If you've read through the preceding sections on the criteria for ranking college football teams-- relevant record, head-to-head results, etc.--then you have most of what you need to know about how to put together a proper top 25. The criteria are the nuts and bolts for ranking one team over another. But how should you actually, physically go about doing it?
My first suggestion: index cards. One for each team that is a candidate for ranking. On each team's index card goes its full schedule. As the season progresses, fill in the scores. On the back, you can put brief notes if you like (injuries, suspensions, or weather that affected a game, and anything that makes a particular score unrepresentative of what happened in the game, such as points in "garbage time").
It's old school, but it can save you time in the long run, as it's easier to review 30-40 cards in your hand than it is to click through all those schedules on your computer. And it's also easier to manipulate and order them on a table top (or anywhere). Most importantly, it's good to be able to see the schedules as you're ordering them. If you're just typing school names into a list, it's all too easy to miss or forget a head-to-head result (which apparently happens to AP poll voters all the time).
Also, you might find it easier, especially late in the season, to rank teams within the same conference first, since they all play the same teams. Once you've done that for all conferences, you can start interlocking them into a national top 25. You should start your rankings from the top (#1) and work down from there. Maybe that goes without saying, but just in case, I said it anyway.
Once you have an initial top 25, congratulations, you've finished your first draft! Now comes the most important part, where the real work comes in: revision. Just like when you were writing term papers in high school. Okay, okay, I know most of you just turned in your first draft when you were in high school. And in college (and grad school too if you went). So did I. But this is vastly more important than your stupid doctoral thesis... this is the college football top 25, man! So get it right. REVISE!
For your first draft, you can labor on it intensely, or you can just do a quick "feel"-based top 25. Unfortunately, every final AP poll looks like the latter, skipping the revision step. Which is why I treat every final AP top 25 as a first draft, and take on the role of editor, correcting the errors myself. In any case, once you have a first draft, you now have a good general idea of where teams will be ranked, and can look at strength of schedule in detail. This is where you want to look carefully at each team's relevant record. You'll also need to make head-to-head adjustments as necessary.
You can do this revision/review in a number of ways, but here's what I suggest. Type up your first draft list, or if you're old school, write it down on a sheet of scratch paper. You should also include teams #26-30, or thereabouts, as one or more of them might make it in on revision. Next to each team, list the teams they lost to, then list the relevant teams they defeated. Ranked teams can be listed just by their ranking number, but also include nearly ranked teams (roughly the next 15 teams, or what you would rank #26-40 if you did a top 40). Here is an example of what this might look like with two teams from the final 2009 AP poll:
| 7) Iowa 11-2 | Losses: 5, X | Wins: 9, 13, 16, + |
| 8) Cincinnati 12-1 | Losses: 3 | Wins: 15, 25, +, +, + |
These are the notation symbols I use. Again, the numbers indicate the ranking of their top 25 opponents. The "X" in Iowa's list of losses indicates an unrated team (in this case, Northwestern). The "+" in both teams' lists of wins indicates a good but unrated team (in this case, I just used the 14 teams in the AP poll's "also receiving votes" section). You can use whatever notation you like. If you want, you can also include close wins over bad teams (touchdown or less). I didn't put this in the example above, but I use a minus symbol in the wins list for this. Iowa would have four of them and Cincinnati would have one.
This gives you a good overview of both strength of schedule and head-to-head results, and you can do a lot of adjusting just by looking at this. As an obvious example, if you have a team with a relevant record that looks like this-- "L: #4, #12; W: #15, #22"-- and you have that team rated #18, you can clearly see that they more likely belong at #13-14.
Next, you may want to circle or highlight any loss to a lower-ranked team and any win over a higher-ranked team (in the example above, that would just be the "X" in Iowa's list of losses). If you have too many circles on one side or the other for a team, you've probably ranked the team incorrectly.
You can also turn these circled results into a simple point system that can give you a good look at where your rankings may need to be adjusted. For each loss to a lower-rated team, subtract one point. For each win against a higher-rated team, add a point. For the teams in the example above, Iowa would be at -1, and Cincinnati would be at 0. You can put these simple scores to the left of every team to see how they stack up against one another. Because there are always plenty of upsets, usually the majority of teams in a late-season top 25 will be at -1, and some will be at -2.
Look for places where a team is ranked behind a team with a lower number than it has. Those are the teams you need to look at to see if they should be ranked higher. In the example above, Cincinnati has a higher number than Iowa (0 to -1). If you don't use this point system, you can just eyeball where the circled losses and wins are. In the example above, you would see that Iowa has a circled loss-- they lost to an unrated team-- and Cincinnati has no circles-- they lost only to #3 Florida.
You can think of Cincinnati as being one game better than Iowa in their relevant records. This isn't because Cincinnati is 12-1 and Iowa is 11-2; it is because Iowa lost to a lower-rated team and Cincinnati did not. If Iowa's 2 losses had come to #1 and #5, then Cincinnati and Iowa would be even in their relevant records, regardless of straight records.
Because Cincinnati has this effective one game lead, you should look carefully to see if Cincinnati should be moved ahead of Iowa. This is where you can get into the details of strength of schedule, performance, and improvement. And this is where you will want to look at the teams' full schedules (such as the index cards), and not just the teams' boiled-down relevant records. You'll want to ask yourself if Iowa played a tougher schedule than Cincinnati, and whether they outperformed Cincinnati, and if so, whether or not the difference is great enough to make up for their upset loss to an unrated team (that Cincinnati did not suffer).
In this case, Iowa's performance on the season was exceedingly poor (barely beating four bad teams), and they don't have improvement on Cincinnati, as their upset loss came in their tenth game. They did defeat some higher rated teams than Cincinnati did, and beat 3 overall to Cincinnati's 2, but Cincinnati defeated 3 near-rated teams to Iowa's 1. It doesn't look like there's enough to keep Cincinnati from being moved ahead of Iowa.
Of course, you should double-check all of the teams, not just teams with fewer upset losses than teams ranked ahead of them. When you look at the relevant records, it can change your view on where a number of teams should be ranked. Look carefully at all the criteria (strength of schedule, performance, improvement) for any two teams that are ranked next to each other and that are not a clear case.
Once you've adjusted and sharpened the entire top 25, are you done? No. Now stop that whining-- you're almost done. If you've made any adjustments (which you should have unless you spent a great deal of time on your first draft), it can affect earlier choices that you made. You need to review your top 25 to see if this happened. If you changed a team's ranking by a significant amount (around 5 or more places), you should check to see if this affects the ranking of any of that team's opponents.
Sticking with the 2009 AP poll as an example of a first draft, they had Pittsburgh at #15 and West Virginia at #25. Both teams would be at -2. Pitt lost to an unranked team and to #25 West Virginia. And they did not defeat a single team in the top 25. West Virginia lost to 3 unranked teams (including a big loss to an unranked team in their bowl game), and the only rated team they beat was Pitt. Looking at these facts, let's suppose that you dropped Pitt a great deal, maybe even behind West Virginia, who beat them. Or maybe you dropped them both out of your top 25 altogether. Given their relevant records, that would be a valid choice.
Well, if you did drop them, or at least one of them, and if you had earlier moved Cincinnati past Iowa, then you might want to revisit that earlier decision. Because as it happens, Pitt and West Virginia are the only rated teams Cincinnati defeated. If Pitt drops a significant amount, and if one or both of them drop out altogether, then Iowa's schedule suddenly looks a lot tougher than Cincinnati's.
If all this sounds time-consuming, well, doing a proper college football top 25 should be time-consuming. That's why the AP poll ballots should be due Sunday evening at the earliest. And really, what's the hurry? Given that their real jobs are most time-demanding on game day, AP poll voters can't possibly spend enough time on their ballots. Which is why they tend to use the terrible method of taking their previous list and mostly just dropping the teams that lost a relatively random number of places (too often carelessly dropping them behind teams they already beat). And why they focus so much on simple straight records.
Which brings me to the last section of
this handy guide... How
Not to Rank Teams