convert kenneth French data to daily datetime format in python

Question

I want to integrate date from Kenneth French's website with the following code, that works fine for monthly data. But since I need now daily data I need to know what I have to put for the variable XYZ (in the last row of code) in order to make it work:

import pandas.io.data
import pandas as pd
from datetime import datetime, date

io = pandas.io.data.DataReader("F-F_Research_Data_Factors_daily", "famafrench")

ff = io[0] # Bestimmung des Dictionary Teils aus der FF Zip Datei
ff.columns = ['Mkt_rf', 'SMB', 'HML', 'rf']    
ff.index = [datetime(d/100, d%100, XYZ) for d in ff.index]

This is how the index looks like:

 Int64Index([19260701, 19260702, 19260706, 19260707, 19260708, 19260709, 19260710, 19260712, 19260713, 19260714, 19260715, 19260716, 19260717, 19260719, 19260720, 19260721, 19260722, 19260723, 19260724, 19260726, 19260727, 19260728, 19260729, 19260730, 19260731, 19260802, 19260803, 19260804, 19260805, 19260806, 19260807, 19260809, 19260810, 19260811, 19260812, 19260813, 19260814, 19260816, 19260817, 19260818, 19260819, 19260820, 19260821, 19260823, 19260824, 19260825, 19260826, 19260827, 19260828, 19260830, 19260831, 19260901, 19260902, 19260903, 19260907, 19260908, 19260909, 19260910, 19260911, 19260913, 19260914, 19260915, 19260916, 19260917, 19260918, 19260920, 19260921, 19260922, 19260923, 19260924, 19260925, 19260927, 19260928, 19260929, 19260930, 19261001, 19261002, 19261004, 19261005, 19261006, 19261007, 19261008, 19261009, 19261011, 19261013, 19261014, 19261015, 19261016, 19261018, 19261019, 19261020, 19261021, 19261022, 19261023, 19261025, 19261026, 19261027, 19261028, 19261029, 19261030, ...], dtype='int64')

And this is how the dataframe looks like:

          Mkt_rf   SMB   HML     rf
19260701    0.10 -0.24 -0.28  0.009   # the integer 19260701 needs to get converted to a datetime copartible format such as 1926/07/01 
...          ...   ...   ...    ...
20150430   -1.11 -1.04  0.73  0.000

[23467 rows x 4 columns]

I use pandas 0.16.1

Any ideas?

Answer 1

To convert your index to a DatetimeIndex, you can use to_datetime:

ff.index = pd.to_datetime(ff.index, format='%Y%m%d')

When having your dates as integers, the above works by specifying format='%Y%m%d' and to_datetime will see the ints as a string of that format. A small example:

In [2]: pd.to_datetime([19260701, 19260702], format='%Y%m%d')
Out[2]: DatetimeIndex(['1926-07-01', '1926-07-02'], dtype='datetime64[ns]', freq=None, tz=None)