Every week this fall, Fan Watch is going to tell you things like "root for Vanderbilt; it matters more than your own game." The fair question is: why the heck should you listen to us? Let's break it down.
We're in the hope business
First, Fan Watch isn't a rankings site. We're not trying to predict what will happen. Full stop.
We're trying to help fans know what to hope happens to most benefit their team. In my family, I am the person people have been asking "what needs to happen for our team to move up in the rankings?" I spent twenty years answering this by spending a lot of time looking at upcoming schedules and figuring out which games matter most. I was always looking for things like hoping a team we beat won the rest of their games to keep our quality wins intact. Or obvious stuff like hoping the team in front of us loses. Or hoping the team in front of us has its quality wins eroded.
So, if my dad asked me what needs to happen for the Ducks to move up, I could say more than "The Ducks need to win." I'd say things like, "Well, the Ducks beat Ohio State, so we want them to beat Michigan this week so our win over them continues to look good." In my family there were also follow-up questions like "well, what channel is that on?", which is why the site has how-to-watch pages too. But, that's a different post.
I also wanted to be realistic. Like, if a team in front of us was playing an FCS team, I didn't hold my breath for the upset, unless it was a giant killer like a South Dakota State or Appalachian State. (If you understood those references, you know how deep down this rabbit hole I've been.)
That's the heart and soul of this site: hope. If there's even a glimmer of it left in your season, the data will find it and we'll show it to you. And the probabilities that ride along with every verdict are there to regulate the emotional size of the hope. Is this a hail-mary, winning-the-lottery kind of thing? A reasonable expectation that SHOULD happen? Or something in between?
So there are two things you'd be trusting: the rooting logic, which outcomes actually help your team, and the probabilities, how much hope to attach. Different claims, different tests. We checked both.
Is the rooting logic right?
The logic lives or dies on the sport's actual rules. Conference tiebreakers are a cascade: head-to-head (in a tie of three or more teams, it only applies if the tied teams all played each other), then common opponents, conference strength of schedule, CFP rank, and computer metrics, in each conference's published order, and when a tie splits, the cascade restarts from the top. Get one branch wrong and our December advice is fiction; we'd be telling you to root for the wrong team in week 13. Selection has its own wiring: five auto-bids, seven at-large, byes to the top four champions.
So we test the wiring. The engine re-ran the last two seasons: championship week with the full regular season's results in hand, and field selection with the results and the committee's final rankings known. Acing that doesn't prove the engine sees the future (it had the answers). It proves the rules are wired right. It went 18 of 18 on conference championship matchups and 24 of 24 on playoff teams, both seasons of the 12-team format.
Are the probabilities honest?
The probabilities are the part we can never get perfect, only honest. The standard is calibration: when we say 80%, it should happen about 80% of the time. So we re-ran the engine "as of" each week of 2024 and 2025 using only what was knowable at the time, and scored every playoff-odds number against reality. Teams we put at 26% made the field 25% of the time; teams at 86% made it 84%. The weekly error (that's the Brier score) drops from about 0.06 in September to about 0.01 by championship week. And measured against the laziest possible forecast, handing every team the historical base rate of making the field, the engine roughly cuts the error in half: a pooled skill score of 0.53, where zero is no better than the lazy guess and one is perfect.
For now, that's a calibration claim, not a we-beat-FPI claim; base rates are the only benchmark we've scored against. The probabilities are also the part of the site we expect to keep improving, and if they get good enough, maybe one day we throw down the gauntlet against the big-boy prediction models. Today the claim is smaller and more useful: when we tell you a hope is a coin flip, it behaves like a coin flip.
The misses are published too. At the very top the engine runs a little hot: teams given 90 to 100% made the field roughly 84 to 92% of the time. Some of that is a small, correlated sample: one late snub echoes across many weeks of scoring. Some of it is real, because injuries are invisible to the model, and so are coaching changes and suspensions. And the 12-team playoff has existed for exactly two seasons, so every season-level claim here carries that asterisk. The full table is on the methodology page, unsmoothed, and we'll keep scoring it every week of 2026, in public, even when it embarrasses us.
That's why a Tuesday-night "your path runs through Vandy beating Texas" deserves more trust than a guy on a message board. Including the guy I used to be.
Where the ratings come from
Honestly, the preseason ratings are a side project. I fitted our own week-1 prior on eleven years of history (returning production, talent composite, recruiting class), and it was a fun learning exercise that happened to work: it beat both a plain carry-over of last season's ratings and CFBD's own week-1 numbers in 10 of 11 seasons. But once the season starts, the engine runs on CFBD's public Elo ratings (that's CollegeFootballData.com, the community stats site), the same numbers anyone can pull.
The edge was never secret ratings. What the calibration table scores is everything wrapped around them: ratings in, then schedule simulation, tiebreaker cascade, committee model, and playoff odds out. Nobody hands you playoff odds at CFBD. That part is the site.
What hope looks like in one season
Oregon, 2025, week by week. Week 6: 91% to make the field, the should-happen kind of hope. Week 11: 68%. Week 12: 65%, the low, the rooting-for-chaos weeks (I've written about those). Then the climb: 77% by week 14, 97% by week 15, and the Ducks made the actual field. The table holds the opposite stories too, the 90-percent teams that missed; that's the 84-to-92 line above. But that's the job: a weekly, honest answer to how much trouble are we in, and what has to happen.
So why trust us?
Not because we promise to out-predict anybody. That was never the offer. The offer is hope, handled honestly. If your team has a chance, we'll show it to you. We'll be real with you about whether it's a hail mary or a should-happen. And when there's something worth hoping for, we'll make sure you know how to watch it unfold.
That's the site. If you have hope for your team this fall, we built this to be your best source for it.