SQL Server 2017

By using db<>fiddle, you agree to license everything you submit by Creative Commons CC0.

CREATE TABLE T
(
  id int,
  value int,
  acc_no int
);

insert into t values 
(1     ,12     ,1), 
(2     ,14     ,1), 
(3     ,15     ,1), 
(4     ,10     ,2), 
(5     ,16     ,2), 
(6     ,19     ,1), 
(7     , 7     ,3), 
(8     ,24     ,2);
8 rows affected

SELECT *,
       Value - LAG(Value, 1, 0) OVER(PARTITION BY acc_no ORDER BY ID) Result
FROM T
ORDER BY ID


id
value
acc_no
Result


1
12
1
12

2
14
1
2

3
15
1
1

4
10
2
10

5
16
2
6

6
19
1
4

7
7
3
7

8
24
2
8


WITH CTE AS
(
  SELECT *,
         ROW_NUMBER() OVER(PARTITION BY acc_no ORDER BY id) RN
  FROM T
)
SELECT T1.id,
       T1.value,
       T1.acc_no,
       T1.value - ISNULL(T2.value, 0) Result
FROM CTE T1 LEFT JOIN CTE T2
     ON T1.acc_no = T2.acc_no
        AND 
        T1.RN = T2.RN + 1
ORDER BY T1.id;


id
value
acc_no
Result


1
12
1
12

2
14
1
2

3
15
1
1

4
10
2
10

5
16
2
6

6
19
1
4

7
7
3
7

8
24
2
8