180 lines
7.6 KiB
Python
180 lines
7.6 KiB
Python
# -*- encoding:utf-8 -*-
|
||
"""
|
||
量化波动程度模块
|
||
"""
|
||
|
||
from __future__ import division
|
||
from __future__ import print_function
|
||
from __future__ import absolute_import
|
||
|
||
import matplotlib.pyplot as plt
|
||
import numpy as np
|
||
import pandas as pd
|
||
|
||
from ..TLineBu.ABuTLine import AbuTLine
|
||
from ..CoreBu.ABuPdHelper import pd_rolling_std, pd_ewm_mean, pd_ewm_std, pd_resample
|
||
from ..UtilBu import ABuStatsUtil
|
||
from ..UtilBu.ABuDTUtil import plt_show
|
||
|
||
|
||
def show_wave_return(kl_pd):
|
||
"""
|
||
可视化收益的移动平均std和加权移动平均std
|
||
注意会修改kl_pd，只做测试使用，内部未做copy处理，
|
||
如不能改动，外部自copy操作，再传递进来
|
||
:param kl_pd: 金融时间序列，pd.DataFrame对象
|
||
"""
|
||
|
||
pre_close = kl_pd['close'].shift(1)
|
||
# noinspection PyTypeChecker
|
||
kl_pd['return'] = np.where(pre_close == 0, 0, np.log(kl_pd['close'] / pre_close))
|
||
kl_pd['mov_std'] = pd_rolling_std(kl_pd['return'], window=20, center=False) * np.sqrt(20)
|
||
kl_pd['std_ewm'] = pd_ewm_std(kl_pd['return'], span=20, min_periods=20, adjust=True) * np.sqrt(20)
|
||
kl_pd[['close', 'mov_std', 'std_ewm', 'return']].plot(subplots=True, figsize=(16, 12), grid=True)
|
||
plt.show()
|
||
|
||
|
||
def calc_wave_std(kl_pd, xd=21, ewm=True, show=True):
|
||
"""
|
||
计算收益的移动平均std或者加权移动平均std技术线，使用
|
||
AbuTLine封装技术线实体，不会修改kl_pd，返回AbuTLine对象
|
||
:param kl_pd: 金融时间序列，pd.DataFrame对象
|
||
:param xd: 计算移动平均std或加权移动平均std使用的窗口参数，默认21
|
||
:param ewm: 是否使用加权移动平均std计算
|
||
:param show: 是否可视化，可视化使用AbuTLine.show接口
|
||
:return: 返回AbuTLine对象
|
||
"""
|
||
|
||
pre_close = kl_pd['close'].shift(1)
|
||
# noinspection PyTypeChecker
|
||
change = np.where(pre_close == 0, 0, np.log(kl_pd['close'] / pre_close))
|
||
if ewm:
|
||
roll_std = pd_ewm_std(change, span=xd, min_periods=1, adjust=True) * np.sqrt(xd)
|
||
else:
|
||
roll_std = pd_rolling_std(change, window=xd, min_periods=1, center=False) * np.sqrt(xd)
|
||
|
||
# min_periods=1还是会有两个nan，填了
|
||
roll_std = pd.Series(roll_std).fillna(method='bfill')
|
||
# 主要目的就是通过roll_std构造AbuTLine对象line
|
||
line = AbuTLine(roll_std, 'wave std')
|
||
if show:
|
||
line.show()
|
||
|
||
return line
|
||
|
||
|
||
def calc_wave_abs(kl_pd, xd=21, show=True):
|
||
"""
|
||
计算金融时间序列kl_pd在的绝对波动，通过参数xd对波动进行重采样
|
||
在默认xd＝21情况下，变成了月震荡幅度，使用ABuStatsUtil.demean对
|
||
月震荡幅度进行去均值操作后得到技术线demean_wave，AbuTLine包裹
|
||
技术线返回
|
||
:param kl_pd: 金融时间序列，pd.DataFrame对象
|
||
:param xd: 对波动进行重采样的周期，单位天，int
|
||
:param show: 是否可视化
|
||
:return: 返回AbuTLine对象
|
||
"""
|
||
# 不考虑正负，只考虑波动，np.abs(kl_pd['p_change'])
|
||
abs_pct_change = np.abs(kl_pd['p_change'])
|
||
xd_resample = '%dD' % xd
|
||
# 通过pd_resample重采样，使用how=sum, 即默认xd＝21情况下，变成了月震荡幅度
|
||
change_ratio_sum = pd_resample(abs_pct_change, xd_resample, how='sum')
|
||
"""
|
||
eg: change_ratio_sum
|
||
2014-07-24 37.13
|
||
2014-08-14 39.33
|
||
2014-09-04 25.16
|
||
2014-09-25 27.53
|
||
2014-10-16 27.78
|
||
...
|
||
2016-04-14 25.17
|
||
2016-05-05 42.07
|
||
2016-05-26 18.93
|
||
2016-06-16 33.25
|
||
2016-07-07 10.79
|
||
"""
|
||
# 使用ABuStatsUtil.demean进行去均值操作
|
||
demean_wave = ABuStatsUtil.demean(change_ratio_sum)
|
||
"""
|
||
eg: demean_wave
|
||
2014-07-24 -1.6303
|
||
2014-08-14 0.5697
|
||
2014-09-04 -13.6003
|
||
2014-09-25 -11.2303
|
||
...
|
||
2016-05-05 3.3097
|
||
2016-05-26 -19.8303
|
||
2016-06-16 -5.5103
|
||
2016-07-07 -27.9703
|
||
"""
|
||
line = AbuTLine(demean_wave, 'demean sum change wave')
|
||
if show:
|
||
# 计算pd_resample how='mean'只是为了_show_wave里面显示价格曲线
|
||
xd_mean_close = pd_resample(kl_pd.close, xd_resample, how='mean')
|
||
# 这里不使用AbuTLine.show，因为需要绘制另一个对比line，价格均线xd_mean_close
|
||
_show_wave(demean_wave, line.high, line.mean, line.low, xd_mean_close)
|
||
# TODO AbuTLine中添加多条线的标准对比方法，左右双轴和数据变化方式
|
||
return line
|
||
|
||
|
||
def calc_wave_weight_abs(kl_pd, xd=21, span=3, show=True):
|
||
"""
|
||
计算金融时间序列kl_pd的绝对波动，通过参数xd对波动进行重采样
|
||
在默认xd＝21情况下，变成了月震荡幅度，使用ABuStatsUtil.demean对
|
||
月震荡幅度进行去均值操作后得到技术线demean_wave，与calc_wave_abs不同，
|
||
使用squared * np.sign(demean_wave)放大了wave，即大的愈加大，且
|
||
继续对squared_wave进行时间加权得到技术线形成ewm_wave，AbuTLine包裹技术线返回
|
||
:param kl_pd: 金融时间序列，pd.DataFrame对象
|
||
:param xd: 对波动进行重采样的周期，单位天，int
|
||
:param span: 对squared_wave进行时间加权的窗口参数，int
|
||
:param show: 是否可视化
|
||
:return: 返回AbuTLine对象
|
||
"""
|
||
|
||
# 不考虑正负，只考虑波动，np.abs(kl_pd['p_change'])
|
||
abs_pct_change = np.abs(kl_pd['p_change'])
|
||
xd_resample = '%dD' % xd
|
||
# 通过pd_resample重采样，使用how=sum, 即默认xd＝21情况下，变成了月震荡幅度
|
||
change_ratio_sum = pd_resample(abs_pct_change, xd_resample, how='sum')
|
||
# 使用ABuStatsUtil.demean进行去均值操作
|
||
demean_wave = ABuStatsUtil.demean(change_ratio_sum)
|
||
# 与calc_wave_abs不同，使用squared * np.sign(demean_wave)放大了wave，即大的愈加大
|
||
squared_wave = (demean_wave ** 2) * np.sign(demean_wave)
|
||
# ewmd的span最后决定了一切， span默认值之对应xd默认值，xd变动, span也要变
|
||
ewm_wave = pd_ewm_mean(squared_wave, span=span, min_periods=span, adjust=True)
|
||
line = AbuTLine(ewm_wave, 'squared ewm wave')
|
||
if show:
|
||
# 计算pd_resample how='mean'只是为了_show_wave里面显示价格曲线
|
||
xd_mean_close = pd_resample(kl_pd.close, xd_resample, how='mean')
|
||
# 这里不使用AbuTLine.show，因为需要绘制另一个对比line，价格均线xd_mean_close
|
||
_show_wave(ewm_wave, line.high, line.mean, line.low, xd_mean_close)
|
||
return line
|
||
|
||
|
||
def _show_wave(wave, above, wave_mean, below, xd_mean_close):
|
||
"""
|
||
calc_wave_abs和calc_wave_weight_abs形成技术线的可视化方法
|
||
不使用AbuTLine.show，因为需要绘制另一个对比line，价格均线xd_mean_close，
|
||
使用双坐标轴的方式进行可视化
|
||
"""
|
||
with plt_show():
|
||
fig, ax1 = plt.subplots()
|
||
plt.plot(wave)
|
||
plt.axhline(above, color='c')
|
||
plt.axhline(wave_mean, color='r')
|
||
plt.axhline(below, color='g')
|
||
_ = plt.setp(plt.gca().get_xticklabels(), rotation=30)
|
||
plt.legend(['wave', 'above', 'wave_mean', 'below'],
|
||
bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
|
||
|
||
# 采用左右两个y轴坐标显示
|
||
# noinspection PyUnusedLocal
|
||
ax2 = ax1.twinx()
|
||
plt.plot(xd_mean_close, c='y')
|
||
plt.plot(xd_mean_close, 'ro')
|
||
plt.legend(['mean close'],
|
||
bbox_to_anchor=(1.05, 1), loc=3, borderaxespad=0.)
|
||
# 当时间序列太长时使用将时间显示倾斜30度角
|
||
_ = plt.setp(plt.gca().get_xticklabels(), rotation=30)
|
||
plt.title('wave line')
|