Question

I have a table with following data:

Id  Name    Value
1   John    100
2   John    -500
3   John    500
4   Smith   10
5   Smith   20
6   Smith   -20
7   Stuart  -10
8   Wills   25

I am looking for an efficient TSQL query which can remove John -500 and Smith -20 (i.e. records with negative value if they have a similar positive value in the same group [group by names]).

Answer 1

I think this is what you need. (SQL DEMO)

delete y
from mytable y join (
  select id,name, value
  from mytable x
  where value > 0) z on y.name = z.name and y.value = -1 * z.value 

select * from mytable
--SELECT RESULTS AFTER DELETING
ID  NAME    VALUE
1   John    100
3   John    500
4   Smith   10
5   Smith   20
7   Stuart  -10
8   Wills   25