Apply function on rolling basis across pandas dataframe

Hi, I need to pass a function (HurstEXP), the Hurst Exponent, to every column in a dataframe on a rolling row basis and assign the values to a new DataFrame.

The HE function I use is below:

#Hurst is essentially a measure of the memory in a particular time series, 
#and this memory can be both mean-reverting and trending at the same time, depending on the time scale (no. of lags)

num_lags = 20

def HurstEXP(ts): 
        if isinstance(ts, np.ndarray):
            ts = ts.tolist()
        pass;
        lags = range( 2, num_lags )                                                              # Create the range of lag values
        tau = [std(subtract(ts[lag:], ts[:-lag])) for lag in lags]   # Calculate the array of the variances of the lagged differences
        m = np.polyfit(np.log(lags), np.log(tau), 1)
        return (m[0])

The issue i am struggling with is how to then apply this to every column in a dataframe with my timeseries data. E.g. 5Year daily pricing data for Tesla & Google;

import pandas as pd
import yfinance as yf
import numpy as np
from datetime import datetime
from dateutil.relativedelta import relativedelta
from numpy import log, polyfit, sqrt, std, subtract 

years = 5
today = datetime.today().strftime('%Y-%m-%d')
lastyeartoday = (datetime.today() - relativedelta(years=years)).strftime('%Y-%m-%d')

df = yf.download(['TSLA', 'GOOG'],
                      start=lastyeartoday, 
                      end=today, 
                      progress=False)
df = df.dropna()
df = df[[u'Close']]
df

So far so good... however now when i try to apply my function it gives NaN values across the df:

df.rolling(20).apply(lambda x: HurstEXP(ts = x), raw=True)

I think the issue is to do with the dtype so i tried converting my df to an ndarray through df.to_numpy() but then i cannot apply the rolling functions:

AttributeError: 'numpy.ndarray' object has no attribute 'rolling'

Please help, been struggling with this for a few days now, very green when it comes to Python! Many thanks