diff --git a/xarray/core/_reductions.py b/xarray/core/_reductions.py
index d312bd41ec5..b4a67ea43bf 100644
--- a/xarray/core/_reductions.py
+++ b/xarray/core/_reductions.py
@@ -3,7 +3,7 @@
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, Any, Callable, Hashable, Sequence
+from typing import TYPE_CHECKING, Any, Callable, Hashable, Sequence, TypeVar
 
 from . import duck_array_ops
 from .options import OPTIONS
@@ -12,7 +12,9 @@
 if TYPE_CHECKING:
     from .dataarray import DataArray
     from .dataset import Dataset
-    from .types import T_DataArray, T_Dataset
+
+    T_Dataset = TypeVar("T_Dataset", bound="DatasetReductions")
+    T_DataArray = TypeVar("T_DataArray", bound="DataArrayReductions")
 
 try:
     import flox
@@ -24,7 +26,7 @@ class DatasetReductions:
     __slots__ = ()
 
     def reduce(
-        self,
+        self: T_Dataset,
         func: Callable[..., Any],
         dim: None | Hashable | Sequence[Hashable] = None,
         *,
@@ -32,7 +34,7 @@ def reduce(
         keep_attrs: bool | None = None,
         keepdims: bool = False,
         **kwargs: Any,
-    ) -> Dataset:
+    ) -> T_Dataset:
         raise NotImplementedError()
 
     def count(
@@ -1034,7 +1036,7 @@ class DataArrayReductions:
     __slots__ = ()
 
     def reduce(
-        self,
+        self: T_DataArray,
         func: Callable[..., Any],
         dim: None | Hashable | Sequence[Hashable] = None,
         *,
@@ -1042,7 +1044,7 @@ def reduce(
         keep_attrs: bool | None = None,
         keepdims: bool = False,
         **kwargs: Any,
-    ) -> DataArray:
+    ) -> T_DataArray:
         raise NotImplementedError()
 
     def count(
diff --git a/xarray/core/rolling.py b/xarray/core/rolling.py
index aef290f6d7f..9600009aee6 100644
--- a/xarray/core/rolling.py
+++ b/xarray/core/rolling.py
@@ -12,6 +12,7 @@
     Hashable,
     Iterator,
     Mapping,
+    Sequence,
     TypeVar,
 )
 
@@ -468,9 +469,8 @@ def reduce(
             obj, rolling_dim, keep_attrs=keep_attrs, fill_value=fillna
         )
 
-        result = windows.reduce(
-            func, dim=list(rolling_dim.values()), keep_attrs=keep_attrs, **kwargs
-        )
+        dim: Sequence[Hashable] = list(rolling_dim.values())
+        result = windows.reduce(func, dim=dim, keep_attrs=keep_attrs, **kwargs)
 
         # Find valid windows based on count.
         counts = self._counts(keep_attrs=False)
@@ -487,6 +487,7 @@ def _counts(self, keep_attrs: bool | None) -> DataArray:
         # array is faster to be reduced than object array.
         # The use of skipna==False is also faster since it does not need to
         # copy the strided array.
+        dim: Sequence[Hashable] = list(rolling_dim.values())
         counts = (
             self.obj.notnull(keep_attrs=keep_attrs)
             .rolling(
@@ -494,7 +495,7 @@ def _counts(self, keep_attrs: bool | None) -> DataArray:
                 center={d: self.center[i] for i, d in enumerate(self.dim)},
             )
             .construct(rolling_dim, fill_value=False, keep_attrs=keep_attrs)
-            .sum(dim=list(rolling_dim.values()), skipna=False, keep_attrs=keep_attrs)
+            .sum(dim=dim, skipna=False, keep_attrs=keep_attrs)
         )
         return counts
 
diff --git a/xarray/util/generate_reductions.py b/xarray/util/generate_reductions.py
index 63e08f2f570..ad225b55481 100644
--- a/xarray/util/generate_reductions.py
+++ b/xarray/util/generate_reductions.py
@@ -22,7 +22,7 @@
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, Any, Callable, Hashable, Sequence
+from typing import TYPE_CHECKING, Any, Callable, Hashable, Sequence, TypeVar
 
 from . import duck_array_ops
 from .options import OPTIONS
@@ -31,7 +31,9 @@
 if TYPE_CHECKING:
     from .dataarray import DataArray
     from .dataset import Dataset
-    from .types import T_DataArray, T_Dataset
+
+    T_Dataset = TypeVar("T_Dataset", bound="DatasetReductions")
+    T_DataArray = TypeVar("T_DataArray", bound="DataArrayReductions")
 
 try:
     import flox
@@ -44,7 +46,7 @@ class {obj}{cls}Reductions:
     __slots__ = ()
 
     def reduce(
-        self,
+        self{self_type_snippet},
         func: Callable[..., Any],
         dim: None | Hashable | Sequence[Hashable] = None,
         *,
@@ -52,7 +54,7 @@ def reduce(
         keep_attrs: bool | None = None,
         keepdims: bool = False,
         **kwargs: Any,
-    ) -> {obj}:
+    ) -> {return_type}:
         raise NotImplementedError()"""
 
 GROUPBY_PREAMBLE = """
@@ -246,7 +248,13 @@ def __init__(
         self.docref = docref
         self.docref_description = docref_description
         self.example_call_preamble = example_call_preamble
-        self.preamble = definition_preamble.format(obj=datastructure.name, cls=cls)
+        self.common_kwargs = dict(
+            obj=self.datastructure.name,
+            cls=cls,
+            self_type_snippet=": " + self.self_type if self.self_type else "",
+            return_type=self.self_type if self.self_type else self.datastructure.name,
+        )
+        self.preamble = definition_preamble.format(**self.common_kwargs)
         if not see_also_obj:
             self.see_also_obj = self.datastructure.name
         else:
@@ -258,12 +266,7 @@ def generate_methods(self):
             yield self.generate_method(method)
 
     def generate_method(self, method):
-        template_kwargs = dict(
-            obj=self.datastructure.name,
-            method=method.name,
-            self_type_snippet=": " + self.self_type if self.self_type else "",
-            return_type=self.self_type if self.self_type else self.datastructure.name,
-        )
+        template_kwargs = dict(self.common_kwargs, method=method.name)
 
         if method.extra_kwargs:
             extra_kwargs = "\n        " + "\n        ".join(