事件发生时在python中收集数据点

如何解决事件发生时在python中收集数据点

当股票价格超过移动平均线（MA）时，我想收集价格点。在这里，我将以Microsoft为例。

我的股票价格为：

Microsoft_Stock = web.DataReader(['MSFT'],'yahoo',start =...,end=...)

，然后将此股票的移动平均线设为：

Microsoft_Stock['Moving_Average_20_Days'] = Microsoft_Stock['Close'].rolling(window = 20,min_periods = 1).mean()

如何获得这两条线交叉的点的列表？

解决方法

可能有一种更聪明的提取价格的方法，但是您可以通过找到收盘价和移动平均线之间的差并提取其绝对值（例如低于1）来找到交叉点的价格。 / p>

MSFT['cross_point'] = MSFT['Adj Close'] - MSFT['MA_20_Days']
MSFT[abs(MSFT['cross_point']) < 1]

Attributes  Adj Close   Close   High    Low Open    Volume  MA_20_Days  cross_point
Date                                
2019-01-04  99.652115   101.930000  102.510002  98.930000   99.720001   44060600.0  100.505001  -0.852886
2019-01-08  100.502670  102.800003  103.970001  101.709999  103.040001  31514400.0  101.146667  -0.643997
2019-01-09  101.939812  104.269997  104.879997  103.239998  103.860001  32280800.0  101.592857  0.346954
2019-01-10  101.284798  103.599998  103.750000  102.379997  103.220001  30067600.0  101.843750  -0.558952
2019-01-15  102.663292  105.010002  105.050003  101.879997  102.510002  31587600.0  102.237273  0.426018
... ... ... ... ... ... ... ... ...
2020-08-14  208.396240  208.899994  209.589996  207.509995  208.759995  17958900.0  208.239999  0.156241
2020-09-28  209.440002  209.440002  212.570007  208.059998  210.880005  32004900.0  209.644500  -0.204498
2020-10-02  206.190002  206.190002  210.990005  205.539993  208.000000  33154800.0  206.369000  -0.178998
2020-10-06  205.910004  205.910004  210.179993  204.820007  208.820007  28554300.0  206.338000  -0.427997
2020-10-27  213.250000  213.250000  214.669998  210.330002  211.589996  36700300.0  214.202501  -0.952501
103 rows × 8 columns

,

由于您尚未指定所需的输出确切是什么，因此我假设您要使用两个值相同的日期或交叉点之后的日期。

import numpy as np
from matplotlib import pyplot as plt
import datetime as dt 
import pandas_datareader.data as web

#your input routine
start,end = dt.datetime(2020,3,1),dt.datetime(2020,5,31)
MS = web.DataReader('MSFT','yahoo',start=start,end=end)
MS['MA_20_D'] = MS['Close'].rolling(window = 20,min_periods = 1).mean()

#identify changes between the curves
MS["CP"] = MS["Close"] - MS["MA_20_D"]
MS["CP_Arr"] = MS["CP"].map(np.sign).diff().abs()
#deal with the special case that they have the same value
#either at one point
MS.loc[MS["CP"].shift() == 0,"CP_Arr"] = 0
#or a series of points
MS.loc[MS["CP"] == 0,"CP_Arr"] = 1

#if CP_Arr is now >0,then there was 
#either a crossover between this and the previous date
#or the values are equal at this specific date
#now we filter for this to get an array of all these points
CP_MS = MS[MS["CP_Arr"].ge(1)]
print(CP_MS)

#to verify our approach,let's plot the curves
plt.plot(MS["Close"],label="Close")
plt.plot(MS["MA_20_D"],label="Moving_Average_20_Days")

#and mark the dates we have identified
for covdate in CP_MS.index.values:
    print(covdate)
    plt.axvline(covdate)

plt.xticks(rotation = 45,ha="right")
plt.legend()
plt.tight_layout()
plt.show()

样本输出：