BUG: Incorrect handling of rolling.cov with offset window

Author: kristopheryahoo

pandas/core/window.py

@@ -81,6 +81,7 @@ def __init__(self, obj, window=None, min_periods=None, freq=None,
         self.freq = freq
         self.center = center
         self.win_type = win_type
+        self.win_freq = None
         self.axis = obj._get_axis_number(axis) if axis is not None else None
         self.validate()
 
@@ -996,7 +997,10 @@ def cov(self, other=None, pairwise=None, ddof=1, **kwargs):
             # only default unset
             pairwise = True if pairwise is None else pairwise
         other = self._shallow_copy(other)
-        window = self._get_window(other)
+        if self.is_freq_type:
+            window = self.win_freq
+        else:
+            window = self._get_window(other)
 
         def _get_cov(X, Y):
             # GH #12373 : rolling functions error on float32 data
@@ -1088,6 +1092,7 @@ def validate(self):
                                           "based windows")
 
             # this will raise ValueError on non-fixed freqs
+            self.win_freq = self.window
             self.window = freq.nanos
             self.win_type = 'freq'
 

pandas/tests/test_window.py

@@ -3396,6 +3396,20 @@ def test_min_periods(self):
         result = df.rolling('2s', min_periods=1).sum()
         tm.assert_frame_equal(result, expected)
 
+    def test_cov(self):
+
+        # xref GH16058
+
+        df = self.regular
+
+        result = df.rolling('2s', min_periods=1).cov()
+        expected = df.rolling(2, min_periods=1).cov()
+        tm.assert_frame_equal(result, expected)
+
+        result = df.rolling('3s', min_periods=1).cov()
+        expected = df.rolling(3, min_periods=1).cov()
+        tm.assert_frame_equal(result, expected)
+
     def test_closed(self):
 
         # xref GH13965