Demo, Sql Server - rextester

Language: Layout:

CREATE TABLE inventory
    (
      timestamp  DATETIME2 DEFAULT (getdate()) NOT NULL
    , Prod_id    INT NOT NULL
    , Loc_ID     nvarchar(3)
    , Price      INT NOT NULL
    , Buy      INT NOT NULL
    , Sell      INT NOT NULL
    )

INSERT INTO inventory
VALUES  ('2017-02-21 12:24:00',50,'A',10,1,0)

INSERT INTO inventory
VALUES  ('2017-02-22 13:15:00',50,'A',10,2,0)

INSERT INTO inventory
VALUES  ('2017-02-23 14:00:00',50,'A',12,0,2)

INSERT INTO inventory
VALUES  ('2017-02-24 12:20:00',55,'B',2,1,0)

INSERT INTO inventory
VALUES  ('2017-02-25 10:04:00',55,'B',2,1,0)

INSERT INTO inventory
VALUES  ('2017-02-26 11:44:00',55,'B',5,0,3)

INSERT INTO inventory
VALUES  ('2017-02-27 15:22:00',60,'C',3,5,0)

INSERT INTO inventory
VALUES  ('2017-02-28 16:24:00',60,'C',4,0,5)

select case when sum(buy) > sum(sell) then max(case when buy > 0 then timestamp end) else max(case when sell > 0 then timestamp end) end as timestamp,
	prod_id,
	Loc_ID,
	case when sum(buy) > sum(sell) then max(case when buy > 0 then price end) else max(case when sell > 0 then price end) end as price,
	case when sum(buy) > sum(sell) then sum(buy) - sum(sell) else 0 end as buy,
	case when sum(sell) > sum(buy) then sum(sell) - sum(buy) else 0 end as sell
from dbo.inventory
group by prod_id,
	Loc_ID
having sum(buy) - sum(sell) <> 0;

View schema
Execution time: 0,02 sec, rows selected: 2, rows affected: 8, absolute service time: 0,16 sec