阅读原文:http://club.jr.jd.com/quant/topic/1425690
京东金融官方资讯QQ群:456448095 有什么想咨询的都可以来询问我们
相当于加了一个开关,站在40日均线上,就执行羊驼策略,否则,就空仓。
源码:
import random
import numpy as np
import pandas as pd
from pandas import Series,DataFrame
import scipy.stats as stats
import math
# 设置股票池,本程序中为所有沪深300的股票
stocks = get_index_stocks('000300.XSHG')
#求出股票池中有多少股票
num=len(stocks)
set_universe(stocks)
#设置benchmark,默认为沪深300
#set_benchmark('510050.XSHG')
#设置回测条件
set_commission(PerTrade(buy_cost=0.0008, sell_cost=0.0015, min_cost=5))
set_slippage(FixedSlippage(0))
#设置初始买入多少只股票
num_of_stocks=10
#设置每次更新时替换多少只股票
num_of_change=3
#设置计算几日收益率
period=2
#用一个列表来保存每天持有的股票代码
stockshold=[]
#判断参数输入是否符合条件,如果不符合,则重置为默认值
if num_of_stocks>num:
log.info('too large num_of_stocks')
num_of_stocks=10
elif num_of_change>num_of_stocks:
log.info('too large num_of_change')
num_of_change=1
num_MA=40
security = '000300.XSHG'
#预处理数据,将没有数据的股票剔除,同时加入收益率
#构成一个列索引为股票名,收益率一行的索引为
#'return'的dataframe,并返回这个dataframe
def process():
#取出每只股票period天的收盘价格
stocks_info=history(period,'1d','close')
#去除信息不全的数据
stocks_info.dropna(axis=0,how='any',thresh=None)
#取出昨天和period天之前的收盘价,计算收益率
a1=list(stocks_info.iloc[0])
a2=list(stocks_info.iloc[period-1])
a1=np.array(a1)
a2=np.array(a2)
#用一个dataframe来保存所有股票的收益率信息
stocks_return=DataFrame(a2/a1,columns=['return'],index=stocks_info.columns)
stocks_info=stocks_info.T
#把收益率的数据加到相应的列
stocks_info=pd.concat([stocks_info,stocks_return],axis=1)
#将股票信息按照收益率从大到小来存储
stocks_info=stocks_info.sort(columns=['return'],ascending=[False])
#返回处理好的dataframe
return stocks_info#股票入池
def BuyStocks(stocks_info,cash):
#计算现在持有的股票数
current_num=len(stockshold)
stocks_info=stocks_info.T
#将已持有的股票从股票池中剔除
for i in range(0,current_num):
if stockshold[i] in stocks_info.columns:
del stocks_info[stockshold[i]]
stocks_info=stocks_info.T
#计算在每只股票上可以支付的现金
cash=cash/num_of_stocks-current_num
for i in range(0,num_of_stocks-current_num):
#取得股票当前的价格
current_price=stocks_info['current_price'][i]
#判断是否有价格数据
if math.isnan(current_price)==False:
#计算可以每只股票可以购买的数量
num_of_shares=int(cash/current_price)
if num_of_shares>0:
order(stocks_info.index[i],+num_of_shares)
log.info('buying %s' %(stocks_info.index[i]))
#将购买的股票代码加到stockhold中
stockshold.append(stocks_info.index[i])
#股票出池
def SellStocks(stocks_info,num_of_change):
stocks_hold=DataFrame()
current_num=len(stockshold)
'log.info('here' % (current_num))'
#用一个dataframe来保存持有的股票的信息
for i in range(0,current_num):
stocks_hold=pd.concat([stocks_hold,stocks_info.loc[stockshold[i]]],axis=1)
stocks_hold=stocks_hold.T
#在持有的股票数不为0时,将持有的股票信息按照收益率大小从小到大排序
if current_num>0:
stocks_hold=stocks_hold.sort(columns=['return'])
#把收益率最低的股票卖空
for k in range(0,min(num_of_change,current_num)):
log.info(num_of_change)
#判断是否停牌
if stocks_hold['paused'][k]=='False':
order_target(stocks_hold.index[k],0)
log.info('Selling %s' % (stocks_hold.index[k]))
#在stockshold中去除已经卖空的股票的信息
stockshold.remove(stocks_hold.index[k])
# 每个单位时间(如果按天回测,则每天调用一次,如果按分钟,则每分钟调用一次)调用一次
def handle_data(context, data):
stocks_info=process()
stocks_num=len(stocks_info.index)
#用一个列表来保存所有股票是否停牌的信息
pause=[]
for i in range(0,stocks_num):
if data[stocks_info.index[i]].paused==True:
pause.append('True')
else:
pause.append('False')
#将列表转换成dataframe以便加入到stocks_info中
paused=DataFrame(pause,columns=['paused'],index=stocks_info.index)
stocks_info=pd.concat([stocks_info,paused],axis=1)
#用一个列表来保存所有股票当前的价格信息
currentprice=[]
for i in range(0,stocks_num):
currentprice.append(data[stocks_info.index[i]].price)
current_price=DataFrame(currentprice,columns=['current_price'],index=stocks_info.index)
#将股票是否停牌,当前价格的信息添加到stocks_info中
stocks_info=pd.concat([stocks_info,current_price],axis=1)
#取得当前现金
cash=context.portfolio.cash
#判断MA的条件
close_price =history(num_MA,'1d', 'close', ['000300.XSHG'])
Closemean=close_price[security].mean()
# 取得当前价格
Current_close = close_price.iloc[-1,-1]
log.info('close %s' % (Current_close))
log.info('mean %s' % Closemean)
if Current_close > Closemean:
#执行卖出股票的函数
SellStocks(stocks_info,3)
#执行买入股票的函数
BuyStocks(stocks_info,cash)
elif Current_close <>
SellStocks(stocks_info,10)