BUG: Setting closed='neither' for rolling function of Series leads to mean() producing nan values

[Safeturn]

• I have checked that this issue has not already been reported.

• [x ] I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Code Sample, a copy-pastable example

import numpy as np
import pandas as pd

series = pd.Series(np.arange(20), index=pd.date_range("2018-01-01", periods=20, freq="M"))
series.rolling(window=5, closed='neither').mean()

Problem description

In the current version of pandas, the call to mean() produces a series of nan-values. For pandas 1.1.5, the output was different providing valid values starting from the fifth entry of the series.

Expected Output

2018-01-31 NaN
2018-02-28 NaN
2018-03-31 NaN
2018-04-30 NaN
2018-05-31 2.0
2018-06-30 3.0
2018-07-31 4.0
2018-08-31 5.0
2018-09-30 6.0
2018-10-31 7.0
2018-11-30 8.0
2018-12-31 9.0
2019-01-31 10.0
2019-02-28 11.0
2019-03-31 12.0
2019-04-30 13.0
2019-05-31 14.0
2019-06-30 15.0
2019-07-31 16.0
2019-08-31 17.0
Freq: M, dtype: float64

Output of pd.show_versions()

INSTALLED VERSIONS

commit : 3e89b4c
python : 3.8.5.final.0
python-bits : 64
OS : Linux
OS-release : 4.19.128-microsoft-standard
Version : #1 SMP Tue Jun 23 12:58:10 UTC 2020
machine : x86_64
processor :
byteorder : little
LC_ALL : None
LANG : C.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.2.0
numpy : 1.19.5
pytz : 2020.5
dateutil : 2.8.1
pip : 20.2.3
setuptools : 50.3.0
Cython : 0.29.17
pytest : 6.0.2
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.18.1
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.3.3
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyxlsb : None
s3fs : None
scipy : 1.6.0
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[phofl]

Hi, thanks for your report. In 1.1.4 the closed parameter was ignored for fixed windows and it defaulted to right. The behavior you are showing is for closed=right. I thinkt your result is correct, since left and right both are not in the window so nan everywhere with min_periods=window size, the min-periods parameter needs to be 4 I think.

At least this behaves like that for time aware windows in 1.1.5

@mroeschke Please correct me, if I am wrong.

[mroeschke]

@phofl is correct, xref #34315,

In this case when using neither and a fixed window length, you would need to adjust the min_periods parameter as mentioned.