我们的基础策略教程教授配对交易策略策略。接下来我们将为大家展示相应策略的策略逻辑，策略信号以及策略表现等。
配对交易策略（pair trading）
1. 数据获取：选取金种子酒（600199.SH）和舍得酒业（600702.SH）作为股票标的进行研究
In [18]:
stocks_pair = ['600199.SH', '600702.SH']
data4=get_price(stocks_pair, '20130103', '20170103', '1d', ['close'],is_panel=1)
data4=data4['close']
data4.plot(figsize=(10,8))
Out[18]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f35d0175080>
2. 策略逻辑
In [19]:
data4.head()
Out[19]:
| 600199.SH | 600702.SH | |
|---|---|---|
| 2013-01-04 | 18.80 | 28.25 |
| 2013-01-07 | 18.66 | 27.91 |
| 2013-01-08 | 19.44 | 29.33 |
| 2013-01-09 | 20.00 | 30.20 |
| 2013-01-10 | 19.79 | 30.09 |
2.1 取价差，并对价差进行标准化
In [20]:
data4['pricedelta']=data4['600199.SH']-data4['600702.SH']
data4['after_zscore']=(data4['pricedelta']-data4['pricedelta'].mean())/np.std(data4['pricedelta'])
In [21]:
data4.head()
Out[21]:
| 600199.SH | 600702.SH | pricedelta | after_zscore | |
|---|---|---|---|---|
| 2013-01-04 | 18.80 | 28.25 | -9.45 | -0.492517 |
| 2013-01-07 | 18.66 | 27.91 | -9.25 | -0.442815 |
| 2013-01-08 | 19.44 | 29.33 | -9.89 | -0.601860 |
| 2013-01-09 | 20.00 | 30.20 | -10.20 | -0.678898 |
| 2013-01-10 | 19.79 | 30.09 | -10.30 | -0.703748 |
In [23]:
threshold=1.5
len(data4[data4['after_zscore'] < - threshold])
Out[23]:
84
2.2 策略逻辑:
当标准化后的价差大于阀门，则600199做空
当标准化后的价差小于负的阀门，则600199做多
当标准化后的价差在半个阀门范围内则平仓
In [27]:
data4['position_600199'] = np.where(data4['after_zscore'] > threshold, -1, np.nan)
data4['position_600199'] = np.where(data4['after_zscore'] < -threshold, 1, data4['position_600199'])
data4['position_600199'] = np.where(abs(data4['after_zscore']) < threshold/2, 0, data4['position_600199'])
data4['position_600199'] = data4['position_600199'].fillna(method = 'ffill')
In [29]:
data4.tail()
Out[29]:
| 600199.SH | 600702.SH | pricedelta | after_zscore | position_600199 | |
|---|---|---|---|---|---|
| 2016-12-27 | 9.06 | 22.40 | -13.34 | -1.459213 | 1.0 |
| 2016-12-28 | 8.98 | 22.34 | -13.36 | -1.464183 | 1.0 |
| 2016-12-29 | 9.00 | 22.40 | -13.40 | -1.474124 | 1.0 |
| 2016-12-30 | 9.18 | 22.60 | -13.42 | -1.479094 | 1.0 |
| 2017-01-03 | 9.21 | 22.57 | -13.36 | -1.464183 | 1.0 |
2.3 其中600702的信号与600199的信号相反
In [31]:
data4['position_600702']=-data4['position_600199']
data4.tail()
Out[31]:
| 600199.SH | 600702.SH | pricedelta | after_zscore | position_600199 | position_600702 | |
|---|---|---|---|---|---|---|
| 2016-12-27 | 9.06 | 22.40 | -13.34 | -1.459213 | 1.0 | -1.0 |
| 2016-12-28 | 8.98 | 22.34 | -13.36 | -1.464183 | 1.0 | -1.0 |
| 2016-12-29 | 9.00 | 22.40 | -13.40 | -1.474124 | 1.0 | -1.0 |
| 2016-12-30 | 9.18 | 22.60 | -13.42 | -1.479094 | 1.0 | -1.0 |
| 2017-01-03 | 9.21 | 22.57 | -13.36 | -1.464183 | 1.0 | -1.0 |
3.策略信号
In [34]:
data4[['position_600702','position_600199']].plot(ylim=[-1.1, 1.1], figsize=(10, 6))
Out[34]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f35d031c3c8>
4.策略展示
data4['return_600199']=np.log(data4['600199.SH']/data4['600199.SH'].shift(1)) data4['return_600702']=np.log(data4['600702.SH']/data4['600702.SH'].shift(1)) data4.head()
In [37]:
data4['strategy']=0.5*(data4['position_600199'].shift(1)*data4['return_600199'])+\
0.5*(data4['position_600702'].shift(1)*data4['return_600702'])
In [42]:
data4[['return_600199','return_600702','strategy']].cumsum().apply(np.exp).plot(figsize=(12, 8))
Out[42]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f35ca819518>
查看以上策略详情请到supermind量化交易官网查看：同花顺Supermind量化交易 经典量化策略基础-配对交易策略
