Is it possible to align asset performance by trading session in python?

user_6342167 0
Viewed 57

I have time series data in one minute intervals for the last 7 business days and I want to calculate and plot the cumulative change by trading session like this:

BTC returns by trading session

The 23-Jun 'Asia' trading session overlaps two days (i.e. 23:00 on 22-Jun > 07:00 on 23-Jun) and I am unable to align the cumulative change calculation to take into account the 23:00 > 23:59 portion of the trading session which falls in the prior day.

My code:

cl1 = con.bdib('CO1 Comdty', start_datetime=start, 
         end_datetime=end, event_type='TRADE', interval=interval)

cl1['Timestamp'] = pd.to_datetime(cl1.index)
cl1['Time'] = cl1['Timestamp'].dt.time

def bin_f(x):
    if x.time() < datetime.time(7):
        return 'Asia'
    elif x.time() < datetime.time(14):
        return 'Europe'
    elif x.time() < datetime.time(22):
        return 'US'
    else:
        return 'Open'

cl1['Timezone'] = cl1['Timestamp'].apply(bin_f)

perf = cl1[['close', 'Timezone']]
perf['Count'] = 1

perf['Minute'] = perf.groupby([perf.index.date, 'Timezone'])['Count'].cumsum()
perf['Session'] = perf.index.date

perf = perf.pivot(index=['Session', 'Minute'], columns='Timezone', values='close')

perf = perf[['Asia', 'Europe', 'US']]

perf = perf.diff()

perf = perf.cumsum()

Result

Result

Raw Data

Raw Data

Desired output:

Two new columns: TimeZone and Session.

enter image description here

bloomberg trading finance python
2 Answers

You can use xbbg to download intraday prices and use pd.concat([...], axis=1) directly to align the time series. Because the price data from xbbg carries timezone info and pandas deals with alignment automatically.

user_13326560

One solution is to split the problem into two parts:

user_106935960
Related