import numpy as np
import pandas as pd
import math
from jqdata import *
import matplotlib.pyplot as plt
from datetime import date, timedelta

#上期所

shang = {'AG8888.XSGE':'白银期货指数', 'PB8888.XSGE':'铅期货指数',
'AU8888.XSGE':'黄金期货指数', 'RB8888.XSGE':'螺纹钢期货指数',
'AL8888.XSGE':'铝期货指数',   'RU8888.XSGE':'天然橡胶期货指数',
'BU8888.XSGE':'石油沥青期货指数', 'SN8888.XSGE':'锡期货指数',
'CU8888.XSGE':'铜期货指数', 
'FU8888.XSGE':'燃料油期货指数', 'ZN8888.XSGE':'锌期货指数',
'HC8888.XSGE':'热轧卷板期货指数', 'NI8888.XSGE':'镍期货指数',
'SP8888.XSGE':'纸浆主力合约',}

#郑商所

zheng = {'RM8888.XZCE':'菜籽粕期货指数',
'CF8888.XZCE':'棉花期货指数',  'FG8888.XZCE':'玻璃期货指数',
'SF8888.XZCE':'硅铁期货指数', 
'SM8888.XZCE':'锰硅期货指数',  'MA8888.XZCE':'甲醇期货指数',
'SR8888.XZCE':'白糖期货指数',  
'TA8888.XZCE':'PTA期货指数',   'OI8888.XZCE':'菜籽油期货指数', 
'ZC8888.XZCE':'动力煤期货指数',  
'AP8888.XZCE':'苹果期货指数',  'CJ8888.XZCE':'红枣合约',}

#大商所

da = {'A8888.XDCE':'豆一期货指数', 'JD8888.XDCE':'鸡蛋期货指数',
'B8888.XDCE':'豆二期货指数', 'JM8888.XDCE':'焦煤期货指数',
'L8888.XDCE':'聚乙烯期货指数',
'C8888.XDCE':'玉米期货指数', 'M8888.XDCE':'豆粕期货指数',
'CS8888.XDCE':'玉米淀粉期货指数', 'P8888.XDCE':'棕榈油期货指数',
'PP8888.XDCE':'聚丙烯期货指数',
'I8888.XDCE':'铁矿石期货指数', 'V8888.XDCE':'聚氯乙烯期货指数',
'J8888.XDCE':'焦炭期货指数', 'Y8888.XDCE':'豆油期货指数',
'EG8888.XDCE':'乙二醇期货指数',}

futures = list(shang.keys()) + list(zheng.keys()) + list(da.keys())
futures[:5]

future_list = []
date_ = date(2014,1,1)
for future in futures:
    start_date = get_security_info(future).start_date
    if start_date < (date_ - timedelta(days=365)):
        future_list.append(future)
print(len(future_list))
future_list[:5]

n=5
trade_days = get_trade_days(start_date='2018-01-01', end_date='2019-06-01')
panel_dict = {}

for i in range(1, len(trade_days)):
    daily_start = str(trade_days[i - 1])+' 21:31:00'
    daily_end = str(trade_days[i])+' 15:05:00'
    
    factor_df_index = []
    factor_df_data = []

    for future in future_list:
        price = get_price(future,start_date=daily_start,end_date=daily_end,frequency='5m',fields=['close'],
                          fq='pre')
        sum_rt2 = 0.0
        sum_rt3 = 0.0
        sum_rt4 = 0.0
        for j in range(1, len(price)):
            pi = math.log(price.iloc[j]['close'])
            pi_1 = math.log(price.iloc[j - 1]['close'])
            rt = pi - pi_1
            sum_rt2 += math.pow(rt, 2)
            sum_rt3 += math.pow(rt, 3)
            sum_rt4 += math.pow(rt, 4)
            
        rd_var = sum_rt2
        if sum_rt3 == 0:
            rd_skew = 0
        else:
            rd_skew = math.sqrt(len(price)) * sum_rt3 / (math.pow(rd_var, 3 / 2))
        
        if sum_rt4 == 0:
            rd_kurt = 0
        else:
            rd_kurt = len(price) * sum_rt4 / (math.pow(rd_var, 2))
            
        factor_df_index.append(future)
        factor_df_data.append([price.close.iloc[-1], rd_var, rd_skew, rd_kurt])
        factor_df = pd.DataFrame(data=factor_df_data, index=factor_df_index, 
                                 columns=['close', 'rd_var', 'rd_skew', 'rd_kurt'])
    panel_dict[trade_days[i]] = factor_df
panel = pd.Panel(panel_dict)
panel

panel.major_xs('AG8888.XSGE')

rvol = np.sqrt(panel.minor_xs('rd_var').T.rolling(5).mean()*242).shift(1)
rvol.head(10)

rskew = panel.minor_xs('rd_skew').T.rolling(5).mean().shift(1)
rskew.tail()

rkurt = panel.minor_xs('rd_kurt').T.rolling(5).mean().shift(1)
rkurt.tail()

import matplotlib.dates as mdate
# 设置字体 用来正常显示中文标签
plt.rcParams['font.sans-serif'] = ['Arial Unicode MS']

# 用来正常显示负号
plt.rcParams['axes.unicode_minus'] = False

"""
绘制直方图
data:必选参数，绘图数据
bins:直方图的长条形数目，可选项，默认为10
density:是否将得到的直方图向量归一化，可选项，默认为0，代表不归一化，显示频数。normed=1，表示归一化，显示频率。
facecolor:长条形的颜色
edgecolor:长条形边框的颜色
alpha:透明度
"""

# 这个颜色是我把研报的图截图用取色器取出来的，为什么画出来还是有色差？
color = "#1F77B4"
plt.hist(rvol, bins=40, density=0, facecolor=color, edgecolor=None, alpha=1)
# 显示横轴标签
plt.xlabel("区间")
# 显示纵轴标签
plt.ylabel("频数")
# 显示图标题
plt.title("个股波动率分布")
plt.show()

plt.hist(rskew, bins=40, density=0, facecolor=color, edgecolor=None, alpha=1)
# 显示横轴标签
plt.xlabel("区间")
# 显示纵轴标签
plt.ylabel("频数")
# 显示图标题
plt.title("个股偏度分布")
plt.show()

plt.hist(rkurt, bins=40, density=0, facecolor=color, edgecolor=None, alpha=1)
# 显示横轴标签
plt.xlabel("区间")
# 显示纵轴标签
plt.ylabel("频数")
# 显示图标题
plt.title("个股峰度分布")
plt.show()

# 百分位走势5档颜色 蓝 橙 绿 红 紫
color_list = ['#5698c6', '#ff9e4a', '#60b760', '#e05c5d', '#ae8ccd']

label_list = ['10', '25', 'median', '75', '90']

all_df = [rvol, rskew, rkurt]
title_name = ['rvol', 'rskew', 'rkurt']

num = 0
for df in all_df:
    # 这里需要用每天的因子数据分档，计算出5个折线
    y_list = [[], [], [], [], []]
    q_list = [0.10, 0.25, 0.50, 0.75, 0.90]
    
    for i in range(len(df)):
        factor = df.iloc[i].sort_values()
        for j in range(len(q_list)):
            num_signal = int(len(q_list) * q_list[j])
            factor_value = factor.iloc[num_signal]
            y_list[j].append(factor_value)
    
    # 可以设置生成图片的大小
    fig = plt.figure(figsize=(12, 8))
    plt.title(title_name[num]+' 百分位走势')
    num += 1
    for i in range(len(y_list)):
        plt.plot(y_list[i], color_list[i], label=label_list[i])
        
        x = np.arange(0, len(df),50)
        x_label = []
        for i in range(0,len(df)):
            if i in x:
                date = list(df.index)[i]
                x_label.append(date)
        plt.xticks(x, x_label, rotation='vertical')
        
        plt.xticks(rotation=360)
        plt.xlabel("TRADE_DT")
        plt.ylabel("因子值")
        legend()

all_df = [rvol, rskew, rkurt]
Q_list = ['Q1', 'Q2', 'Q3', 'Q4', 'Q5']
df_list = []
num = 0
for i in range(len(all_df)):
    price_close = panel.minor_xs('close').T
    price_pct = price_close.pct_change().dropna().iloc[4:]
    groups = [[], [], [], [], []]
    length = int(len(all_df[i].T) / 5)
    for j in range(len(price_pct)):
        df = all_df[i].dropna().iloc[0:]
        index_list = list(df.iloc[j].sort_values().index)
        daily_price = price_pct[index_list].iloc[j]
        groups[0].append(daily_price[0:5].mean())
        groups[1].append(daily_price[5:9].mean())
        groups[2].append(daily_price[9:14].mean())
        groups[3].append(daily_price[14:19].mean())
        groups[4].append(daily_price[19:].mean())
    df_group = pd.DataFrame(groups).T
    df_group.index = price_pct.index
    df_group = df_group.cumsum()
    df_group.columns = ['Q1', 'Q2', 'Q3', 'Q4', 'Q5']
    df_list.append(df_group)
    
    fig = plt.figure(figsize=(12, 8))
    plt.title(title_name[num]+' 累计收益率')
    
    plt.plot(df_group['Q1'])
    plt.plot(df_group['Q2'])
    plt.plot(df_group['Q3'])
    plt.plot(df_group['Q4'])
    plt.plot(df_group['Q5'])
    plt.legend()
    num += 1

num = 0
for df in df_list:
    df = df['Q3'] - df['Q2']
    
    plt.figure(figsize=(12, 8))
    plt.title(title_name[num]+' 累计收益率')
    plt.plot(df, label=title_name[num])
    plt.legend()
    num += 1

import scipy.stats as st
all_df = [rvol, rskew, rkurt]
name = ['rvol', 'rskew', 'rkurt']
color_list = ['#2B4C80', '#B00004']
label_list = ['IC', 'IC均值(12期)']

for i in range(len(all_df)):
    # 每天的ic
    ic_list = []
    # ic均值(12期)
    ic_ma_list = []
    y_list = [ic_list, ic_ma_list]
    for j in range(len(all_df[i].iloc[5:])):
        ic = st.pearsonr(price_pct.iloc[j].values, all_df[i].iloc[5:].iloc[j].values)[0]
        ic_list.append(ic)
    ic_list = np.array(ic_list)
    print("%s ic 小于0的个数占比：%s" % (name[i], np.sum(ic_list < 0) / len(ic_list)))
    for z in range(len(ic_list)):
        if z < 12:
            ic_ma_list.append(np.nan)
            continue
        ic_ma = np.array(ic_list[z - 12:z]).mean()
        ic_ma_list.append(ic_ma)
        

    fig = plt.figure(figsize=(12, 8))
    ax = fig.add_subplot(1, 1, 1)
    ax.set_title(name[i] + "因子ic")

    for i in range(len(y_list)):
        yi = y_list[i]
        ax.plot(yi, color_list[i], label=label_list[i])

    # 绘制Y轴的网格线便于查看IC
    plt.grid(axis='y')
    plt.show()