算法因子

1. 读取本地数据
2. 设计与添加算法
3. 可视化与获取因子信号数据
4. 外部因子编写

1_读取本地数据

import numpy as np
from jaqs.data import DataView
from jaqs.research import SignalDigger
import warnings
warnings.filterwarnings("ignore")
dataview_folder = 'JAQS_Data/hs300'
dv = DataView()
dv.load_dataview(dataview_folder)
mask = dv.get_ts('mask_index_member')
can_enter = dv.get_ts('can_enter')
can_exit = dv.get_ts('can_exit')

D:\Anaconda3\lib\site-packages\statsmodels\compat\pandas.py:56: FutureWarning: The pandas.core.datetools module is deprecated and will be removed in a future version. Please use the pandas.tseries module instead.
  from pandas.core import datetools


Dataview loaded successfully.

2_设计与添加算法

StdDev(x,n)
Delay(x,n)
Delta(x,n)
Return(x,n,log)
Ts_Skewness(x,n)
Product(x,n)
Ts_Mean(x,n)

1. 过去五天的平均成交量大于五天前的过去五天平均成交量变化率

Ts_Mean(volume,5)/Delay(Ts_Mean(volume,5),5)

2. 最低价的累积变化率上升

Product(low/Delay(low,1),5) - 1

3. 价格波动率上升

StdDev(Return(close,1,True), 10)

4. 收益率偏度

Ts_Skewness(Return(low,5),5)

5. 价量协方差

Covariance(low, volume, 10)

dv.remove_field('Cov')

Field name [Cov] does not exist.

def add_factor():
    dv.add_formula('Volume_Up','Ts_Mean(volume,5)/Delay(Ts_Mean(volume,5),5)*-1*Return(close,5)',is_quarterly=False)    
    dv.add_formula('Low_Product','-1*Product(low/Delay(low,1),5) - 1',is_quarterly=False)
    dv.add_formula('STD', '-1*StdDev(Return(close,1), 10)', is_quarterly=False)
    dv.add_formula('Skewness', 'Ts_Skewness(Return(volume,5),5)', is_quarterly=False)
    dv.add_formula('Cov','-1*Covariance(low,volume,10)',is_quarterly=False)
add_factor()

3_可视化与获取因子信号数据

import matplotlib.pyplot as plt
def cal_obj(signal, name, period, quantile):
    price = dv.get_ts('close_adj')
    price_bench = dv.data_benchmark
    obj = SignalDigger(output_folder="hs300/%s" % name,
                       output_format='pdf')
    obj.process_signal_before_analysis(signal,
                                   price=price,
                                   n_quantiles=quantile, period=period,
                                   benchmark_price=price_bench,
                                   can_enter = can_enter,
                                   can_exit = can_exit,
                                   mask=mask
                                   )
    obj.create_full_report()
    return obj
def plot_pfm(signal, name, period=5, quantile=5):
    obj = cal_obj(signal, name, period, quantile)
    plt.show()
def signal_data(signal, name, period=5, quantile=5):
    obj = cal_obj(signal, name, period, quantile)
    return obj.signal_data

signal_dict = {'sig_1' : dv.get_ts('Volume_Up'),
               'sig_2' : dv.get_ts('Low_Product'),
               'sig_3' : dv.get_ts('STD'),
               'sig_4' : dv.get_ts('Skewness'),
               'sig_5' : dv.get_ts('Cov')}

data_signal = {key: signal_data(frame, key) for key, frame in signal_dict.items()}

Nan Data Count (should be zero) : 0;  Percentage of effective data: 53%


Value of signals of Different Quantiles Statistics
                   min           max          mean           std  count  \
quantile                                                                  
1        -4.448244e+16  1.895836e-01 -6.499444e+12  3.282543e+14  82946   
2        -3.929699e-01  2.325106e-01 -2.226987e-02  4.620945e-02  82360   
3        -2.352434e-01  2.659026e-01 -1.242107e-03  3.827854e-02  82350   
4        -1.652057e-01  3.359749e-01  1.508382e-02  3.799451e-02  82360   
5        -9.601820e-02  3.760106e+16  3.462099e+12  2.003023e+14  81770   

            count %  
quantile             
1         20.142987  
2         20.000680  
3         19.998252  
4         20.000680  
5         19.857402  
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_1\returns_report.pdf
Information Analysis
                ic
IC Mean      0.044
IC Std.      0.174
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.254
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_1\information_report.pdf
Nan Data Count (should be zero) : 0;  Percentage of effective data: 53%


Value of signals of Different Quantiles Statistics
               min       max      mean       std  count    count %
quantile                                                          
1        -2.668837 -1.794997 -2.060149  0.057623  83290  20.143464
2        -2.259160 -1.770604 -2.015925  0.037586  82694  19.999323
3        -2.182598 -1.746186 -1.998437  0.037362  82687  19.997630
4        -2.134259 -1.699563 -1.982646  0.039398  82694  19.999323
5        -2.076210 -1.265291 -1.947343  0.065847  82119  19.860261
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_2\returns_report.pdf
Information Analysis
                ic
IC Mean      0.040
IC Std.      0.178
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.223
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_2\information_report.pdf
Nan Data Count (should be zero) : 0;  Percentage of effective data: 53%


Value of signals of Different Quantiles Statistics
               min       max      mean       std  count    count %
quantile                                                          
1        -0.230151 -0.013882 -0.038593  0.021056  83288  20.143711
2        -0.091371 -0.010489 -0.025736  0.010736  82691  19.999323
3        -0.085854 -0.008099 -0.021115  0.009997  82683  19.997388
4        -0.078773 -0.005999 -0.017388  0.009108  82691  19.999323
5        -0.068259 -0.000000 -0.012369  0.007607  82116  19.860256
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_3\returns_report.pdf
Information Analysis
                ic
IC Mean      0.026
IC Std.      0.198
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.130
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_3\information_report.pdf
Nan Data Count (should be zero) : 0;  Percentage of effective data: 52%


Value of signals of Different Quantiles Statistics
                   min           max           mean           std  count  \
quantile                                                                   
1        -1.458161e+10  1.948145e+00 -263277.198399  5.238580e+07  81342   
2        -1.081012e+00  2.108829e+00       0.062442  2.855984e-01  80772   
3        -5.775263e-01  2.170505e+00       0.558179  2.597112e-01  80764   
4        -1.450277e-01  2.203910e+00       1.076562  2.916370e-01  80772   
5         3.750683e-01  4.399282e+09  131702.733601  1.792908e+07  80186   

            count %  
quantile             
1         20.142335  
2         20.001189  
3         19.999208  
4         20.001189  
5         19.856080  
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_4\returns_report.pdf
Information Analysis
                ic
IC Mean      0.000
IC Std.      0.081
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.005
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_4\information_report.pdf
Nan Data Count (should be zero) : 0;  Percentage of effective data: 53%


Value of signals of Different Quantiles Statistics
                   min           max          mean           std  count  \
quantile                                                                  
1        -8.818676e+08  6.625718e+05 -9.376437e+06  2.467965e+07  83288   
2        -8.217876e+07  1.246660e+07 -1.482240e+06  3.203704e+06  82691   
3        -5.520948e+07  2.433595e+07 -4.138087e+05  1.780147e+06  82685   
4        -1.401159e+07  4.165598e+07  2.652617e+05  2.253609e+06  82691   
5        -2.725820e+06  1.762297e+09  5.196908e+06  2.913466e+07  82117   

            count %  
quantile             
1         20.143565  
2         19.999178  
3         19.997727  
4         19.999178  
5         19.860353  
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_5\returns_report.pdf
Information Analysis
                ic
IC Mean      0.042
IC Std.      0.138
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.307
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\sig_5\information_report.pdf

plot_pfm(signal_dict.get('sig_5'), 'sig_5')

4_外部因子编写

import talib as ta
import pandas as pd
close_adj = dv.get_ts('close_adj')
close = close_adj.dropna(how='all', axis=1)
slope_df = pd.DataFrame({name: -ta.LINEARREG_SLOPE(value.values, 3) for name, value in close.iteritems()}, index=close.index)

dv.remove_field('slope')

Field name [slope] does not exist.

dv.append_df(slope_df,'slope')

plot_pfm(dv.get_ts('slope'), 'slope')

Nan Data Count (should be zero) : 0;  Percentage of effective data: 53%


Value of signals of Different Quantiles Statistics
                min        max      mean       std  count    count %
quantile                                                            
1        -57.788764   0.924510 -0.807342  1.487646  83172  20.143327
2         -3.229754   1.582843 -0.109449  0.256158  82577  19.999225
3         -1.875339   2.531553  0.012089  0.219972  82570  19.997530
4         -1.387552   4.972311  0.131095  0.322812  82577  19.999225
5         -0.757552  51.460955  0.739474  1.514192  82005  19.860693
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\slope\returns_report.pdf
Information Analysis
                ic
IC Mean      0.045
IC Std.      0.162
t-stat(IC)     NaN
p-value(IC)    NaN
IC Skew        NaN
IC Kurtosis    NaN
Ann. IR      0.278
Figure saved: C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\hs300\slope\information_report.pdf

dv.save_dataview('stockdata/hs300')

Store data...
Dataview has been successfully saved to:
C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\stockdata\hs300

You can load it with load_dataview('C:\Users\small\OneDrive\notebook\Internet_Course\Selection\JAQS\stockdata\hs300')