UPDATE a
   SET a.DOC2 = b.DOC2
FROM myTable a
 CROSS APPLY(SELECT TOP(1) b.DOC2 FROM myTable b WHERE b.DOC = a.DOC ORDER BY b.VERSION) b

I think this is an equivalent query
;WITH cte as
(
    SELECT tkt.OrderHeader,a.*,sum(value) OVER (partition by tkt.orderheader,date) as Profit
      FROM dbo.[OrderDetails] ...

USE Test
GO
DECLARE @p1 AS INT = 9, @p2 AS INT = 2;
SELECT CAST(@p1 AS NUMERIC(12, 2))/ CAST(@p2 AS NUMERIC(12, 2)) ReturnResult
INTO dbo.TestDivisionReturnType;
GO
EXEC sp_help 'dbo.TestDivisionReturnType'
GO
DROP TABLE dbo.TestDivisionReturnType

Ideally you would have a primary key on your table, an identity column or somthing similar so you won't find the same row overlapping with itself:
select t1.enrollid,
   ...

Your query looks ok, exists and not exists are fast methods of filtering.
A few of points:
 I would give every table an alias and use it on every...