I am trying to create a Beta for stocks on 756 days, calculating the Covariance of the stocks Weighted on an index and divided by the Variance of the index.
When i run it for a DF with a single stock, without the groupby arguments, it runs and creates the Beta Column, but i need to do it for all the stocks on my DF and the Groupby was a solution that i read here on Stack Overflow, but i am not able to use it yet.
This is the line that causes the error
df2['Beta 756d'] = df2.groupby('CODNEG').apply(df2['Retorno_Ação'].rolling(window=756,center=False).cov(df2['Retorno_Ibov']) / df2['Retorno_Ibov'].rolling(window=756,center=False).var())
--
This error comes up when the code gets to the line above
TypeError: 'Series' objects are mutable, thus they cannot be hashed
--
Here is an example of df2
TIPREG DATE CODBDI CODNEG TPMERC NOMRES ESPECI PRAZOT MODREF PREABE PREMAX PREMIN PREMED PREULT PREOFC PREOFV TOTNEG QUATOT VOLTOT PREEXE INDOPC DATVEN FATCOT PTOEXE CODISI DISMES Retorno_Ação Retorno_Ibov
0 1 1995-01-02 2 ACE 3 10 ACESITA ON *INT R$ 6300 6300 6300 6300 6300 6300 6500 1 200000 1260000 0 0 99991231 1000 0 ACESACON 119 NaN NaN
105 1 1995-01-02 2 PET 3 10 PETROBRAS ON * R$ 6400 6400 6250 6287 6250 6250 6750 2 40000 251500 0 0 99991231 1000 0 PETRACON 132 NaN NaN
106 1 1995-01-02 2 PET 4 10 PETROBRAS PN * R$ 10700 11000 10400 10599 10500 10500 10650 234 31210000 330805170 0 0 99991231 1000 0 PETRACPN 133 NaN NaN
107 1 1995-01-02 2 BRD 4 10 PETROBRAS BR PN * R$ 4600 4600 4540 4591 4540 4333 4500 13 18200000 83566000 0 0 99991231 1000 0 BRDTACPN 102 NaN NaN
108 1 1995-01-02 2 PTN 4 10 PETTENATI PN * R$ 5189 5189 5189 5189 5189 4700 5280 1 5000000 25945000 0 0 99991231 1000 0 BRPTNTACNPR3 21 NaN NaN
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
1826575 1 2020-08-19 2 HGTX3 10 CIA HERING ON NM R$ 1576 1598 1547 1575 1575 1574 1575 14582 4640200 7309217700 0 0 99991231 1 0 BRHGTXACNOR9 151 0.005105 0.000788
1826576 1 2020-08-19 2 HOME34 10 HOME DEPOT DRN R$ 77798 78612 77798 78413 78612 55000 0 2 1720 134870760 0 0 99991231 1 0 BRHOMEBDR002 133 0.005719 0.000788
1826577 1 2020-08-19 2 HONB34 10 HONEYWELL DRN ED R$ 86347 86347 86347 86347 86347 0 0 1 100 8634700 0 0 99991231 1 0 BRHONBBDR006 127 0.085375 0.000788
1826579 1 2020-08-19 2 HYPE3 10 HYPERA ON NM R$ 3248 3265 3167 3199 3175 3175 3177 15332 3082700 9864018200 0 0 99991231 1 0 BRHYPEACNOR0 122 -0.034367 0.000788
1826517 1 2020-08-19 2 EMAE4 10 EMAE PN R$ 2869 2954 2831 2888 2952 2832 2953 10 1100 3177600 0 0 99991231 1 0 BREMAEACNPR1 114 0.040903 0.000788
I am using the last two columns (Retorno_Ação, Retorno_Ibov) to calculate the Cov and the Var to generate the Beta.
Can anyone tell me what is causing the error?
This line works fine when the DF has only one stock:
df2['Beta 756d'] = df2['Retorno_Ação'].rolling(window=756,center=False).cov(df2['Retorno_Ibov']) / df2['Retorno_Ibov'].rolling(window=756,center=False).var()
--
The error happens when i use df2['Beta 756d'] = df2.groupby('CODNEG').apply(
