2019-04-17

NSLog(@"hello world");        

# -*- coding: utf-8 -*-

import numpy as np

import pandas as pd

from matplotlib import pylab as plt

import math

group1 = ['300460', '002619', '603809','000426','002600','000008','002217','300652','600505','002078']

group2 =['603533','603928','600693','603811','300613','300621','000988','600839','300502','000933','600685','002907','002922',

        '600284','601998','000505','600726','000591','002798','601127']

group3 = ['300697', '300035', '603322','601229','002681','002121','600615','600367','002695','002301','300130','603920','300325',

          '300468','002061','002897','002014','000628','300034','002023','002199','000882','000401','002598','300558','600691',

          '300392','601901','300353','002077']

# IF_IH_IC = ['IF','IH','IC']  # 三大期货指数

# IF_IH = ['IF','IH']

# IF_IC = ['IF','IC']

# IH_IC = ['IH','IC']

IF = 'IF'

IH = 'IH'

IC = 'IC'

# 定义函数：得到投资组合x从时间start到时间end的日收益率序列（x：group  start:20190101, end:20190401）

def get_returndaily(x,start,end):

#:step1：先用交易日日历得到从时间start到时间end的交易日日期

    from CAL.PyCAL import *

    data=DataAPI.TradeCalGet(exchangeCD=u"XSHG",beginDate=str(start),endDate=str(end),field=['calendarDate','isOpen'])                   

    # DataAPI.TradeCalGet即交易所交易日历函数

    data = data[data['isOpen'] == 1]

    date= map(lambda x: x[0:4]+x[5:7]+x[8:10], data['calendarDate'].values.tolist())

    # 从时间start到时间end的每个交易日日期，得到后存为列表date

# step2：调用投资组合x每只股票每个交易日的市值以及收盘价

    returndaily=np.zeros(len(date)-1)

    for i in range(len(date)-1):    # 即 i=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11...]为了从date列表里面依次定位取第几个日子

        inf1=DataAPI.MktEqudAdjGet(tradeDate=date[i],ticker=x,field=u"ticker,closePrice").set_index('ticker')  # 前一个交易日的收盘价

        inf2=DataAPI.MktEqudAdjGet(tradeDate=date[i+1],ticker=x,field=u"ticker,marketValue,closePrice").set_index('ticker')  # 当前交易日的收盘价和市值

        # 将以上两个表横向合并成一张有股票代码、市值、上个交易日收盘价和当前交易日收盘价组成的表Return

        Return=pd.concat([inf2,inf1],axis=1)

        # print date[i+1]

        # print Return

    # step3：计算每只股票收益率和市值加权的权重以及两者的乘积

        # 每只股票权重不等

        """

        Return.columns=['Weight','Return','WReturn']  # 改下表的列名称（就是第一行的字段），下面分别计算各列数据

        Return['Weight']=Return['Weight']/Return['Weight'].sum() # 按市值在该组总市值占比分配权重

        Return['Return']=Return['Return']/Return['WReturn']-1    # 即定义出三者逻辑运算关系

        Return['WReturn']=Return['Weight']*Return['Return']      # 同上，这个WReturn即我们需要的个股日收益率表

        returndaily[i]=Return['WReturn'].sum()                  # 按日依次汇总得出组合的日收益率表returndaily

        #    Return['Return']=Return['Return']/Return['Return']-1    # 即定义出三者逻辑运算关系       

        """

        # 每只股票等权重

        Return.columns=['Weight','Return','WReturn']

        Return['Weight'] = 1.0

        Return['Return']=Return['Return']/Return['WReturn']-1    # 即定义出三者逻辑运算关系

        Return['WReturn']=Return['Weight']*Return['Return']      # 同上，这个WReturn即我们需要的个股日收益率表

        returndaily[i]=Return['WReturn'].sum()       

    return returndaily

# 定义函数：得到期货指数y从时间start到时间end的日收益率序列（y：IF,IH,IC  start:20190101, end:20190401）

def get_returnfutures_daily(y,start,end):

# step1：先用交易日日历得到从时间start到时间end的交易日日期

    from CAL.PyCAL import *

    data=DataAPI.TradeCalGet(exchangeCD=u"XSHG",beginDate=str(start),endDate=str(end),field=['calendarDate','isOpen'])                   

    data = data[data['isOpen'] == 1]

    date= map(lambda x: x[0:4]+x[5:7]+x[8:10], data['calendarDate'].values.tolist())

    # step2：调用指数y收盘价用来计算收益率

    returndaily=np.zeros(len(date)-1)

    for i in range(len(date)-1):  # 即 i=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11...]为了从date列表里面依次定位取第几个日子

        # inf1= DataAPI.MktFutdGet(tradeDate=date[i],ticker=y,field=u"ticker,closePrice",pandas="1").set_index('ticker')   

        # inf2= DataAPI.MktFutdGet(tradeDate=date[i+1],ticker=y,field=u"ticker,openInt,closePrice",pandas="1").set_index('ticker')  # 这两个返回空值，有时间再研究

        inf1 = DataAPI.MktMFutdGet(tradeDate=date[i],mainCon=u"1",contractMark=u"",contractObject=y,field=u"ticker,closePrice",pandas="1").set_index('ticker')

        inf2 = DataAPI.MktMFutdGet(tradeDate=date[i+1],mainCon=u"1",contractMark=u"",contractObject=y,field=u"ticker,openInt,closePrice",pandas="1").set_index('ticker')

        Return=pd.concat([inf2,inf1],axis=1)

        # step3：计算每种期货指数收益率和持仓量加权的权重以及两者的乘积

        # 如果对冲指数包含多个期货指数

        if type(y) is list: 

            Return.columns=['Weight','Return','WReturn'] # 改下表的列名称（就是第一行的字段），下面分别计算各列数据

            Return['Weight']=Return['Weight']/Return['Weight'].sum() # 按市值在该组总市值占比分配权重

            Return['Return']=Return['Return']/Return['WReturn']-1    # 即定义出三者逻辑运算关系

            Return['WReturn']=Return['Weight']*Return['Return']      # 同上，这个WReturn即我们需要的个股日收益率表

            returndaily[i]=Return['WReturn'].sum()                  # 按日依次汇总得出组合的日收益率表returndaily

        # 如果只用一种期货指数对冲

        else:

            Return.columns = ['Weight','Return','WReturn']

            # Return['Weight'] = 1.0/(len(date)-1)  # 等权

            Return['Return'] = Return['Return']/Return['WReturn']-1

            # Return['WReturn'] = Return['Weight']*Return['Return']

            returndaily[i] = Return['Return'].sum()

    return returndaily   

# 定义函数：填充ndarray中的nan值

def fill_ndarray(t1):

    nan_num = np.count_nonzero(t1==t1)

    if nan_num != 0: # 不为0,说明当前这一列中有nan

        temp_not_nan=t1[t1==t1] # 当前不为nan的array

        # 选中当前为nan的位置,把值赋值为该列除nan以外的数的均值

        t1[np.isnan(t1)] = temp_not_nan.mean()

    return t1

# 股票组合收益率       

group1_ret = get_returndaily(group1,20190101,20190401)

group2_ret = get_returndaily(group2,20190101,20190401)

group3_ret = get_returndaily(group3,20190101,20190401)

print group3_ret

print "*"*100

# 各个指数收益率

# IF_IH_IC_ret = get_returnfutures_daily(IF_IH_IC,20190101,20190401)  # 三大指数

# IF_IH_ret = get_returnfutures_daily(IF_IH,20190101,20190401)

# IF_IC_ret = get_returnfutures_daily(IF_IC,20190101,20190401)

# IH_IC_ret = get_returnfutures_daily(IH_IC,20190101,20190401)

IF_ret = get_returnfutures_daily(IF,20190101,20190401)

IF_ret = fill_ndarray(IF_ret)

IH_ret = get_returnfutures_daily(IH,20190101,20190401)

IH_ret = fill_ndarray(IH_ret)

IC_ret = get_returnfutures_daily(IC,20190101,20190401)

IC_ret = fill_ndarray(IC_ret)

print IC_ret

print type(IC_ret)

# 定义相关系数函数

def corr(stock_list):    # stock_list:股票组合收益率

    # 期货收益率序列

    # IF_IH_IC_series = pd.Series(IF_IH_IC_ret)

    # IF_IH_series = pd.Series(IF_IH_ret)

    # IF_IC_series = pd.Series(IF_IC_ret)

    # IH_IC_series = pd.Series(IH_IC_ret)

    IF_ret_series = pd.Series(IF_ret)

    IH_ret_series = pd.Series(IH_ret)

    IC_ret_series = pd.Series(IC_ret)

    # 股票收益率序列

    stock_group = pd.Series(stock_list)

    # 计算相关系数，保留四位小数

    # corr_IF_IH_IC = round(stock_group.corr(IF_IH_IC_series), 4)

    # corr_IF_IH = round(stock_group.corr(IF_IH_series), 4)

    # corr_IF_IC = round(stock_group.corr(IF_IC_series), 4)

    # corr_IH_IC = round(stock_group.corr(IH_IC_series), 4)

    corr_IF = round(stock_group.corr(IF_ret_series), 4)

    corr_IH = round(stock_group.corr(IH_ret_series), 4)

    corr_IC = round(stock_group.corr(IC_ret_series), 4)

    # 绘制散点图

#    plt.scatter(stock_list, IF_IH_IC_ret)

#    plt.title('corr_IF_IH_IC :' + str(corr_IF_IH_IC), fontproperties='SimHei')

#    print('corr_IF_IH_IC :', corr_IF_IH_IC)

#    plt.show()

#    plt.scatter(stock_list, IF_IH_ret)

#    plt.title('corr_IF_IH :' + str(corr_IF_IH), fontproperties='SimHei')

#    print('corr_IF_IH :', corr_IF_IH)

#    plt.show()

#    plt.scatter(stock_list, IF_IC_ret)

#    plt.title('corr_IF_IC :' + str(corr_IF_IC), fontproperties='SimHei')

#    print('corr_IF_IC :', corr_IF_IC)

#    plt.show()

#    plt.scatter(stock_list, IH_IC_ret)

#    plt.title('corr_IH_IC :' + str(corr_IH_IC), fontproperties='SimHei')

#    print('corr_IH_IC :', corr_IH_IC)

#    plt.show()

    plt.scatter(stock_list, IF_ret)

    plt.title('corr_IF :' + str(corr_IF), fontproperties='SimHei')

    print('corr_IF :', corr_IF)

    plt.show()

    plt.scatter(stock_list, IH_ret)

    plt.title('corr_IH :' + str(corr_IH), fontproperties='SimHei')

    print('corr_IH :', corr_IH)

    plt.show()

    plt.scatter(stock_list, IC_ret)

    plt.title('corr_IC :' + str(corr_IC), fontproperties='SimHei')

    print('corr_IC :', corr_IC)

    plt.show()   

    return corr_IF,corr_IH,corr_IC

group1_corr = corr(group1_ret)  # 组合的各种相关系数

group2_corr = corr(group2_ret)

group3_corr = corr(group3_ret)

print group1_corr

print "*"*100

print group2_corr

print "*"*100

print group3_corr

# 结论：有结果可知，三组股票均与IC的相关性最高。当然了，在与指数池做相关性分析时，发现：第二组用IH\IC混合对冲较好，由于相对复杂，时间关系，这里不深入研究，日后再探讨。

# 下面是求贝塔的过程

cov_mat = np.cov(group3_ret,IC_ret)

print cov_mat

# 协方差

cov = cov_mat[1,1]

print 'cov(group3_ret,IC_ret) =', cov

print "var =", IC_ret.var()  # 方差

# 得出贝塔

B=cov/IC_ret.var()

print 'β =', B

# 期指总市值

def matketValue(y, date):

    inf = DataAPI.MktMFutdGet(tradeDate=date,mainCon=u"1",contractMark=u"",contractObject=y,field=u"ticker,closePrice",pandas="1").set_index('ticker')

    closeprice = inf.values

    tol_index_equd = closeprice * 300

    return tol_index_equd

# 股票组合总市值(废弃)

# def group_matketValue(group, date):

#    group_equd = 0

#    for stock in group:

#        equd=DataAPI.MktEqudAdjGet(tradeDate=date,ticker=stock,field=u"ticker,marketValue").set_index('ticker')

#        equd = equd["marketValue"].values

#        # print equd

#        group_equd += equd

#    return group_equd

Vs = len(group3)*1000000

Vf = matketValue(IC, '20190401')

print 'Vs=', Vs  # 股票组合市值

print 'Vf=', Vf  # 一手期指市值

N = B*Vs/Vf  # 期指对冲的手数

print 'N=', N