#核心代码
import pandas as pd import dataApi as DA import cPickle as pickle import threading import glob import os import time import lock from pylab import * from multiprocessing import Process class account(object): def __init__(self): self.universe=[] self.current_date='' self.cash=0 self.secpos={} self.valid_secpos={} self.referencePrice={} self.referencePortfolioValue=0 self.blotter=[] self.days_counter=0 class StrategyBase(object): ''' @var start:回测开始时间 @var end:回测结束时间 @int capital_base:起始资金 @var commission:手续费标准 @var slippage:滑点标准 ''' def __init__(self,start,end,hist,banchmark,universe,capital_base,commission,slippage): self.start = start #开始时间 self.end = end #结束时间 self.hist=hist #历史数据长度 self.banchmark =banchmark #业绩基准 self.universe =universe #股票池 self.capital_base = capital_base #起始资金 self.commission = commission self.slippage = slippage self.account = account() self.today_data= pd.DataFrame() self.date_range = pd.DataFrame() self.hist_range = pd.DataFrame() self.last_hist =[] self.stop_change_stock={} self.day_hist=[] #日期记录 self.value_hist =[] #资产记录 self.return_hist =[] #回报记录 self.return_view =pd.DataFrame() self.botter_list= [] #交易记录 self.__init__ =self.init() def get_cal_date(self):#获取日历 try: res=DA.Api() date_range=res.getTradeCal('XSHG','','','calendarDate,isOpen,prevTradeDate') hist_range=date_range[date_range['calendarDate']<self.start] hist_range=hist_range[hist_range['isOpen']=='1']#isOpen为字符串 hist_range=hist_range.iloc[hist_range.shape[0]-self.hist:hist_range.shape[0]] date_range=date_range[date_range['calendarDate']>=self.start] date_range=date_range[date_range['calendarDate']<=self.end] self.date_range=date_range self.hist_range=hist_range except Exception as e: raise e def init(self): self.get_cal_date()#获取日历 p1 = Process(target=chack_cache,args=(self.date_range,)) #print "start download data!" p2 = Process(target=chack_cache,args=(self.hist_range,)) p1.start() p2.start() self.get_hist_day() def get_hist_day(self): hist_data_list=[] for i in range(self.hist_range.shape[0]): temp=self.read_cache(self.hist_range.iloc[i]['calendarDate'].replace('-','')) hist_data_list.append(temp) self.last_hist=hist_data_list def save_hist(self): self.last_hist.append(self.today_data) while len(self.last_hist)>self.hist: self.last_hist.pop(0) def get_last_data(self): last_data=pd.concat(self.last_hist) return last_data def read_cache(self,name):#读缓存函数 abs_path=os.path.abspath('..') while glob.glob("%s\data\%s.pkl"%(abs_path,name)) == []: time.sleep(6) t1=time.time() while True: try: f1=file('../data/%s.pkl'%name,'rb') p1=pickle.load(f1) f1.close() return p1 break except: time.sleep(6) def get_singal_price(self,aymbol): rows=self.today_data[self.today_data.secID==aymbol] self.tempdata=rows return float(rows.iloc[0]['openPrice']) def run(self): #self.my_thread(self.chack_cache(self.date_range)) #self.my_thread(self.orap()) print "start strategy..." len=self.date_range.shape[0] self.account.cash=self.capital_base self.initialize() for i in range(len): if self.date_range.iloc[i]['isOpen']=='1': self.account.days_counter=i+1 self.account.current_date=self.date_range.iloc[i]['calendarDate'].replace('-','') self.today_data=self.read_cache(self.account.current_date) self.handle_data() self.save_hist()#更新30天数据池 self.day_record() data={'time': self.day_hist, 'return_back':self.return_hist, 'value':self.value_hist } self.return_view=pd.DataFrame(data,index=data['time']) self.return_view['return_back'].plot() show() def day_record(self): self.account.cash value=0 for stock in self.account.secpos: stock_data=self.today_data[self.today_data['secID']==stock] price=float(stock_data.iloc[0]['closePrice']) if price == 0 : if stock in self.stop_change_stock: price=float(self.stop_change_stock[stock])#查询是否属于停牌股票，获得停牌前价格 else: price=float(stock_data.iloc[0]['preclosePrice']) #首次停牌股票，获得停牌前价格 self.stop_change_stock[stock]=price else: if stock in self.stop_change_stock: self.stop_change_stock.pop(stock) else: pass value+=(int(self.account.secpos[stock])*price) self.account.referencePortfolioValue=value+self.account.cash value_return=float(self.account.referencePortfolioValue-self.capital_base)/self.capital_base self.return_hist.append(value_return) self.value_hist.append(self.account.referencePortfolioValue) self.day_hist.append(self.account.current_date) def initialize(self): pass def handle_data(self): pass def order(self,aymbol,amount): price=self.get_singal_price(aymbol) if amount>0 and price>0: #买入 volum=int(amount/100)*100 order_money=(price*volum)*(1+float(self.commission[0])) if order_money<=self.account.cash: self.account.cash-=order_money if aymbol in self.account.secpos: self.account.secpos[aymbol]+=volum else: self.account.secpos[aymbol]=volum self.botter=[aymbol,self.tempdata.iloc[0]['secShortName'],'buy',self.account.current_date,price,volum,order_money] self.botter_list.append(self.botter) return self.botter else : print "botter_error:no enough cash!" return 0 elif amount<0 and price>0: #卖出 volum=int(abs(amount)/100)*100 order_money=(price*volum)*(1-float(self.commission[1])) if aymbol in self.account.secpos and self.account.secpos[aymbol]>=volum : self.account.cash+=order_money self.account.secpos[aymbol]-=volum self.botter=[aymbol,self.tempdata.iloc[0]['secShortName'],'sell',self.account.current_date,price,volum,order_money] self.botter_list.append(self.botter) return self.botter else: print "botter_error:no enough stock!" return 0 else: pass def get_botter_list(self): botter_list=pd.DataFrame(self.botter_list,columns=[u'证券代码',u'证券名称',u'买/卖',u'时间',u'成交价格',u'交易数量',u'交易金额']) return botter_list def order_to(self,symbol,amount): pass def max_buy(self,aymbol,cash): pass def chack_cache(date_range):#检查数据缓存 try: date_range=date_range[date_range['isOpen']=='1'] abs_path=os.path.abspath('..') for i in range(date_range.shape[0]): day=date_range.iloc[i]['calendarDate'].replace('-','') have_file=glob.glob("%s\data\%s.pkl"%(abs_path,day)) if have_file == []: write_cache(day) print "download data: %s done!"%day else: size=os.path.getsize("%s\data\%s.pkl"%(abs_path,day)) if size < 10: write_cache(day) except Exception as e: print e def write_cache(name):#写入数据缓存 f1=file('../data/%s.pkl'%name,'wb') p1=get_all_market_data(name) if p1 is not None: pickle.dump(p1,f1,True) f1.close() def get_all_market_data(current_data): res=DA.Api() data=res.getMktEqud('','',current_data,'','') #print data return data if __name__ == '__main__': pass #data=pd.read_csv('../data/600006.csv') #new_obj=take_back('2006-01-09','2007-01-09',data,cash=100000) #new_obj.handle() #res=DA.Api() #print res.getTradeCal('XSHG','','','calendarDate,isOpen,prevTradeDate') #mystrategy=StrategyBase(start='2012-01-01',end='2014-01-01',banchmark='HS300',universe='A',capital_base=10000000,commission=0.01,slippage=0) #date_range=mystrategy.get_cal_date() #date_range=date_range[date_range['isOpen']=='1'] #for i in range(date_range.shape[0]): #name=date_range.iloc[i]['calendarDate'].replace('-','') #print name #mystrategy.write_cache(name) #print mystrategy.get_all_market_data('20150902')
