You want the .name attribute and pass this to get_loc:
In [131]:
dates = pd.date_range('1/1/2000', periods=8)
df = pd.DataFrame(np.random.randn(8, 4), index=dates, columns=['A', 'B', 'C', 'D'])
df
Out[131]:
A B C D
2000-01-01 0.095234 -1.000863 0.899732 -1.742152
2000-01-02 -0.517544 -1.274137 1.734024 -1.369487
2000-01-03 0.134112 1.964386 -0.120282 0.573676
2000-01-04 -0.737499 -0.581444 0.528500 -0.737697
2000-01-05 -1.777800 0.795093 0.120681 0.524045
2000-01-06 -0.048432 -0.751365 -0.760417 -0.181658
2000-01-07 -0.570800 0.248608 -1.428998 -0.662014
2000-01-08 -0.147326 0.717392 3.138620 1.208639
In [133]:
window_stop_row = df[df.index < '2000-01-04'].iloc[-1]
window_stop_row.name
Out[133]:
Timestamp('2000-01-03 00:00:00', offset="D")
In [134]:
df.index.get_loc(window_stop_row.name)
Out[134]:
2
get_loc returns the ordinal position of the label in your index which is what you want:
In [135]:
df.iloc[df.index.get_loc(window_stop_row.name)]
Out[135]:
A 0.134112
B 1.964386
C -0.120282
D 0.573676
Name: 2000-01-03 00:00:00, dtype: float64
if you just want to search the index then so long as it is sorted then you can use searchsorted:
In [142]:
df.index.searchsorted('2000-01-04') - 1
Out[142]:
2