So, you actually want to know how this thing works?  Well, you have more intellectual curiosity than I assumed.  Let me see if I can satisfy it.

Each week the results are entered into a massive spreadsheet with all kinds of functions that link the results of each team together.  What you end up is something that looks like this:

You can see how each team is under a certain record (these are all the 4-0 teams from week 4 in 2011) and then next to them is the scores of each game, the opponent, and the opponent’s record in order of the week the game was played.

Since the RPM is a “blind” system, all the team names are then removed from the spreadsheet and the main team is given a random number by a random number generator and then sorted so that a brand new order is created.  This is what it then looks like:

As you can see, any indicator of what team is which has been eliminated.  Now, if you really wanted to figure it out, you could easily compare this with the above and see that Team 1 is clearly Oklahoma St.  But for my purposes, this is as good as it gets.

Now, above is the perfect scenario with no FCS games, no losses, no byes, and no inter-record matches.  That is not so below, where all four come into play:

 
As you can see, there are multiple cases to deal with here.  First is the most obvious, losses.  Losses are indicated with the inverse score (losing score first) and a red highlight.  Next are the FCS games, which are marked with a placeholder.  FCS losses include the placeholder plus the score, as well as being highlighted in red.  Thirdly, byes are marked by another placeholder (-).  Lastly are inter-record matches.  As you can see above, a special mark (œ) has been applied for Maryland and Miami FL (red for Miami since they lost the match).  This mark, as well as all the other indicators, remains there when the data is converted into the usable form.

In the end you wind up with a spreadsheet of all 120 teams and their records looking like the revised 4-0 section mentioned earlier.  From there the ranking begins.

------------------------

The ranking process is much more difficult to illustrate.  As the season progresses and the resumes become more robust, it gets easier to separate resumes, but it is much harder in the early weeks.  I briefly explain the issue in my welcome post, but I will go into it in a bit more depth here.

This is the only part of the RPM that is subjective and is, in the end, what separates it from computer polls.  I have to look at every resume and deem which ones are more impressive than the others.  This is not necessarily an easy task, and often deciding single positions is splitting hairs between resumes.

Let’s take another look at that 4-0 section:

Let’s look at these teams in a vacuum, as if they are the only 5 teams to rank.  How do you rank them?  There is no set procedure, as every circumstance merits different methods, but there are some easy tools.

Firstly, look for the big wins.  In this case 3-1 wins are the best possible wins a team could have.  Unfortunately, none of these teams have that win, so let’s move onto the next best win, 2.5-1.  As you can see, 4 of these teams (1,2,5,7)  have that win, so let’s focus on those for now.  Team 7 has two of these wins but the rest of their wins aren’t so impressive, so let’s leave them at #1 for now and revisit that later.  Team 2 has no other 2 true (FBS) win teams so let’s take a look at Teams 1 and 5.  Both have a single 2 true win team, though Team 1’s win is 2-1, not 2-2.  But, Team 5’s 3^rd and 4^th best wins are superior to Team 1’s, so we can say that Team 5 has the superior resume.  We can then compare Team 5 with Team 7 (the team we set aside because they have two 2.5-1 wins).  When looking at the full resumes, these two are hard to pull apart.  Team 7 has the better top tier wins, but Team 5 has the better lower tier wins.  In situations like this, I tend to favor the bigger wins.  So, I would rank Team 7 #1 and Team 5 #2.

Now we can compare Teams 2 and 1.  While Team 1 has a 2 true win team, Team 2 is not far off  with a  1.5-1 win, plus superior 3^rd and 4^th teir wins.  I would rank Team 2 #3.  Finally, we can bring back Team 9 and compare it to Team 1.  Team 1 has a better win on every tier and is clearly the better resume.  Team 1 would be ranked #4 and Team 9 #5.

As you can tell, this is a very drawn-out process that is done for all teams (usually focusing on the clear top 35 or so).  Also, it is fairly easy to see that there are parts of the process where one could pick a different resume over the other.  A case could easily be made for Team 5 to be ranked #1 over Team 7, and I would probably not argue with it.  But this is often the case in nearly ever polling system and is not at all unique to the RPM

In the end, you are left with the final RPM poll, a ranking of 25 resumes.  As always, feel free to send me any comments, questions or ideas and I will try to get to as many as virtually possible.

- Rampage