sql server - Finding difference of two columns from two different rows in an SQL table -

i trying figure out query returns difference of 2 columns belongs 2 different rows of single table

for example have following data in table

id      market  grade       term     bid     offer 1       heavy   abc         jun14   -19.5     -17 2       heavy   abc         jul14   -20      -17.5 3       sour    xyz         jun14   -30       -17 4       sour    xyz         jul14   -32       -27 

now, want following results

(as market , grade same id=1,2)

bid of id=1 - offer of id=2 offer of id=1- bid of id=2 

(as market , grade same id=3,4)

bid of id=3 - offer of id=4 offer of id=3- bid of id=4 

something following

market     term          bid                   offer heavy/abc  jun14/jul14   (-19.5-(-17.5))=-2    (-17-(-20))=3 sour/xyz  jun14/jul14    (-30-(-27))=-3        (-17-(-32))=15 

what vest way ms sql.

using row_number function

create table things (id int, market varchar(30), grade varchar(30), term varchar(30), bid decimal, offer decimal)  insert things values (1,'heavy','abc','jun14',-19.5 , -17) ,(2,'heavy','abc','jul14',-20   ,-17.5) ,(3,'sour','xyz','jun14',-30   , -17) ,(4,'sour','xyz','jul14',-32   , -27)  ;with numbered (     select id, market, grade, term, bid, offer, row_number() on (partition market, grade order bid desc)     things ) --select * numbered select r1.market + '/' + r1.grade market, r1.term + '/' + r2.term term, r1.bid - r2.offer [bid], r1.offer - r2.bid [offer]  numbered r1  join numbered r2 on r1.market = r2.market , r1.grade = r2.grade , r1.i < r2.i