How to return a column value based on integer index and another column value in pandas? - pandas

Question

I have a df that I want to groupby Date and apply a function to it.
Date Symbol Shares
0 1990-01-01 A 0.0
1 1990-01-01 B 0.0
2 1990-01-01 C 0.0
3 1990-01-01 D 0.0
4 1990-01-02 A 50.0
5 1990-01-02 B 100.0
6 1990-01-02 C 66.0
7 1990-01-02 D 7.0
8 1990-01-03 A 11.0
9 1990-01-03 B 123.0
10 1990-01-03 C 11.0
11 1990-01-03 D 11.0
I should be able to access the Shares value for a Symbol from the previous Date in the function. How can I do that? Creating df[prev_shares] like df.groupby('Symbol')['Shares'].shift(1) before applying the function is not an option because the function calculates Shares row by row. It should look like:
def calcs(x):
x.loc[some_condition, 'Shares'] = ...
x.loc[other_condition, 'Shares'] = # return 'Shares' from previous 'Date' for this 'Symbol'
df = df.groupby('Date').apply(calcs)
Any help appreciated.
EDIT:
I post the function that I created.
Equity = 10000
def calcs(x):
global Equity
if x.index[0]==0: return x
x.loc[x['condition'] == True, 'Shares'] = np.floor((Equity * 0.02 / x['ATR']).astype(float))
x.loc[x['condition'] == False, 'Shares'] = # locate Symbol for the previous Date and return its Shares value
x['Closed_P/L'] = x['Shares'] * x['Close']
Equity += x['Closed_P/L'].sum()
return x
data = data.groupby('Date').apply(calcs)

Answer 1

IIUC something like this might help, although what you're trying to accomplish in the end is unclear. Below you will find the data grouped by 'Symbol' and 'Date'. Column 'diff' shows the incremental P/L.
please advise what you need additionally.
df = data.set_index(['Symbol', 'Date']).sort_index()[['Shares']]
df['diff'] = np.nan
idx = pd.IndexSlice
for ix in df.index.levels[0]:
df.loc[ idx[ix,:], 'diff'] = df.loc[idx[ix,:], 'Shares' ].diff()
df.fillna(0)
Shares diff
Symbol Date
A 1990-01-01 0.0 0.0
1990-01-02 50.0 50.0
1990-01-03 11.0 -39.0
B 1990-01-01 0.0 0.0
1990-01-02 100.0 100.0
1990-01-03 123.0 23.0
C 1990-01-01 0.0 0.0
1990-01-02 66.0 66.0
1990-01-03 11.0 -55.0
D 1990-01-01 0.0 0.0
1990-01-02 7.0 7.0
1990-01-03 11.0 4.0