daveskis
Reputation: 153
Python - Portfolio Open Position Flag
I have a dataframe below:
ticker date buy_fg sell_fg
50 ABC 2010-01-01 0.0 2.0
51 ABC 2010-01-08 0.0 2.0
52 ABC 2010-01-15 0.0 2.0
53 ABC 2010-01-22 0.0 2.0
1512 ABC 2010-01-29 1.0 0.0
54 ABC 2010-02-05 0.0 2.0
55 ABC 2010-02-12 0.0 2.0
56 ABC 2010-02-19 0.0 2.0
57 ABC 2010-02-26 0.0 2.0
58 ABC 2010-03-05 0.0 2.0
59 ABC 2010-03-12 0.0 2.0
60 ABC 2010-03-19 0.0 2.0
61 ABC 2010-03-26 0.0 2.0
1417 ABC 2010-04-02 1.0 0.0
1409 ABC 2010-04-09 1.0 0.0
62 ABC 2010-04-23 0.0 2.0
63 ABC 2010-05-07 0.0 2.0
64 ABC 2010-05-14 0.0 2.0
65 ABC 2010-05-21 0.0 2.0
66 ABC 2010-05-28 0.0 2.0
I am trying to develop a column that provides a summary of the action that my script needs to take. I have tried to develop a cumulative score approach and haven't been able to get it to work. I have also tried lambda, however the logic isn't right as it just repeats the buy_fg column.
# Initiate the global, default buy flag
hold_flag = False
# Custom function to flick the global flag based on `buy_fg` column
def hold(value):
global hold_flag
if value == 1:
if hold_flag is False:
hold_flag = True
return '3'
if hold_flag is True:
hold_flag = False
return '1'
else:
return '0'
new_report['hold_fg'] = new_report['buy_fg'].apply(hold)
This created hold_fg column. But now I need to expand upon this to consider all three *_fg columns to provide the final action.
ticker date buy_fg sell_fg action hold_fg
50 ABC 2010-01-01 0.0 2.0 0 0
51 ABC 2010-01-08 0.0 2.0 0 0
52 ABC 2010-01-15 0.0 2.0 0 0
53 ABC 2010-01-22 0.0 2.0 0 0
1512 ABC 2010-01-29 1.0 0.0 0 1
54 ABC 2010-02-05 0.0 2.0 0 0
55 ABC 2010-02-12 0.0 2.0 0 0
56 ABC 2010-02-19 0.0 2.0 0 0
57 ABC 2010-02-26 0.0 2.0 0 0
58 ABC 2010-03-05 0.0 2.0 0 0
59 ABC 2010-03-12 0.0 2.0 0 0
60 ABC 2010-03-19 0.0 2.0 0 0
61 ABC 2010-03-26 0.0 2.0 0 0
1417 ABC 2010-04-02 1.0 0.0 0 1
1409 ABC 2010-04-09 1.0 0.0 0 3
62 ABC 2010-04-23 0.0 2.0 0 0
63 ABC 2010-05-07 0.0 2.0 0 0
64 ABC 2010-05-14 0.0 2.0 0 0
65 ABC 2010-05-21 0.0 2.0 0 0
66 ABC 2010-05-28 0.0 2.0 0 0
I have tried the following, but this doesn't work:
new_report['action'] = 0
buy_flag = False
def action(value):
global buy_flag
if new_report['hold_fg'] == 1:
buy_flag = True
return 'BUY'
if new_report['hold_fg'] == 3:
buy_flag = True
return 'HOLD'
if new_report['sell_fg'] == 2:
buy_flag = False
return 'SELL'
else:
return '0'
Any ideas on how I can solve this?
updated expected output:
ticker date buy_fg sell_fg action hold_fg
50 ABC 2010-01-01 0.0 2.0 NIL 0
51 ABC 2010-01-08 0.0 2.0 NIL 0
52 ABC 2010-01-15 0.0 2.0 NIL 0
53 ABC 2010-01-22 0.0 2.0 NIL 0
1512 ABC 2010-01-29 1.0 0.0 BUY 1
54 ABC 2010-02-05 0.0 2.0 SELL 0
55 ABC 2010-02-12 0.0 2.0 NIL 0
56 ABC 2010-02-19 0.0 2.0 NIL 0
57 ABC 2010-02-26 0.0 2.0 NIL 0
58 ABC 2010-03-05 0.0 2.0 NIL 0
59 ABC 2010-03-12 0.0 2.0 NIL 0
60 ABC 2010-03-19 0.0 2.0 NIL 0
61 ABC 2010-03-26 0.0 2.0 NIL 0
1417 ABC 2010-04-02 1.0 0.0 BUY 1
1409 ABC 2010-04-09 1.0 0.0 HOLD 3
62 ABC 2010-04-23 0.0 2.0 SELL 0
63 ABC 2010-05-07 0.0 2.0 NIL 0
64 ABC 2010-05-14 0.0 2.0 NIL 0
65 ABC 2010-05-21 0.0 2.0 NIL 0
66 ABC 2010-05-28 0.0 2.0 NIL 0
Upvotes: 0
Views: 107
Answers (1)
Danail Petrov
Reputation: 1875
I think you'd need to keep track of the status and apply your required logic based on that. Here is an example using a custom function to do that. Hope this helps!
# Initiate the global, default buy flag
>>> buy_flag = False
# Custom function to flick the global flag based on `buy_fg` column
>>> def sell_or_buy(value):
global buy_flag
if value == 1:
if buy_flag is False:
buy_flag = True
return 'BUY'
if buy_flag is True:
return 'HOLD'
if value == 0:
if buy_flag == 1:
buy_flag = False
return 'SELL'
if buy_flag == 0:
return 'NIL'
>>> df['action'] = df['buy_fg'].apply(sell_or_buy)
>>> df
ticker date buy_fg sell_fg action
0 50 ABC 2010-01-01 0.0 2.0 NIL
1 51 ABC 2010-01-08 0.0 2.0 NIL
2 52 ABC 2010-01-15 0.0 2.0 NIL
3 53 ABC 2010-01-22 0.0 2.0 NIL
4 1512 ABC 2010-01-29 1.0 0.0 BUY
5 54 ABC 2010-02-05 0.0 2.0 SELL
6 55 ABC 2010-02-12 0.0 2.0 NIL
7 56 ABC 2010-02-19 0.0 2.0 NIL
8 57 ABC 2010-02-26 0.0 2.0 NIL
9 58 ABC 2010-03-05 0.0 2.0 NIL
10 59 ABC 2010-03-12 0.0 2.0 NIL
11 60 ABC 2010-03-19 0.0 2.0 NIL
12 61 ABC 2010-03-26 0.0 2.0 NIL
13 1417 ABC 2010-04-02 1.0 0.0 BUY
14 1409 ABC 2010-04-09 1.0 0.0 HOLD
15 62 ABC 2010-04-23 0.0 2.0 SELL
16 63 ABC 2010-05-07 0.0 2.0 NIL
17 64 ABC 2010-05-14 0.0 2.0 NIL
18 65 ABC 2010-05-21 0.0 2.0 NIL
19 66 ABC 2010-05-28 0.0 2.0 NIL
EDIT: This is to accommodate your latest requirement. As it gets a bit messy now with the way you've updated the question, I will just modify my answer to add to the above dataframe.
>>> d={'NIL': 0, 'BUY': 1, 'SELL': 2, 'HOLD' : 3}
>>> df['hold_fg'] = df.action.map(d)
>>> df
ticker date buy_fg sell_fg action hold_fg
0 50 ABC 2010-01-01 0.0 2.0 NIL 0
1 51 ABC 2010-01-08 0.0 2.0 NIL 0
2 52 ABC 2010-01-15 0.0 2.0 NIL 0
3 53 ABC 2010-01-22 0.0 2.0 NIL 0
4 1512 ABC 2010-01-29 1.0 0.0 BUY 1
5 54 ABC 2010-02-05 0.0 2.0 SELL 2
6 55 ABC 2010-02-12 0.0 2.0 NIL 0
7 56 ABC 2010-02-19 0.0 2.0 NIL 0
8 57 ABC 2010-02-26 0.0 2.0 NIL 0
9 58 ABC 2010-03-05 0.0 2.0 NIL 0
10 59 ABC 2010-03-12 0.0 2.0 NIL 0
11 60 ABC 2010-03-19 0.0 2.0 NIL 0
12 61 ABC 2010-03-26 0.0 2.0 NIL 0
13 1417 ABC 2010-04-02 1.0 0.0 BUY 1
14 1409 ABC 2010-04-09 1.0 0.0 HOLD 3
15 62 ABC 2010-04-23 0.0 2.0 SELL 2
16 63 ABC 2010-05-07 0.0 2.0 NIL 0
17 64 ABC 2010-05-14 0.0 2.0 NIL 0
18 65 ABC 2010-05-21 0.0 2.0 NIL 0
19 66 ABC 2010-05-28 0.0 2.0 NIL 0
Upvotes: 1
Related Questions
- Portfolio Turnover with Pandas
- conditional flagging in pandas
- Backtesting Trading Strategy with python and pandas - Recognizing only one open position at a time
- Pandas create flag if any condition fails
- Positional indexing
- Calculate stock position balance
- How to create a simple flag in Python
- Calculate stock returns - when in position
- python pandas portfolio return
- Row flagging in Python