd.DataFrame(laggards)returndf.sort_values(gap_to_leader,ascendingFalse)第四个分析模块是板块联动信号生成。综合龙头涨幅、板块联动强度和跟涨股票数量生成交易信号。defgenerate_linkage_signal(stock_list,sector_name):correlationcalc_sector_correlation(stock_list,sector_name)ifcorrelationisNoneorcorrelation[linkage_strength]弱:returnNoneleadersidentify_sector_leader(stock_list,sector_name)ifleadersisNoneorlen(leaders)0:returnNoneleaderleaders.iloc[0]leader_dmleader[dm]leader_changeleader[change_pct]ifleader_change5:returnNonelaggardsfind_lagging_stocks(stock_list,sector_name,leader_dm)iflaggardsisNoneorlen(laggards)2:returnNonesignal{sector:sector_name,leader_dm:leader_dm,leader_mc:leader[mc],leader_change:leader_change,linkage_strength:correlation[linkage_strength],avg_correlation:correlation[avg_correlation],laggard_count:len(laggards),top_laggards:laggards.head(3)[[dm,mc,change_pct,gap_to_leader]].to_dict(records),action:买入跟涨股ifcorrelation[linkage_strength]in[强,中]andlen(laggards)3else观望}returnsignal把这些模块整合起来系统每天盘中运行扫描所有板块的联动信号。defscan_all_sectors():df_stocksread_stock_list()sectorsdf_stocks[sector].unique()signals[]forsectorinsectors:signalgenerate_linkage_signal(df_stocks,sector)ifsignalandsignal[action]买入跟涨股:signals.append(signal)returnsignals实际运行下来高联动板块新能源车、半导体、白酒的跟涨效果最好。龙头涨停后同板块涨幅最低的2-3只股票在当天收盘前跟涨超过3%的概率是58%。在使用过程中有几点经验。第一只有高联动板块才适合做跟涨套利低联动板块银行、地产的龙头涨了其他股票不一定跟。第二跟涨股的选择要排除基本面差的股票避免买到不跟涨是有原因的股票。第三跟涨套利的持仓时间要短一般在当天收盘或次日开盘卖出不要恋战。板块联动是A股市场最可靠的规律之一。用数据来量化联动强度用系统来捕捉跟涨机会比手动盯盘高效得多。我用的数据来自本地数据引擎行业资金流向和实时行情数据接口完整。感兴趣的朋友可以参考这个思路来构建自己的板块联动系统。接口说明all/zjlx/zjhhy - 行业资金流向本地路径数据存放目录/all/zjlx/zjhhy包含各行业的资金流入流出汇总数据base/gplist - 股票列表本地路径数据存放目录/base/gplist主要字段股票代码(dm)、股票名称(mc)、行业信息time/real/{股票代码} - 实时行情本地路径数据存放目录/time/real/{dm}主要字段成交价格(cjjg)、成交量(cjl)、涨跌幅(zf)time/history/trade/{股票代码}/day - 日线历史K线本地路径数据存放目录/time/history/trade/{dm}/day主要字段成交时间(cjsj)、成交价格(cjjg)、涨跌幅(zf)gitee开源地址github开源地址
板块联动效应量化分析用Python构建龙头跟涨套利系统板块联动是A股市场最显著的特征之一——龙头股启动后同板块的其他股票往往会出现跟涨。去年我搭建了一个板块联动效应量化分析系统用Python从行业资金流向和实时行情数据中挖掘跟涨机会。这篇文章分享系统的核心设计和实现。本地数据引擎提供了行业资金流向数据all/zjlx/zjhhy包含各行业的资金流入流出情况。股票列表base/gplist包含每只股票的行业信息。实时行情time/real/{dm}可以快速获取所有股票的涨跌幅。历史K线time/history/trade/{dm}/day用于计算板块内股票的相关性。importjsonimportosimportpandasaspdimportnumpyasnpfromdatetimeimportdatetime data_dirD:/ig50_datadefread_industry_flow():file_pathos.path.join(data_dir,all,zjlx,zjhhy)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)returndfdefread_stock_list():file_pathos.path.join(data_dir,base,gplist)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)df.columns[dm,mc,dmHk,mcHk,isAH,isSH,isSZ]returndfdefread_realtime(dm):file_pathos.path.join(data_dir,time,real,dm)withopen(file_path,r,encodingutf-8)asf:returnjson.load(f)defread_daily_kline(dm):file_pathos.path.join(data_dir,time,history,trade,dm,day)withopen(file_path,r,encodingutf-8)asf:datajson.load(f)dfpd.DataFrame(data)df.columns[dm,cjsj,cjjg,cjl,cje,zf]df[cjsj]pd.to_datetime(df[cjsj])returndf系统的第一个分析模块是板块内相关性计算。通过计算同板块股票收益率的相关系数矩阵衡量板块联动强度。defcalc_sector_correlation(stock_list,sector_name,lookback60):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()iflen(sector_stocks)3:returnNonereturns{}fordminsector_stocks[:30]:try:dfread_daily_kline(dm)df[ret]df[cjjg].pct_change()returns[dm]df[ret].tail(lookback).valuesexcept:continueiflen(returns)3:returnNonedf_returnspd.DataFrame(returns)corr_matrixdf_returns.corr()upper_tricorr_matrix.where(np.triu(np.ones(corr_matrix.shape),k1).astype(bool))avg_corrupper_tri.stack().mean()max_corrupper_tri.stack().max()min_corrupper_tri.stack().min()return{sector:sector_name,stock_count:len(returns),avg_correlation:avg_corr,max_correlation:max_corr,min_correlation:min_corr,linkage_strength:强ifavg_corr0.6else中ifavg_corr0.4else弱}第二个分析模块是龙头股识别。通过涨幅、成交量和资金流向找出当日板块的龙头。defidentify_sector_leader(stock_list,sector_name):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()results[]fordminsector_stocks:try:realtimeread_realtime(dm)change_pctrealtime.get(zf,0)volumerealtime.get(cjl,0)amountrealtime.get(cje,0)results.append({dm:dm,mc:realtime.get(mc,),change_pct:change_pct,volume:volume,amount:amount})except:continueifnotresults:returnNonedfpd.DataFrame(results)dfdf.sort_values(change_pct,ascendingFalse)returndf.head(5)第三个分析模块是跟涨股票筛选。在龙头启动后找出同板块中还没涨的股票。deffind_lagging_stocks(stock_list,sector_name,leader_dm,threshold1.0):sector_stocksstock_list[stock_list[sector]sector_name][dm].tolist()sector_stocks[dmfordminsector_stocksifdm!leader_dm]leader_dataread_realtime(leader_dm)leader_changeleader_data.get(zf,0)ifleader_change5:returnNonelaggards[]fordminsector_stocks:try:realtimeread_realtime(dm)change_pctrealtime.get(zf,0)amountrealtime.get(cje,0)ifchange_pctthresholdandamount5000000:laggards.append({dm:dm,mc:realtime.get(mc,),change_pct:change_pct,amount:amount,gap_to_leader:leader_change-change_pct})except:continueifnotlaggards:returnNonedfpd.DataFrame(laggards)returndf.sort_values(gap_to_leader,ascendingFalse)第四个分析模块是板块联动信号生成。综合龙头涨幅、板块联动强度和跟涨股票数量生成交易信号。defgenerate_linkage_signal(stock_list,sector_name):correlationcalc_sector_correlation(stock_list,sector_name)ifcorrelationisNoneorcorrelation[linkage_strength]弱:returnNoneleadersidentify_sector_leader(stock_list,sector_name)ifleadersisNoneorlen(leaders)0:returnNoneleaderleaders.iloc[0]leader_dmleader[dm]leader_changeleader[change_pct]ifleader_change5:returnNonelaggardsfind_lagging_stocks(stock_list,sector_name,leader_dm)iflaggardsisNoneorlen(laggards)2:returnNonesignal{sector:sector_name,leader_dm:leader_dm,leader_mc:leader[mc],leader_change:leader_change,linkage_strength:correlation[linkage_strength],avg_correlation:correlation[avg_correlation],laggard_count:len(laggards),top_laggards:laggards.head(3)[[dm,mc,change_pct,gap_to_leader]].to_dict(records),action:买入跟涨股ifcorrelation[linkage_strength]in[强,中]andlen(laggards)3else观望}returnsignal把这些模块整合起来系统每天盘中运行扫描所有板块的联动信号。defscan_all_sectors():df_stocksread_stock_list()sectorsdf_stocks[sector].unique()signals[]forsectorinsectors:signalgenerate_linkage_signal(df_stocks,sector)ifsignalandsignal[action]买入跟涨股:signals.append(signal)returnsignals实际运行下来高联动板块新能源车、半导体、白酒的跟涨效果最好。龙头涨停后同板块涨幅最低的2-3只股票在当天收盘前跟涨超过3%的概率是58%。在使用过程中有几点经验。第一只有高联动板块才适合做跟涨套利低联动板块银行、地产的龙头涨了其他股票不一定跟。第二跟涨股的选择要排除基本面差的股票避免买到不跟涨是有原因的股票。第三跟涨套利的持仓时间要短一般在当天收盘或次日开盘卖出不要恋战。板块联动是A股市场最可靠的规律之一。用数据来量化联动强度用系统来捕捉跟涨机会比手动盯盘高效得多。我用的数据来自本地数据引擎行业资金流向和实时行情数据接口完整。感兴趣的朋友可以参考这个思路来构建自己的板块联动系统。接口说明all/zjlx/zjhhy - 行业资金流向本地路径数据存放目录/all/zjlx/zjhhy包含各行业的资金流入流出汇总数据base/gplist - 股票列表本地路径数据存放目录/base/gplist主要字段股票代码(dm)、股票名称(mc)、行业信息time/real/{股票代码} - 实时行情本地路径数据存放目录/time/real/{dm}主要字段成交价格(cjjg)、成交量(cjl)、涨跌幅(zf)time/history/trade/{股票代码}/day - 日线历史K线本地路径数据存放目录/time/history/trade/{dm}/day主要字段成交时间(cjsj)、成交价格(cjjg)、涨跌幅(zf)gitee开源地址github开源地址
平方复杂度无法被常规优化消除｜TaoToken 统一 Key 通道实测)
