Help! Combining two sorting variables of a portfolio

[Sheharyar24]

Please can you tell me how can I make a table following the description below (PDF attached pg 21, 22 containing the table i want to replicate). I want the commands, to sort them out first on the basis of past returns(losers, Neutrals, winners) and then further sorting them into sizes and in the end get the table with the p1,p2,p3(three return portfolios) on the left hand side and sizes s1,s2....s10 in the upper column and then returns in the centre with t values. I have attached the pdf file which will show the exact table follwoing this description, table which i want is on page 21,22. I will really be grateful. Thanks in advance.

Momentum Strategies Based on Raw Returns and Sorted by Size (1980-2004)

This table reports the average monthly returns of momentum portfolios formed based on the previous K

months’ raw returns and held for K months (K=3, 6, 9,and 12). Each month, NYSE and AMEX

stocks are ranked in ascending order based on the lagged returns and divided into three equal-weighted

portfolios. Portfolio P1 is formed by the loser stocks and P3 is formed by the winner stocks over the

previous K months. This table presents the time-series average monthly returns of holding these

portfolios for K months. On the other hand, stocks are sorted into ten portfolios each month based on

their previous capitalizations. Portfolio S1 includes the smallest firms and S10 includes the largest

firms. Within each decile, firms are equally sorted into three portfolios (P1 to P3) based on their lag

raw returns. This table also reports the time-series average monthly return of holding these portfolios

in each size class for K months. Median sizes are in millions and t-statistics are in parenthesis.

[Patrick]

There are at least 2 questions here. 1) How do I calculate all these measures, 2) How do I create a report over this measures in the required format

A lot of people in this forum here are very supportive in helping to you to get over obstacles when you get stuck in your work. It's another story when it comes to doing your job for you - and this is what you're asking for here.

I suggest that you get started in solving your problem and come back here when you're facing issues with some concrete problems. If so then post a data step which creates sample data, all the code you've already developed and then explain where it doesn't work and how the result should look like.