FAQ (Updated Nov 10, 2025)

What makes these rankings different than all of the other rankings?

TL;DR - Nothing, really. But more rankings > less rankings!

Let's start off with the assumption that the worldwide community of sports lovers does not agree on what rankings similar to these are intended to, well, rank. Some ranking systems are intended to identify the best teams, accepting inconsistent performance. Some systems are intended to rank teams with the best overall performance, regardless of competition. Some systems are heavily focused on identifying SoS (Strength of Schedule) and weigh season performance accordingly. But what do we really want to identify? The best teams, or the teams that are most "deserving" of a high ranking?

I would suggest considering the results of the 1998 college football season. Virtually every ranking system, other than human voting, would identify Ohio State as the best team, probably by a wide margin. Unfortunately for OSU, they suffered 5 turnovers in an arguably fluke loss to Michigan State, while Tennessee finished undefeated against similar competition and was unanimously voted our national champion. At the same time, Florida State, Texas A&M and UCLA all could be in the argument in spite of having at least 2 losses each, given they played (by my own calculations) the toughest schedules in the country.

Or put another way, human voters (and some performance systems) are likely to rank an undefeated team that only had a single top 25 win over an 8-5 team, even if that 8-5 team played every single game against teams that finished in the top 15.

But what if those two teams were to play against each other? Who would be the projected winner?

That's what my rankings are intended to do: Predict future performance based entirely on past performance. However, I readily admit that many other systems are intended to do this as well, and Jeff Sagarin even publishes a version of his own rankings that do just this. My rankings are really not unique, but they're just another drop of data in the overall ranking soup.

So that's what you're doing? Predicting future results?

TL;DR - Yes. Sort of.

More or less. The results published on this site include a power ranking. To predict a point spread on a neutral field, just compute the delta between the power rankings.

With that said, the college sports rankings published on this site are produced using a few modules that help to preserve merit of ranking over pure predictability.

Overall, they're just another predictor system, like the dozens already available.

Why bother, then?

TL;DR - I'm retired and have nothing else to do.

Back in the 1990's, a lovely friend (we miss you, Randy) and I had a nerdy competition every year to create a system to predict future game results. Every year we produced new systems, measuring their efficacy by the percentage of correct winners they chose over an entire season. Somewhat interestingly, the system that finished with the highest predictor score only considered rushing yardage -- it was not fed any information about scores, wins, or losses, only rushing data and opponents.

One algorithm I had dreamed up early on during those years of friendly competition was too difficult to implement with my meager coding skills, but I remained confident that it would work, and work well.

Fast forward a couple of decades.

A few years back, I was asked if I had ever written a football ranking system, and many of those old ideas came flooding back, including the one that I had never previously implemented. Having plenty of free time, better coding skills and a stubborn bent to prove that the algorithm would work, and work well, I set to task and, well, here we are.

How does it work?

TL;DR - No.

Like many systems, the splunty ranking system considers only scores. The algorithm works on any score-based sport, and obviously becomes stronger as the degrees of separation shrink between competitors (e.g., as more games are played).

How well does this system actually work?

TL;DR - "Pretty dang."

The college football version, which has always been my passion, regularly beats Vegas, averaging about 61% of the time last year when used with a confidence methodology. The NFL version is less accurate for two reasons. First, NFL outcomes are easier to predict, so Vegas spreads tend to be more accurate. Second, the code used to produce the NFL version is a basic implementation of the algorithm while the college football version is littered with (experimental? cutting-edge? crackpot?) modules that consider certain effects that are more impactful than they are in professional sports, such as home field advantage.

Home field advantage, which I've given up calculating.

Home field advantage? What is it?

TL;DR - For college football, on average, about 4.35 in week zero, and 0.83 in week 15.

Home field advantage absolutely matters when trying to predict who will win a close game. Unfortunately, it's difficult to calculate an efficacious home field advantage for several reasons.

• Every team (or home field, perhaps) has a different effective advantage.

• Every team is affected differently by traveling.

• Home field advantage changes dramatically throughout the season.

Last season, I implemented an unorthodox method of measure, which essentially asked these three questions:

1) What "generic" home field advantage number would have resulted in the highest overall efficacy when predicting against Vegas spreads the previous week?

2) For a given specific team, considering the entire season while playing at home, what delta against that generic HFA would have resulted in an even higher overall efficacy?

3) For a given specific team, considering the entire season while playing on the road, what delta against that generic HFA for their opponents would have resulted in higher overall efficacy?

What I found is that some teams had virtually no advantage at home, while others maintained a strong advantage all year (for example, TAMU). Some teams traveled very well, while others performed miserably against predicted spreads on the road (for example, TAMU).

And, overall, the advantage of playing at home shrank remarkably as the season progressed! Do people get tired of yelling as the season goes on? Do teams get used to travel and noise? I've speculated this has more to do with underclassmen getting their first taste of playing away games and becoming seasoned (jaded?) as the experience is repeated. (The NFL home field advantage also shrinks during the season, when computed the same way, but only by about .2 points over the course of the entire season.)

Unfortunately, I wound up having to rewrite all of the ranking code multiple times. For the 2025 seasons, I simply took the average HFA from each week last year and plugged it in this year, because rewriting all of that code is beyond the scope of where my interest keeps me these days.

So this is actually fun for you?

TL;DR - Yes!

This is probably much more fun than it ought to be, to be honest. If you bothered to read this, you're probably already aware of Kenneth Massey's ranking systems.

https://masseyratings.com/

In addition to wildly interesting rankings of his own, Kenneth also publishes a composite of dozens of ranking systems, including these, for various sports. While interesting on their own, the composites are incredibly interesting to somebody who works on algorithms such as my own. It's fascinating to me to see how these ranking systems, all of which produce meritable results, each produce different results and different outliers. And it's the outliers that seem to fascinate me the most. "Why does my system think State Valley Tech is in the top 25, while no other system does? Is it right? Should I bet the farm on SVT?"

That's really all there is to it. A lot of effort and time devoted to producing yet another ranking system that gets asked all of the same questions. But it's fun, and that's all that matters to me.