Tushare Day5——超简单的策略实现

股票分析

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tushare as ts

用tushare包获取某支股票的历史行情数据

查询茅台酒的股票编码并获取数据

df=ts.get_k_data("600519",start='1988-01-01')  
df.to_csv("600519.csv")
df.head(5)

本接口即将停止更新，请尽快使用Pro版接口：https://waditu.com/document/2

	date	open	close	high	low	volume	code
0	2001-08-27	5.392	5.554	5.902	5.132	406318.00	600519
1	2001-08-28	5.467	5.759	5.781	5.407	129647.79	600519
2	2001-08-29	5.777	5.684	5.781	5.640	53252.75	600519
3	2001-08-30	5.668	5.796	5.860	5.624	48013.06	600519
4	2001-08-31	5.804	5.782	5.877	5.749	23231.48	600519

只显示感兴趣的四列（开、收盘价，最高、最低价）

#将date列设置为索引，并把它变为时间对象，最后只显示保留四个特征的矩阵
df=pd.read_csv("600519.csv",index_col="date",parse_dates=["date"])[['open','close','high','low']]
df.head(5)

	open	close	high	low
date
2001-08-27	5.392	5.554	5.902	5.132
2001-08-28	5.467	5.759	5.781	5.407
2001-08-29	5.777	5.684	5.781	5.640
2001-08-30	5.668	5.796	5.860	5.624
2001-08-31	5.804	5.782	5.877	5.749

输出该股票所有收盘比开盘上涨大于等于3%及以上的日期

该股票所有收盘比开盘上涨大于等于3%的行

(df["close"]-df["open"])/df["open"] #增长率
df[(df["close"]-df["open"])/df["open"]>=0.03 ].head(5)#将符合算式的行都输出出来

	open	close	high	low
date
2001-08-27	5.392	5.554	5.902	5.132
2001-08-28	5.467	5.759	5.781	5.407
2001-09-10	5.531	5.734	5.757	5.470
2001-12-21	5.421	5.604	5.620	5.421
2002-01-18	5.437	5.726	5.762	5.421

该股票所有收盘比开盘上涨大于等于3%的日期

df[(df["close"]-df["open"])/df["open"]>=0.03 ].index #将符合算式的列输出来

DatetimeIndex(['2001-08-27', '2001-08-28', '2001-09-10', '2001-12-21',
               '2002-01-18', '2002-01-31', '2003-01-14', '2003-10-29',
               '2004-01-05', '2004-01-14',
               ...
               '2020-07-06', '2020-07-07', '2020-07-13', '2020-12-30',
               '2021-01-05', '2021-01-12', '2021-01-25', '2021-02-04',
               '2021-02-09', '2021-02-10'],
              dtype='datetime64[ns]', name='date', length=322, freq=None)

该股票所有收盘比开盘上涨大于等于10%的行

df[(df["close"]-df["open"])/df["open"]>=0.1 ].head(5)

	open	close	high	low
date
2004-03-02	5.463	6.031	6.079	5.463
2005-06-08	11.383	12.555	12.639	11.383

最后发现只有上涨等于10%的股票，是因为中国的涨幅上限为10%

输出该股票所有开盘比前日收盘跌幅超过2%的日期

所有列错位下移

df.shift(1).head(5) #用shift下移

	open	close	high	low
date
2001-08-27	NaN	NaN	NaN	NaN
2001-08-28	5.392	5.554	5.902	5.132
2001-08-29	5.467	5.759	5.781	5.407
2001-08-30	5.777	5.684	5.781	5.640
2001-08-31	5.668	5.796	5.860	5.624

所有列上移

df.shift(-1).head(5)  #用shift上移

	open	close	high	low
date
2001-08-27	5.467	5.759	5.781	5.407
2001-08-28	5.777	5.684	5.781	5.640
2001-08-29	5.668	5.796	5.860	5.624
2001-08-30	5.804	5.782	5.877	5.749
2001-08-31	5.812	5.779	5.870	5.757

收盘价单列下移

df["close"].shift(1).head(5)

date
2001-08-27      NaN
2001-08-28    5.554
2001-08-29    5.759
2001-08-30    5.684
2001-08-31    5.796
Name: close, dtype: float64

输出该股票所有开盘比前日收盘跌幅超过2%的行

df[(df["open"]-df["close"].shift(1))/df["close"].shift(1)<=-0.02].head(5)

	open	close	high	low
date
2001-09-12	5.520	5.621	5.656	5.515
2002-06-26	5.824	5.757	5.843	5.712
2002-12-13	4.508	4.628	4.670	4.508
2004-07-01	6.357	6.792	6.810	6.355
2004-10-29	9.658	9.899	9.986	9.658

df[(df["open"]-df["close"].shift(1))/df["close"].shift(1)<=-0.02].shape

(85, 4)

输出该股票所有开盘比前日收盘跌幅超过2%的日期

df[(df["open"]-df["close"].shift(1))/df["close"].shift(1)<=-0.02].head(5).index

DatetimeIndex(['2001-09-12', '2002-06-26', '2002-12-13', '2004-07-01',
               '2004-10-29'],
              dtype='datetime64[ns]', name='date', freq=None)

策略实现：假如从tushare有记录日期开始，每月第一个交易日买入一手股票，每年最后一个交易日卖出所有股票，到今天为止，我的收益如何？

剔除无用数据（最前最后的月数据不全），并尝试各月汇总

df=df["2001-09":"2017-11"]#剔除所有无用数据
df
df.resample("M").sum().head(5) #重新取样，设置频率为月,将这个月加总

	open	close	high	low
date
2001-09-30	113.752	113.933	115.265	112.730
2001-10-31	99.743	98.974	100.669	97.761
2001-11-30	119.185	119.028	120.324	117.904
2001-12-31	118.281	118.624	119.923	117.189
2002-01-31	113.996	114.204	116.206	112.190

取出每月月初第一支股票信息

df.resample("MS").last().head(5)

	open	close	high	low
date
2001-09-01	5.765	5.795	5.812	5.702
2001-10-01	5.468	5.476	5.482	5.437
2001-11-01	5.442	5.521	5.538	5.437
2001-12-01	5.885	6.023	6.140	5.852
2002-01-01	5.624	5.874	5.937	5.624

df_monthly = df.resample("MS").first() #MS为每月的开始
df_monthly.head(5) 

	open	close	high	low
date
2001-09-01	5.812	5.779	5.870	5.757
2001-10-01	5.781	5.715	5.809	5.663
2001-11-01	5.476	5.557	5.568	5.460
2001-12-01	5.523	5.651	5.667	5.521
2002-01-01	6.049	5.866	6.062	5.851

resample的用法

• print(ts.resample(‘5D’).mean(),‘→ 求平均值\n’)
• print(ts.resample(‘5D’).max(),‘→ 求最大值\n’)
• print(ts.resample(‘5D’).min(),‘→ 求最小值\n’)
• print(ts.resample(‘5D’).median(),‘→ 求中值\n’)
• print(ts.resample(‘5D’).first(),‘→ 返回第一个值\n’)
• print(ts.resample(‘5D’).last(),‘→ 返回最后一个值\n’)
• print(ts.resample(‘5D’).ohlc(),‘→ OHLC重采样\n’)

取出每年的最后一支股票

df.resample('A').last()  #年用A表示，虽然date显示的是每年12月31号的值，但实际是表格里有的最后一天的值（如果12.31没有记录就显示当年最后一天的值）

	open	close	high	low
date
2001-12-31	5.885	6.023	6.140	5.852
2002-12-31	4.473	4.448	4.504	4.447
2003-12-31	4.940	4.921	4.940	4.888
2004-12-31	9.325	9.310	9.579	9.168
2005-12-31	14.309	14.039	14.316	13.817
2006-12-31	53.482	54.946	57.617	52.900
2007-12-31	139.495	144.783	144.846	137.085
2008-12-31	68.502	68.818	69.318	68.058
2009-12-31	107.993	108.369	108.516	107.718
2010-12-31	117.103	118.469	118.701	116.620
2011-12-31	138.039	138.468	139.600	136.105
2012-12-31	155.208	152.087	156.292	150.144
2013-12-31	93.188	96.480	97.179	92.061
2014-12-31	157.642	161.056	161.379	157.132
2015-12-31	207.487	207.458	208.704	207.106
2016-12-31	317.239	324.563	325.670	317.239
2017-12-31	626.613	622.166	639.401	617.236

去掉信息不全的17年的最后一支股票

df_yearly = df.resample('A').last()[:-1]#由于17年最后一天还没有，所以切掉倒数第一行
df_yearly

	open	close	high	low
date
2001-12-31	5.885	6.023	6.140	5.852
2002-12-31	4.473	4.448	4.504	4.447
2003-12-31	4.940	4.921	4.940	4.888
2004-12-31	9.325	9.310	9.579	9.168
2005-12-31	14.309	14.039	14.316	13.817
2006-12-31	53.482	54.946	57.617	52.900
2007-12-31	139.495	144.783	144.846	137.085
2008-12-31	68.502	68.818	69.318	68.058
2009-12-31	107.993	108.369	108.516	107.718
2010-12-31	117.103	118.469	118.701	116.620
2011-12-31	138.039	138.468	139.600	136.105
2012-12-31	155.208	152.087	156.292	150.144
2013-12-31	93.188	96.480	97.179	92.061
2014-12-31	157.642	161.056	161.379	157.132
2015-12-31	207.487	207.458	208.704	207.106
2016-12-31	317.239	324.563	325.670	317.239

把2001所有的月初股票信息取出来

df_monthly["2001"]

	open	close	high	low
date
2001-09-01	5.812	5.779	5.870	5.757
2001-10-01	5.781	5.715	5.809	5.663
2001-11-01	5.476	5.557	5.568	5.460
2001-12-01	5.523	5.651	5.667	5.521

取得最后未卖出股票的价值(获取股票最后一天的开盘价）

price_last=df['open'][-1]  #最后一天每卖出股票的开盘价

策略代码（输出每年的收益及最终收益）

cost_money = 0
hold = 0
for year in range(2001,2018):
    
    cost_money += df_monthly[str(year)]['open'].sum()*100 #df_monthly[str(year)]['open'].sum()*100是计算每年花的钱
    hold += len(df_monthly[str(year)]['open'])*100  #计算共有多少支股票已入手
    if year != 2017:  #2017年买的还没卖出去
        cost_money -= df_yearly[str(year)]['open'][0]*hold    #卖出去的价格
        hold=0
        print(-cost_money)

cost_money -= hold * price_last
print(-cost_money)

94.80000000000018
-1082.6999999999998
-493.4999999999982
2027.2000000000025
3889.5000000000055
38129.100000000006
110589.8
70515.90000000001
96941.2
116591.6
123173.59999999998
116746.5
78106.39999999998
121609.49999999996
143901.09999999998
212044.10000000003
425573.8

最终收益为42.55738万

取出某行信息：

df_yearly['2001']

	open	close	high	low
date
2001-12-31	5.885	6.023	6.14	5.852