Last account per customer group, Sql Server

Language: Layout:

create table mytable (acctnumber int, date date, cust1id int, cust2id int);

insert into mytable values 
(10000, '20160201', 1110, null),
(10001, '20160201', 1111, null),
(10050, '20170201', 1111, null),
(10008, '20160201', 1120, null),
(10058, '20170203', 1120, 1121),
(10002, '20160201', 1112, null),
(10052, '20170202', 1113, 1112),
(10003, '20160202', 1114, 1115),
( 7060, '20170204', 1115, 1114),
(10004, '20160202', 1116, 1117),
(10067, '20170205', 1117, null),
(10005, '20160201', 1118, null),
(10054, '20170203', 1118, 1119),
(10101, '20170602', 1119, null),
(10007, '20160201', 1122, 1123),
(10057, '20170203', 1123, 1124),
(10107, '20170602', 1124, 1125);

with all_cust(custid) as
(
  select cust1id from mytable
  union
  select cust2id from mytable where cust2id is not null
)
, cte(c1, c2, sofar) as
(
  select custid, custid, '<' + cast(custid as varchar(max)) + '>' from all_cust
  union all
  select cte.c1, case when cte.c2 = m.cust1id then m.cust2id else m.cust1id end,
     cte.sofar + '<' + cast(case when cte.c2 = m.cust1id then m.cust2id else m.cust1id end as varchar(max)) + '>'
  from mytable m
  join cte on cte.c2 in (m.cust1id, m.cust2id)
  and cte.sofar not like '%' + cast(case when cte.c2 = m.cust1id then m.cust2id else m.cust1id end as varchar(max)) + '%'
)
, groups(custid, grp) as
(
  select c1, min(c2) from cte group by c1
)
, ranked as
(
  select *, row_number() over (partition by g.grp order by date desc) as rn 
  from groups g
  join mytable m on g.custid in (m.cust1id, m.cust2id)
)
select acctnumber, date, cust1id, cust2id
from ranked
where rn = 1
order by cust1id;

drop table mytable;

View schema
Execution time: 0,03 sec, rows selected: 8, rows affected: 17, absolute service time: 0,19 sec