学习记录656@python之MPT现代/均值-方差资产组合理论

最新推荐文章于 2023-09-23 13:11:28 发布

教练我想学编程

最新推荐文章于 2023-09-23 13:11:28 发布

阅读量635

点赞数 1

文章标签： python MPT投资组合理论

本文链接：https://blog.csdn.net/weixin_44663675/article/details/127178629

版权

理论就不多加叙述了，感兴趣的可以去查找MPT现代/均值-方差资产组合理论。
均值-方差理论的核心思想是同时考察资产组合的预期收益和风险，研究当我们有一系列可选资产的时候，应如何对其配置资金权重，从而可以得到最好的收益风险比，也就是最好的夏普比率。
在这里插入图片描述

以下模拟投资组合的预期收益率和波动情况

import pandas as pd
import numpy as np
%matplotlib inline
import matplotlib.pyplot as plt
plt.style.use('ggplot')

dates=pd.date_range(start='2010/1/1',end='2010/12/31')
df_rtn=pd.DataFrame(index=dates)
df_rtn['s1']=np.random.normal(0.1, 0.05, len(dates))
df_rtn['s2']=np.random.normal(0.2, 0.10, len(dates))
df_rtn['s3']=np.random.normal(0.3, 0.15, len(dates))
df_rtn['s4']=np.random.normal(0.4, 0.2, len(dates))
df_rtn['s5']=np.random.normal(-0.2, 0.10, len(dates))
df_rtn

	s1	s2	s3	s4	s5
2010-01-01	0.165496	0.054865	0.550757	0.416184	-0.241358
2010-01-02	0.123811	0.066198	0.236836	0.269195	-0.276284
2010-01-03	0.059512	0.321974	0.194933	0.144246	-0.250190
2010-01-04	0.056678	0.003651	0.499698	0.579080	-0.235689
2010-01-05	0.065828	0.129830	0.291351	0.397176	0.001602
...	...	...	...	...	...
2010-12-27	0.104600	0.335536	0.303462	0.346302	-0.243276
2010-12-28	0.047231	0.199015	0.475520	-0.043252	-0.236914
2010-12-29	0.130282	0.131229	0.386482	0.614867	0.059841
2010-12-30	0.093123	0.288365	0.442705	0.326410	-0.126081
2010-12-31	0.118626	0.213211	0.601449	0.605520	-0.116745

365 rows × 5 columns

df_rtn.plot(figsize=(10,6))

<AxesSubplot:>

在这里插入图片描述

df_rtn.mean()*252

s1     25.781191
s2     49.999966
s3     78.002242
s4    102.503923
s5    -50.197467
dtype: float64

df_rtn.cov()*252

	s1	s2	s3	s4	s5
s1	0.677835	-0.049152	-0.003292	-0.264586	0.008482
s2	-0.049152	2.617314	-0.143490	0.321991	0.113303
s3	-0.003292	-0.143490	5.242705	0.443050	0.032384
s4	-0.264586	0.321991	0.443050	9.358566	0.395169
s5	0.008482	0.113303	0.032384	0.395169	2.581551

随机一个投资组合权重

n=len(df_rtn.columns)
w=np.random.random(n)
w=w/np.sum(w)
w

array([0.21695834, 0.14631267, 0.1290212 , 0.20732207, 0.30038572])

# 投资组合收益率,s1~s5是每个资产的收益情况，每个资产的均值乘以不同的权重计算出投资组合的收益率
p_ret= np.sum(df_rtn.mean()*w)*252
p_ret

29.145739519785984

#投资组合波动率
p_vol=np.sqrt(np.dot(w.T,np.dot(df_rtn.cov()*252,w)))
p_vol
# 以上是某书籍中的波动率的衡量，感觉不太对，因为不是之前学过的标准的投资组合的风险计算方式，不知道是不是量化中特有的

0.9401632467991151

模拟多个权重的情况下，收益率和风险的变化情况，并作图，这个图就是经典的投资组合有效边界图

n=len(df_rtn.columns)

# 保存一系列权重对应的投资组合收益率
p_rets=[]

# 保存一系列权重对应的投资组合波动率
p_vols=[]

# 随机生成10000组权重
for i in range(10000):
    w=np.random.random(n)
    w/=np.sum(w)

    # # 投资组合收益率
    p_ret= np.sum(df_rtn.mean()*w)*252

    #投资组合波动率（需要注意的是，使用np生成的一维数据，实际上是N*1矩阵，而不是1*N矩阵）
    p_vol=np.sqrt(np.dot(w.T,np.dot(df_rtn.cov()*252,w)))

    p_rets.append(p_ret)
    p_vols.append(p_vol)

p_rets=np.array(p_rets)
p_vols=np.array(p_vols)

p_rets

array([37.37827321, 24.83548494, 27.16777023, ..., 51.47163399,
       26.03826012, 31.35167142])

p_vols

array([1.05508155, 0.89835554, 0.97479329, ..., 1.09094976, 0.78924616,
       0.91253653])

plt.figure(figsize=(10,6))
plt.scatter(p_vols, p_rets, c=p_rets/p_vols,marker='o')
plt.grid(True)

# x轴是风险，y轴是收益
plt.xlabel('volatility')
plt.ylabel('return')

# 夏普比率y/x的值，越黄的部分夏普比率越高，说明收益风险比最优秀
plt.colorbar(label='Sharpe ratio')

<matplotlib.colorbar.Colorbar at 0x1d6bd551588>

在这里插入图片描述

接下来求解最优投资组合权重，也就是求解最大的夏普比率下的投资组合权重

def portfolio_stat(weights):
    """获取投资组合的各种统计值

参数:
        weights：分配的资金权重

返回值：
        p_ret：投资组合的收益率
        p_vol：投资组合的波动率
        p_sr：投资组合的夏普比率

注意：rets是全局变量，在函数外部定义
    """

    w=np.array(weights)
    p_ret=np.sum(rets.mean()*w)*252
    p_vol=np.sqrt(np.dot(w.T,np.dot(rets.cov()*252,w)))
    p_sr=p_ret/p_vol

    return np.array([p_ret,p_vol,p_sr])

# 该函数返回了夏普比率的负值，换句话说，我们需要对该函数进行最小化操作。之所以使用这种形式，是因为scipy.optimization只有minimize函数（机器学习中优化目标通常是最小化损失函数），所以这里使用了这种看起来有点奇怪的形式。
def min_func_sharpe(weights):
    """优化的目标函数，最小化夏普比率的负值，即最大化夏普比率
    """
    return -portfolio_stat(weights)[2]

import scipy.optimize as sco

# 投资组合
rets=df_rtn

n=len(rets.columns)

# 优化的约束条件：资金的权重和为1，下面表示np.sum(x)-1 eq 0
cons=({'type':'eq','fun':lambda x:np.sum(x)-1})
 #参数取值范围
bnds=tuple((0,1) for x in range(n))

# 生成初始权重
w_initial=n*[1./n,]
# 计算最有权重
opts_sharpe=sco.minimize(min_func_sharpe,w_initial,method='SLSQP',bounds=bnds,constraints=cons)
opts_sharpe

     fun: -65.88146585394263
     jac: array([ 4.17900085e-03, -1.56126022e-02,  7.96985626e-03,  3.96728516e-04,
        7.72509413e+01])
 message: 'Optimization terminated successfully'
    nfev: 82
     nit: 12
    njev: 12
  status: 0
 success: True
       x: array([4.94327710e-01, 2.19114359e-01, 1.64106232e-01, 1.22451699e-01,
       1.55918074e-11])

# fun对应最优夏普比率,注意取负值，x为权重
opts_sharpe['x'].round(2)

array([0.49, 0.22, 0.16, 0.12, 0.  ])

接下来我们来实战一下，根据近一年比亚迪，长安汽车，长城汽车近一年的收益率情况，搭建投资组合，计算最有的投资权重。

import tushare as ts
ts.set_token('d56646c5b067c6715534aff6a6487d736a545e525bbb3bd5abf7f9ed')
df_byd = ts.pro_bar(ts_code='002594.SZ', adj='qfq', start_date='20220427', end_date='20220627')
df_byd.sort_values(by="trade_date",inplace=True)
df_byd

	ts_code	trade_date	open	high	low	close	pre_close	change	pct_chg	vol	amount
39	002594.SZ	20220427	226.61	235.81	218.81	235.01	231.81	3.20	1.3804	254341.67	5.794106e+06
38	002594.SZ	20220428	234.60	235.50	230.00	231.98	235.01	-3.03	-1.2893	146372.13	3.405250e+06
37	002594.SZ	20220429	233.80	243.83	230.75	243.00	231.98	11.02	4.7504	248163.10	5.912448e+06
36	002594.SZ	20220505	246.20	252.39	241.77	248.80	243.00	5.80	2.3868	253258.79	6.282898e+06
35	002594.SZ	20220506	241.02	246.00	240.10	242.08	248.80	-6.72	-2.7010	145488.66	3.530097e+06
34	002594.SZ	20220509	231.00	236.89	231.00	232.73	242.08	-9.35	-3.8624	175116.35	4.086614e+06
33	002594.SZ	20220510	224.10	233.66	222.28	233.33	232.73	0.60	0.2578	209696.64	4.792959e+06
32	002594.SZ	20220511	232.00	256.66	231.50	252.70	233.33	19.37	8.3015	459525.71	1.148248e+07
31	002594.SZ	20220512	249.99	257.70	248.71	254.09	252.70	1.39	0.5501	241156.06	6.107995e+06
30	002594.SZ	20220513	259.80	271.50	255.91	270.08	254.09	15.99	6.2930	367120.61	9.733352e+06
29	002594.SZ	20220516	271.11	273.00	265.30	267.12	270.08	-2.96	-1.0960	202368.80	5.441552e+06
28	002594.SZ	20220517	268.98	285.00	268.50	282.50	267.12	15.38	5.7577	293489.03	8.193619e+06
27	002594.SZ	20220518	282.50	288.00	279.00	281.42	282.50	-1.08	-0.3823	205302.70	5.812001e+06
26	002594.SZ	20220519	275.31	285.35	274.30	284.51	281.42	3.09	1.0980	173636.51	4.879757e+06
25	002594.SZ	20220520	287.01	291.61	281.20	287.70	284.51	3.19	1.1212	178683.36	5.125700e+06
24	002594.SZ	20220523	292.70	298.00	287.81	295.20	287.70	7.50	2.6069	255588.16	7.496178e+06
23	002594.SZ	20220524	290.00	292.96	281.00	281.00	295.20	-14.20	-4.8103	290496.53	8.303004e+06
22	002594.SZ	20220525	281.26	288.65	278.03	287.20	281.00	6.20	2.2064	230857.73	6.563212e+06
21	002594.SZ	20220526	287.37	291.22	282.50	286.64	287.20	-0.56	-0.1950	143094.99	4.106312e+06
20	002594.SZ	20220527	292.95	297.61	286.65	288.88	286.64	2.24	0.7815	153186.85	4.467796e+06
19	002594.SZ	20220530	292.00	299.70	291.08	298.12	288.88	9.24	3.1986	218523.50	6.485732e+06
18	002594.SZ	20220531	300.00	300.96	293.51	295.91	298.12	-2.21	-0.7413	177011.19	5.242211e+06
17	002594.SZ	20220601	298.00	304.80	293.12	299.98	295.91	4.07	1.3754	216911.75	6.501902e+06
16	002594.SZ	20220602	296.00	305.00	295.22	301.02	299.98	1.04	0.3467	177727.77	5.355774e+06
15	002594.SZ	20220606	304.00	320.47	302.55	318.23	301.02	17.21	5.7172	294757.29	9.260503e+06
14	002594.SZ	20220607	318.00	319.87	311.11	314.89	318.23	-3.34	-1.0496	200220.89	6.296538e+06
13	002594.SZ	20220608	318.60	328.06	313.01	327.41	314.89	12.52	3.9760	290513.40	9.321933e+06
12	002594.SZ	20220609	334.03	334.80	321.00	322.41	327.41	-5.00	-1.5271	307541.72	1.005064e+07
11	002594.SZ	20220610	321.50	350.00	321.06	348.80	322.41	26.39	8.1852	438249.00	1.479062e+07
10	002594.SZ	20220613	346.00	358.86	340.60	348.80	348.80	0.00	0.0000	360681.32	1.257353e+07
9	002594.SZ	20220614	342.00	344.79	327.23	343.59	348.80	-5.21	-1.4937	404789.34	1.363462e+07
8	002594.SZ	20220615	343.97	346.23	332.52	335.00	343.59	-8.59	-2.5001	303947.87	1.028586e+07
7	002594.SZ	20220616	335.50	342.70	333.41	336.00	335.00	1.00	0.2985	190172.45	6.410739e+06
6	002594.SZ	20220617	330.10	341.20	327.07	338.00	336.00	2.00	0.5952	262147.59	8.830739e+06
5	002594.SZ	20220620	338.08	348.66	335.02	342.18	338.00	4.18	1.2367	245415.57	8.405362e+06
4	002594.SZ	20220621	339.76	341.80	332.20	341.00	342.18	-1.18	-0.3448	197827.39	6.666970e+06
3	002594.SZ	20220622	342.42	347.63	338.00	339.88	341.00	-1.12	-0.3284	186787.83	6.412052e+06
2	002594.SZ	20220623	344.70	354.05	338.00	353.50	339.88	13.62	4.0073	323142.26	1.125178e+07
1	002594.SZ	20220624	352.99	356.28	346.01	351.05	353.50	-2.45	-0.6931	235983.55	8.273168e+06
0	002594.SZ	20220627	352.07	357.00	348.33	349.69	351.05	-1.36	-0.3874	217155.24	7.630685e+06

df_ca = ts.pro_bar(ts_code='000625.SZ', adj='qfq', start_date='20220427', end_date='20220627')
df_ca.sort_values(by="trade_date",inplace=True)
df_ca

	ts_code	trade_date	open	high	low	close	pre_close	change	pct_chg	vol	amount
39	000625.SZ	20220427	6.6975	7.0399	6.5072	6.9715	6.8877	0.0838	1.2167	958948.85	8.507858e+05
38	000625.SZ	20220428	7.0171	7.3520	6.9410	7.2454	6.9715	0.2739	3.9289	1002657.73	9.455602e+05
37	000625.SZ	20220429	7.9685	7.9685	7.9685	7.9685	7.2454	0.7231	9.9801	298727.48	3.127677e+05
36	000625.SZ	20220505	8.6002	8.7676	8.4251	8.7676	7.9685	0.7991	10.0282	1115967.05	1.268370e+06
35	000625.SZ	20220506	9.0568	9.3917	8.8437	8.9655	8.7676	0.1979	2.2572	3547775.54	4.247638e+06
34	000625.SZ	20220509	8.9655	9.1253	8.5240	8.6610	8.9655	-0.3045	-3.3964	2144091.46	2.474558e+06
33	000625.SZ	20220510	8.4251	8.6915	8.1207	8.5849	8.6610	-0.0761	-0.8787	1858279.40	2.059471e+06
32	000625.SZ	20220511	8.5621	9.4449	8.5393	9.3080	8.5849	0.7231	8.4229	3463359.41	4.132147e+06
31	000625.SZ	20220512	9.0416	9.2851	8.9350	9.1025	9.3080	-0.2055	-2.2078	2401654.43	2.867413e+06
30	000625.SZ	20220513	9.1025	10.0158	9.0111	10.0158	9.1025	0.9133	10.0335	3869386.93	4.886108e+06
29	000625.SZ	20220516	10.0614	10.1451	9.7418	9.8103	10.0158	-0.2055	-2.0518	3231046.57	4.195913e+06
28	000625.SZ	20220517	9.7189	10.6398	9.7037	10.2821	9.8103	0.4718	4.8092	3594912.97	4.840887e+06
27	000625.SZ	20220518	10.2821	11.0051	9.9777	10.1756	10.2821	-0.1065	-1.0358	3949048.20	5.460881e+06
26	000625.SZ	20220519	9.9473	10.5713	9.8864	10.5105	10.1756	0.3349	3.2912	3292688.10	4.464683e+06
25	000625.SZ	20220520	10.3887	10.4800	10.1604	10.2669	10.5105	-0.2436	-2.3177	2470887.87	3.342778e+06
24	000625.SZ	20220523	10.2669	10.6322	10.1375	10.4343	10.2669	0.1674	1.6305	2488559.94	3.394024e+06
23	000625.SZ	20220524	10.7083	11.3400	10.6551	10.7159	10.4343	0.2816	2.6988	4130361.13	5.962842e+06
22	000625.SZ	20220525	10.5409	11.2944	10.2745	11.0660	10.7159	0.3501	3.2671	3396696.86	4.816627e+06
21	000625.SZ	20220526	10.8834	10.9443	10.5789	10.7388	11.0660	-0.3272	-2.9568	1916191.31	2.708203e+06
20	000625.SZ	20220527	10.8225	11.1117	10.5485	10.7692	10.7388	0.0304	0.2831	1781256.44	2.532129e+06
19	000625.SZ	20220530	10.8910	10.9443	10.5789	10.7616	10.7692	-0.0076	-0.0706	1570927.16	2.223216e+06
18	000625.SZ	20220531	10.7007	10.7083	10.3582	10.6170	10.7616	-0.1446	-1.3437	1892289.81	2.621740e+06
17	000625.SZ	20220601	10.8834	11.3020	10.8453	11.0508	10.6170	0.4338	4.0859	2930464.92	4.245724e+06
16	000625.SZ	20220602	10.8377	12.0859	10.8073	11.8043	11.0508	0.7535	6.8185	3655846.09	5.568282e+06
15	000625.SZ	20220606	11.7358	12.1620	11.5683	11.9184	11.8043	0.1141	0.9666	2941937.33	4.578257e+06
14	000625.SZ	20220607	11.7586	11.7814	11.4009	11.6292	11.9184	-0.2892	-2.4265	1913290.76	2.911673e+06
13	000625.SZ	20220608	11.6216	12.0250	11.4922	11.9413	11.6292	0.3121	2.6838	2254946.82	3.489635e+06
12	000625.SZ	20220609	11.8347	11.8347	11.4466	11.4922	11.9413	-0.4491	-3.7609	1467712.92	2.230526e+06
11	000625.SZ	20220610	11.3400	12.0250	11.2944	11.8728	11.4922	0.3806	3.3118	1876558.32	2.895509e+06
10	000625.SZ	20220613	11.7967	13.0524	11.7129	12.9307	11.8728	1.0579	8.9103	3696010.89	6.177846e+06
9	000625.SZ	20220614	12.6871	14.2245	12.6415	14.2245	12.9307	1.2938	10.0056	4249183.95	7.488059e+06
8	000625.SZ	20220615	14.3843	15.2215	14.1712	14.1712	14.2245	-0.0533	-0.3747	4727900.37	9.067054e+06
7	000625.SZ	20220616	14.1712	14.6888	13.9962	14.1408	14.1712	-0.0304	-0.2145	2970896.59	5.601199e+06
6	000625.SZ	20220617	14.0799	14.4604	13.8211	14.3995	14.1408	0.2587	1.8295	2463060.39	4.609731e+06
5	000625.SZ	20220620	14.8410	15.0312	14.3387	14.5670	14.3995	0.1675	1.1632	2629303.35	5.064239e+06
4	000625.SZ	20220621	14.3082	15.4498	14.2930	15.3737	14.5670	0.8067	5.5379	3363875.71	6.578857e+06
3	000625.SZ	20220622	15.3889	15.5868	14.9932	15.0312	15.3737	-0.3425	-2.2278	2619720.99	5.236025e+06
2	000625.SZ	20220623	15.2215	16.5382	15.0693	16.5382	15.0312	1.5070	10.0258	3512679.33	7.302952e+06
1	000625.SZ	20220624	16.7056	17.4134	16.2261	17.0100	16.5382	0.4718	2.8528	3504568.61	7.700924e+06
0	000625.SZ	20220627	17.5000	18.7100	16.7100	18.5000	17.0100	1.4900	8.7596	5753928.73	1.027395e+07

df_cc = ts.pro_bar(ts_code='601633.SH', adj='qfq', start_date='20220427', end_date='20220627')
df_cc.sort_values(by="trade_date",inplace=True)
df_cc

	ts_code	trade_date	open	high	low	close	pre_close	change	pct_chg	vol	amount
39	601633.SH	20220427	21.9337	22.9316	21.3051	22.7720	22.2630	0.5090	2.2863	341470.02	757962.925
38	601633.SH	20220428	22.5524	23.2410	22.4526	22.9316	22.7720	0.1596	0.7009	235284.09	538410.271
37	601633.SH	20220429	23.6102	24.9274	23.0813	24.7677	22.9316	1.8361	8.0069	417307.70	1005261.810
36	601633.SH	20220505	24.2488	25.1170	23.9794	24.7079	24.7677	-0.0598	-0.2414	376912.00	927626.444
35	601633.SH	20220506	23.8996	24.0093	23.6102	23.7599	24.7079	-0.9480	-3.8368	227978.49	544080.521
34	601633.SH	20220509	23.4505	23.8197	23.0214	23.2809	23.7599	-0.4790	-2.0160	167211.01	390822.373
33	601633.SH	20220510	22.4526	23.1711	22.1533	23.0015	23.2809	-0.2794	-1.2001	252818.28	574889.646
32	601633.SH	20220511	23.0015	25.3066	23.0015	25.3066	23.0015	2.3051	10.0215	312912.35	770334.450
31	601633.SH	20220512	25.7457	27.5219	25.3964	26.8234	25.3066	1.5168	5.9937	1049395.34	2777063.170
30	601633.SH	20220513	26.4941	29.5077	26.2446	29.5077	26.8234	2.6843	10.0073	1133286.69	3169086.473
29	601633.SH	20220516	30.3160	30.3160	28.5198	28.9389	29.5077	-0.5688	-1.9276	929327.24	2710805.031
28	601633.SH	20220517	28.6995	31.8328	28.5597	31.8328	28.9389	2.8939	10.0000	857405.12	2633159.103
27	601633.SH	20220518	31.9326	35.0161	31.1842	33.5991	31.8328	1.7663	5.5487	1370972.18	4593804.484
26	601633.SH	20220519	32.8108	36.9221	32.4615	35.6847	33.5991	2.0856	6.2073	1127528.77	3805122.945
25	601633.SH	20220520	34.9045	35.8443	34.4146	34.5245	35.6844	-1.1599	-3.2504	866565.33	3026872.497
24	601633.SH	20220523	34.3846	35.6144	33.1548	34.0046	34.5245	-0.5199	-1.5059	1033487.10	3530852.856
23	601633.SH	20220524	35.1944	35.9643	33.6747	34.3646	34.0046	0.3600	1.0587	1154648.95	4062426.732
22	601633.SH	20220525	33.5247	37.6940	32.7948	37.0941	34.3646	2.7295	7.9428	1149982.27	4056564.929
21	601633.SH	20220526	36.1143	36.5942	35.0145	35.2244	37.0941	-1.8697	-5.0404	843588.02	3010855.738
20	601633.SH	20220527	35.0245	36.6542	34.0346	34.8445	35.2244	-0.3799	-1.0785	778321.46	2748631.008
19	601633.SH	20220530	34.9200	35.3000	33.5000	34.0200	34.8500	-0.8300	-2.3816	685725.84	2339470.648
18	601633.SH	20220531	34.0300	34.0400	32.1400	33.2000	34.0200	-0.8200	-2.4103	763013.82	2518375.766
17	601633.SH	20220601	34.0000	34.8000	33.6900	33.7900	33.2000	0.5900	1.7771	766761.94	2619704.685
16	601633.SH	20220602	33.2400	36.4900	33.2200	35.4700	33.7900	1.6800	4.9719	907016.54	3196874.468
15	601633.SH	20220606	34.8200	37.2400	34.8000	36.3600	35.4700	0.8900	2.5092	891169.06	3198770.206
14	601633.SH	20220607	36.0000	36.1600	34.5800	35.0900	36.3600	-1.2700	-3.4928	579871.20	2045914.414
13	601633.SH	20220608	35.0500	37.5000	34.8500	37.3900	35.0900	2.3000	6.5546	873766.34	3183634.937
12	601633.SH	20220609	37.2500	37.2500	35.5500	35.5800	37.3900	-1.8100	-4.8409	638815.97	2302870.364
11	601633.SH	20220610	35.1800	37.2000	34.9500	36.5200	35.5800	0.9400	2.6419	720066.72	2617492.454
10	601633.SH	20220613	35.9000	38.9700	35.5000	38.4400	36.5200	1.9200	5.2574	864491.14	3286654.568
9	601633.SH	20220614	37.6000	41.9900	36.9900	41.2300	38.4400	2.7900	7.2581	923584.90	3606184.652
8	601633.SH	20220615	41.0000	44.5200	39.4000	39.4400	41.2300	-1.7900	-4.3415	1122114.77	4677419.539
7	601633.SH	20220616	39.1400	40.2600	38.3600	38.6100	39.4400	-0.8300	-2.1045	591751.33	2318071.822
6	601633.SH	20220617	38.1100	39.7900	38.0000	39.1000	38.6100	0.4900	1.2691	506979.99	1981230.766
5	601633.SH	20220620	39.1000	39.7100	38.4000	38.9400	39.1000	-0.1600	-0.4092	410093.72	1600731.789
4	601633.SH	20220621	38.7600	38.9200	37.0800	38.3200	38.9400	-0.6200	-1.5922	514513.57	1956043.991
3	601633.SH	20220622	38.3100	39.2800	37.8900	37.9900	38.3200	-0.3300	-0.8612	382944.07	1475721.830
2	601633.SH	20220623	38.5000	40.4900	38.5000	40.4100	37.9900	2.4200	6.3701	737531.29	2922807.528
1	601633.SH	20220624	40.8000	41.1900	39.4900	39.9100	40.4100	-0.5000	-1.2373	594487.88	2390130.687
0	601633.SH	20220627	39.9500	42.3300	39.0600	40.9500	39.9100	1.0400	2.6059	722113.37	2927615.901

# 计算收益率并且合并数据
df_byd['r_byd']=(df_byd['close'] - df_byd['close'].shift(1)) / df_byd['close'].shift(1)
df_ca['r_ca']=(df_ca['close'] - df_ca['close'].shift(1)) / df_ca['close'].shift(1)
df_cc['r_cc']=(df_cc['close'] - df_cc['close'].shift(1)) / df_cc['close'].shift(1)
df_byd=df_byd.dropna()
df_ca=df_ca.dropna()
df_cc=df_cc.dropna()

df_portfolio=pd.DataFrame([df_byd['r_byd'],df_ca['r_ca'],df_cc['r_cc']]).T
df_portfolio.head()

	r_byd	r_ca	r_cc
38	-0.012893	0.039289	0.007009
37	0.047504	0.099801	0.080069
36	0.023868	0.100282	-0.002414
35	-0.027010	0.022572	-0.038368
34	-0.038624	-0.033964	-0.020160

min(df_portfolio.r_byd)

-0.04810298102981026

import scipy.optimize as sco

# 投资组合
rets=df_portfolio

n=len(rets.columns)

# 优化的约束条件：资金的权重和为1，下面表示np.sum(x)-1 eq 0
cons=({'type':'eq','fun':lambda x:np.sum(x)-1})
 #参数取值范围
bnds=tuple((0,1) for x in range(n))

# 生成初始权重
w_initial=n*[1./n,]
# 计算最有权重
opts_sharpe=sco.minimize(min_func_sharpe,w_initial,method='SLSQP',bounds=bnds,constraints=cons)
opts_sharpe

     fun: -9.70554120401185
     jac: array([-6.16312027e-05,  2.01463699e-05,  1.61020207e+00])
 message: 'Optimization terminated successfully'
    nfev: 25
     nit: 6
    njev: 6
  status: 0
 success: True
       x: array([0.24684656, 0.75315344, 0.        ])

# fun对应最优夏普比率,注意取负值，x为权重
opts_sharpe['x'].round(2)

array([0.25, 0.75, 0.  ])

# 以上面投资组合计算预期收益率

rets=df_portfolio
portfolio_stat([0.25,0.75,0])
# 得到p_ret,p_vol,p_sr 预期收益率，风险，夏普比率
# 根据结果而言，这年画收益率是真的不高，我还是故意选择的这三个股票在这段时间涨的比较好的。
# 这里我没有检验各个样本是否符合正态分布，可能是组合表现不佳的原因之一，就不多加叙述了

array([5.63126075, 0.58021321, 9.70550256])

教练我想学编程

关注

1
点赞
踩
3

收藏

觉得还不错? 一键收藏
0
评论
学习记录656@python之MPT现代/均值-方差资产组合理论

均值-方差理论的核心思想是同时考察资产组合的预期收益和风险，研究当我们有一系列可选资产的时候，应如何对其配置资金权重，从而可以得到最好的收益风险比，也就是最好的夏普比率。接下来我们来实战一下，根据近一年比亚迪，长安汽车，长城汽车近一年的收益率情况，搭建投资组合，计算最有的投资权重。模拟多个权重的情况下，收益率和风险的变化情况，并作图，这个图就是经典的投资组合有效边界图。接下来求解最优投资组合权重，也就是求解最大的夏普比率下的投资组合权重。以下模拟投资组合的预期收益率和波动情况。随机一个投资组合权重。
复制链接

扫一扫