Simple fix

A quick fix I found is to rename the columns:

data.index.name = 'Date'
data.columns = ['Adj Close', 'Close', 'High', 'Low', 'Open', 'Volume']

I can't reproduce integer index - is that the problem?

>>> yf.download('SPY', session=session)
[*********************100%***********************]  1 of 1 completed
Price                       Adj Close       Close        High         Low        Open    Volume
Ticker                            SPY         SPY         SPY         SPY         SPY       SPY
Date                                                                                           
1993-01-29 00:00:00+00:00   24.608625   43.937500   43.968750   43.750000   43.968750   1003200
...

Integer index wasnt the problem for me, it was the ticker index. his PR solves the issue.

If you are only fetching one symbol, why not use yf.Ticker('SPY').history()? @antoniouaa
@cyrom0

The output is different from 0.2.47 to 0.2.48, I have downgraded to 0.2.47 and the output looks good. This is the output of 0.2.47.

data = yf.download('SPY')
[100%**] 1 of 1 completed
data.head()
Adj Close Close High Low Open Volume
Date
1993-01-29 00:00:00+00:00 24.608624 43.93750 43.96875 43.75000 43.96875 1003200
1993-02-01 00:00:00+00:00 24.783661 44.25000 44.25000 43.96875 43.96875 480500
1993-02-02 00:00:00+00:00 24.836142 44.34375 44.37500 44.12500 44.21875 201300
1993-02-03 00:00:00+00:00 25.098690 44.81250 44.84375 44.37500 44.40625 529400
1993-02-04 00:00:00+00:00 25.203718 45.00000 45.09375 44.46875 44.96875 531500

As you can see, the header/index from 0.2.48 is different with:
Price Adj Close Close High Low Open Volume
Ticker SPY SPY SPY SPY SPY SPY
Date

If you are only fetching one symbol, why not use yf.Ticker('SPY').history()? @antoniouaa @cyrom0

I wasn't aware that I can do the same using yf.Ticker('SPY').history(). I upgraded to 0.2.49 and sees that
data = yf.Ticker('SPY').history(period="max", interval='1d', actions=False, auto_adjust=False) and
data.index = pd.to_datetime(data.index, utc=True)
will do the equivalent output as yf.download('SPY') for my purposes.

Tracing the code a bit and based on my limited understand, yf.download("SPY") has wrapper to handle multiple tickers so yf.Ticker('SPY').history() looks like is more efficient for one ticker case. I will modify my code to use yf.Ticker('SPY').history().
Basically, there are some additional processing in yf.download('SPY'), such pd.to_datetime(data.index, utc=True) compares to yf.Ticker('SPY').history(). The output of yf.Ticker('SPY').history() will not be able to treated the index as datetime without calling to pd.to_datetime(data.index, utc=True).

Thanks for your help.