1

I have just started my first project in python. I have a pandas dataframe of cryptocurrency data with four columns "id", "date", "close", and "volume".

I want to calculate the On Balance Volume (OBV) for each group (id) in my dataframe. My goal is to have a new column OBV which contains the OBV for each group.The formuala for OBV is as follows:

If today's close > yesterday's close then: OBV = Yesterday’s OBV + Today’s Volume; If today’s close < than yesterday’s close then: OBV = Yesterday’s OBV – Today’s Volume; If today’s close = to yesterday’s close then: OBV = Yesterday’s OBV

On Balance Volume is calculated by adding the day's volume to a cumulative total when the security's price closes up, and subtracting the day's volume when the security's price closes down.

My dataframe df looks like this:

|id |last_updated |close  |volume 
|-- |------------ |-----  |------
|1  |2021-10-08   |0.0113 |134.431757         
|1  |2021-10-09   |0.0115 |57.222013     
|1  |2021-10-10   |0.0115 |75.848861    
|1  |2021-10-11   |0.0120 |77.568503      
|1  |2021-10-12   |0.0118 |19.712469 
|2  |2021-10-08   |0.1104 | 0.000000      
|2  |2021-10-09   |0.1632 | 0.176570      
|2  |2021-10-10   |0.1603 | 7.636593     
|2  |2021-10-11   |0.1509 |193.415714     
|2  |2021-10-12   |0.1046 |  0.000000

Since I have a large dateset of cryptocurrency data, I want to run vectorized computation in order to to express batch operations on data without writing any for loops:

df['OBV']=np.where(df['close'] > df['close'].shift(1), df['volume'], np.where(df['close'] < df['close'].shift(1), -df['volume'], 0)).cumsum()

With this approach, however, I only obtain the correct calculation of OBV for id=1. The calculation of the OBV for id=2 uses the last value of OBV from id=1 instead of restarting excecuting the calculations from the first row of id=2.

I have also tried the OBV calculation with a loop but I also do not know how to iterate with the loop through each group, that is to excecute the loop within each group.

OBV=[]
OBV.append(0)
#Loop through the data set (close price)
for i in range(1, len(df.close)):
    if  df.close[i] > df.close[i-1]:
        OBV.append(OBV[-1] + df.vol[i])
    elif df.close[i] < df.close[i-1]:
        OBV.append(OBV[-1] - df.vol[i])
    else:
        OBV.append(OBV[-1])

df['OBV']=OBV

So how do I obtain a new colum OBV in my dataframe df which contains the OBV within each group?

Any help is very much appreciated. Thank you!

Lars
  • 11
  • 1
  • possible duplicate of https://stackoverflow.com/questions/52671594/calculating-stockss-on-balance-volume-obv-in-python?rq=1 – Meitham Dec 01 '21 at 21:12
  • It is similar, yes. But the issue here is not the implementation of OBV per se, but rather how to calculate it per group. How to iterate over groups using numpy.where or the if condition is still a mystery for me having just started with python ... – Lars Dec 01 '21 at 21:39

0 Answers0