Alphalens parameter timezone mismatch issue

Technical Help
#1

I have been trying to debug an error resulting from using the Alphalens module in zipline

The issue seems to be that for some reason the timezone attribute of the parameters to alphalens do not match even though the tests beforehand show that they do!

here is an example. The df at the top is just an output of a pipeline call.

Any help is appreciated. I have tried everything on my end!


data = df.copy(deep=True)

data.index.set_levels(pd.to_datetime(data.index.get_level_values(0),  utc=True).values, level=0, inplace=True)


print(data[['MLfactor']].head())
print(data[['yesterday_close']].head())
print()

my_factor = data.tz_localize('UTC', level=0,).tz_convert(timezone('UTC'), level=0)[['MLfactor']]
closes = data.tz_localize('UTC', level=0,).tz_convert(timezone('UTC'), level=0)[['yesterday_close']]

print(my_factor.index.levels[0].tz, closes.index.levels[0].tz)
print(my_factor.index.levels[0].tz == closes.index.levels[0].tz)
factor_data = al.utils.get_clean_factor_and_forward_returns(factor=my_factor,
                                                            prices=closes,
                                                           # groupby=sectors,
                                                            #groupby_labels=MORNINGSTAR_SECTOR_CODES,
                                                            #periods=(1,5,10),
                                                           max_loss=10000,
                                                           )

output


                                         MLfactor
2021-01-04 Equity(FIBBG000C2V3D6 [A])     0.139287
           Equity(FIBBG00B3T3HD3 [AA])    0.150880
           Equity(FIBBG005P7Q881 [AAL])   0.127909
           Equity(FIBBG00WCNDCZ6 [AAN])   0.137289
           Equity(FIBBG000C2LZP3 [AAON])  0.132719
                                          yesterday_close
2021-01-04 Equity(FIBBG000C2V3D6 [A])             117.725
           Equity(FIBBG00B3T3HD3 [AA])             23.150
           Equity(FIBBG005P7Q881 [AAL])            16.052
           Equity(FIBBG00WCNDCZ6 [AAN])            17.820
           Equity(FIBBG000C2LZP3 [AAON])           66.675

UTC UTC
True
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-340-76baf73d515f> in <module>()
     19                                                             #groupby_labels=MORNINGSTAR_SECTOR_CODES,
     20                                                             #periods=(1,5,10),
---> 21                                                            max_loss=10000,
     22                                                            )
     23 

/opt/conda/envs/zipline/lib/python3.6/site-packages/alphalens/utils.py in get_clean_factor_and_forward_returns(factor, prices, groupby, binning_by_group, quantiles, bins, periods, filter_zscore, groupby_labels, max_loss, zero_aware, cumulative_returns)
    838         periods,
    839         filter_zscore,
--> 840         cumulative_returns,
    841     )
    842 

/opt/conda/envs/zipline/lib/python3.6/site-packages/alphalens/utils.py in compute_forward_returns(factor, prices, periods, filter_zscore, cumulative_returns)
    261 
    262     factor_dateindex = factor.index.levels[0]
--> 263     if factor_dateindex.tz != prices.index.tz:
    264         raise NonMatchingTimezoneError("The timezone of 'factor' is not the "
    265                                        "same as the timezone of 'prices'. See "

AttributeError: 'MultiIndex' object has no attribute 'tz'
#2

prices should not have a multiindex. Why not follow the Alphalens convention used in the docs? Your approach will have a lookahead bias anyway.

(Note that the forum is mainly for addressing our bugs not yours. Debugging user code is outside of the scope of what we can offer.)