金融量化 - 技术分析策略和交易系统_SMA+CCI交易系统

双技术指标：SMA+CCI交易系统

以SMA作为开平仓信号，同时增加CCI作为过滤器；
当股价上穿SMA，同时CCI要小于-100，说明是在超卖的情况下，上穿SMA，做多；交易信号更可信；
当股价下穿SMA，同时CCI要大于+100，说明是在超买的情况下，下穿SMA，做空；交易信号更可信；

import numpy as np
import pandas as pd
import talib as ta
import tushare as ts
import matplotlib as mpl
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
# 确保‘-’号显示正常
mpl.rcParams['axes.unicode_minus']=False
# 确保中文显示正常
mpl.rcParams['font.sans-serif'] = ['SimHei']  

1. 数据准备

# 获取数据
stock_index = ts.get_k_data('hs300', '2016-01-01', '2017-06-30')
stock_index.set_index('date', inplace=True)
stock_index.sort_index(inplace = True)
stock_index.head()

	open	close	high	low	volume	code
date
2016-01-04	3725.86	3470.41	3726.25	3469.01	115370674.0	hs300
2016-01-05	3382.18	3478.78	3518.22	3377.28	162116984.0	hs300
2016-01-06	3482.41	3539.81	3543.74	3468.47	145966144.0	hs300
2016-01-07	3481.15	3294.38	3481.15	3284.74	44102641.0	hs300
2016-01-08	3371.87	3361.56	3418.85	3237.93	185959451.0	hs300

# 计算指标sma，cci
stock_index['sma'] = ta.SMA(np.asarray(stock_index['close']), 5)
stock_index['cci'] = ta.CCI(np.asarray(stock_index['high']), np.asarray(stock_index['low']), np.asarray(stock_index['close']), timeperiod=20)

plt.subplot(2,1,1)
plt.title('沪深300 sma cci指标图')
plt.gca().axes.get_xaxis().set_visible(False)
stock_index['close'].plot(figsize = (10,8))
stock_index['sma'].plot(figsize=(10,8))
plt.legend()
plt.subplot(2,1,2)
stock_index['cci'].plot(figsize = (10,8))
plt.legend()
plt.show()

2. 交易信号、持仓信号和策略逻辑

2.1 交易信号

# 产生开仓信号时应使用昨日及前日数据，以避免未来数据
stock_index['yes_close'] = stock_index['close'].shift(1)
stock_index['yes_sma'] = stock_index['sma'].shift(1)
stock_index['yes_cci'] = stock_index['cci'].shift(1)   #CCI是作为策略的一个过滤器；
stock_index['daybeforeyes_close'] = stock_index['close'].shift(2)
stock_index['daybeforeyes_sma'] = stock_index['sma'].shift(2)

stock_index.head()

# 产生交易信号
# sma开多信号：昨天股价上穿SMA；
stock_index['sma_signal'] = np.where(np.logical_and(stock_index['daybeforeyes_close']<stock_index['daybeforeyes_sma'],
                                                              stock_index['yes_close']>stock_index['yes_sma']), 1, 0)
# sma开空信号：昨天股价下穿SMA
stock_index['sma_signal'] = np.where(np.logical_and(stock_index['daybeforeyes_close']>stock_index['daybeforeyes_sma'],
                                                               stock_index['yes_close']<stock_index['yes_sma']),-1,stock_index['sma_signal'])
# 产生cci做多过滤信号
stock_index['cci_filter'] = np.where(stock_index['yes_cci'] < -100, 1, 0)
# 产生cci做空过滤信号
stock_index['cci_filter']  = np.where(stock_index['yes_cci'] > 100,-1, stock_index['cci_filter'])
# 过滤后的开多信号
stock_index['filtered_signal'] = np.where(stock_index['sma_signal']+stock_index['cci_filter']==2, 1, 0)
# 过滤后的开空信号
stock_index['filtered_signal'] = np.where(stock_index['sma_signal']+stock_index['cci_filter']==-2, -1,stock_index['filtered_signal'])
# 生成交易信号

stock_index.tail()

	open	close	high	low	volume	code	sma	cci	yes_close	yes_sma	yes_cci	daybeforeyes_close	daybeforeyes_sma	sma_signal	cci_filter	filtered_signal
date
2017-06-26	3627.02	3668.09	3671.94	3627.02	134637995.0	hs300	3603.152	190.662244	3622.88	3580.268	127.766972	3590.34	3559.444	0	-1	0
2017-06-27	3665.58	3674.72	3676.53	3648.76	97558702.0	hs300	3628.798	180.056339	3668.09	3603.152	190.662244	3622.88	3580.268	0	-1	0
2017-06-28	3664.16	3646.17	3672.19	3644.03	97920858.0	hs300	3640.440	138.038475	3674.72	3628.798	180.056339	3668.09	3603.152	0	-1	0
2017-06-29	3649.25	3668.83	3669.13	3644.73	85589498.0	hs300	3656.138	130.685000	3646.17	3640.440	138.038475	3674.72	3628.798	0	-1	0
2017-06-30	3654.73	3666.80	3669.76	3646.23	81510028.0	hs300	3664.922	118.314672	3668.83	3656.138	130.685000	3646.17	3640.440	0	-1	0

# 绘制cci，sma指标图
plt.subplot(3,1,1)
plt.title('沪深300 CCI SMA指标图')
plt.gca().axes.get_xaxis().set_visible(False)
stock_index['close'].plot(figsize=(12,8))
stock_index['sma'].plot()
plt.legend(loc='upper left')
plt.subplot(3,1,2)
stock_index['cci'].plot(figsize=(12, 8))
plt.legend(loc='upper left')
plt.subplot(3,1,3)
stock_index['filtered_signal'].plot(figsize=(12, 8), marker='o', linestyle='')
plt.legend(loc='upper left')
plt.show()

2.2 持仓信号

# 记录持仓情况，默认为0
position = 0
# 对每一交易日进行循环
for i, item in stock_index.iterrows():
    # 判断交易信号
    if item['filtered_signal'] == 1:
        # 交易信号为1，则记录仓位为1
        position = 1
    elif item['filtered_signal'] == -1:
        # 交易信号为-1， 则记录仓位为-1
        position = -1
    else:
        pass
    # 记录每日持仓情况
    stock_index.loc[i, 'position'] = position

stock_index.head()

plt.subplot(3, 1, 1)
plt.title('600030 CCI持仓图')
plt.gca().axes.get_xaxis().set_visible(False)
stock_index['close'].plot(figsize = (12,12))
plt.legend()
plt.subplot(3, 1, 2)
stock_index['cci'].plot(figsize = (12,12))
plt.legend()
plt.gca().axes.get_xaxis().set_visible(False)
plt.subplot(3, 1, 3)
stock_index['position'].plot(marker='o', figsize=(12,12),linestyle='')
plt.legend()
plt.show()

3.策略收益和数据可视化

# 计算策略收益
# 计算股票每日收益率
stock_index['pct_change'] = stock_index['close'].pct_change()
# 计算策略每日收益率
stock_index['strategy_return'] = stock_index['pct_change'] * stock_index['position']
# 计算股票累积收益率
stock_index['return'] = (stock_index['pct_change']+1).cumprod()
# 计算策略累积收益率
stock_index['strategy_cum_return'] = (1 + stock_index['strategy_return']).cumprod()

stock_index.head()

	open	close	high	low	volume	code	sma	cci	yes_close	yes_sma	...	daybeforeyes_close	daybeforeyes_sma	sma_signal	cci_filter	filtered_signal	position	pct_change	strategy_return	return	strategy_cum_return
date
2016-01-04	3725.86	3470.41	3726.25	3469.01	115370674.0	hs300	NaN	NaN	NaN	NaN	...	NaN	NaN	0	0	0	0.0	NaN	NaN	NaN	NaN
2016-01-05	3382.18	3478.78	3518.22	3377.28	162116984.0	hs300	NaN	NaN	3470.41	NaN	...	NaN	NaN	0	0	0	0.0	0.002412	0.0	1.002412	1.0
2016-01-06	3482.41	3539.81	3543.74	3468.47	145966144.0	hs300	NaN	NaN	3478.78	NaN	...	3470.41	NaN	0	0	0	0.0	0.017544	0.0	1.019998	1.0
2016-01-07	3481.15	3294.38	3481.15	3284.74	44102641.0	hs300	NaN	NaN	3539.81	NaN	...	3478.78	NaN	0	0	0	0.0	-0.069334	-0.0	0.949277	1.0
2016-01-08	3371.87	3361.56	3418.85	3237.93	185959451.0	hs300	3428.988	NaN	3294.38	NaN	...	3539.81	NaN	0	0	0	0.0	0.020392	0.0	0.968635	1.0

5 rows × 21 columns

# 将股票累积收益率和策略累积收益率绘图
stock_index[['return', 'strategy_cum_return']].plot(figsize = (12,6))

plt.title('600030 CCI收益图')
plt.legend()
plt.show()