Is there such thing as a "good loss?"

My system currently treats all losses the same: as a zero to average against. A loss also damages the SRS component because your PF/PA ratio gets smaller. So it seems to me that losses are accounted for, and will show up in the rankings.

That said, I do penalize for playing certain types of schools. If you only get 0.5pts for a G5 win, shouldn't you get penalized more for losing to one? The problem is how to have this effect only affect P5 schools in negatively while simultaneously affecting P5 schools positively. Here's what I mean:

• When a P5 beats a G5, they only get 0.5pts and the G5 gets a 0. This seems OK?
• Should a G5 get rewarded for losing to a P5? If yes, why? And how?
• I still don't know what the probability is of a P5 beating a G5. It should be around 80% I think, so that 0.5 may get changed.
• When a G5 beats a P5, they get 1.0pts and the P5 gets a 0.
• The G5 is rewarded with more points for "playing up."
• The P5 is not penalized for playing down. Should they be? How much?
• Some P5 schools are worse programs than some G5 schools. How can we parse this?
• Should margin of victory matter? I feel this is handled in the SRS.
• If it matters, where's the cutoff? Is it different between P5 and G5 teams?

At some point in here a judgment call will be made to segregate teams based not on numbers. This is unacceptable to me. And, I do not have anything like the stats background or time to do that research, if it's even possible. 

Maybe we just use conferences? But there are teams in pretty much all G5 conferences that can bring the pain: FAU, Cincy, USF, Memphis (this year), Northern Illinois (from a few years ago), App State, Boise State, Air Force (sometimes). 


So, can losing ever really be good? Miami only lost to UF by 4. At the time the Gators were ranked #8 and the 'canes were unranked. Seems like we could call that a "good loss." It could be considered a "we played them tough" game. Yet Florida is currently RR #13 at 10-2 while Miami is RR #40 and 6-6. Considering Miami is 633-363-19 all time (.633), finishing at .500 can't be good, or even average. It's bad. Still call it a good loss? Maybe they would be worse if they'd gotten beaten worse? How can we know? 

We pick an arbitrary number for the margin of victory to call a "good" loss. Let's say 4pts. Now we go find every instance of a lower ranked team (how much lower? I don't know - you tell me) losing to a higher ranked team by 4 or less. Once we have that data, we look at that season and see if:

1. Did the losing team's record exceed, meet, or fail to meet the historical record?
2. Did the losing team's record exceed, meet, or fail to meet the historical record with that coach in charge?

OK, we have a framework for figuring if there's such thing as a "good loss." We can even determine how much it might be worth if we find that every "good" loss leads to X more wins. But keep in mind that X is probably going to have to be at least 1 standard deviation out to be distinguishable from noise. 

I don't know if I can parse all that data. Not easily, and not quickly, but maybe by next season. I'll get started and see if I can spot a trend going back 10 years to begin. 

What about a bad loss? Not many worse losses than UM to FCS App State in 2007. The Wolverines were ranked #5 when itty-bitty App State kicked the winning field goal with 26 seconds left in the Big House. UM went 9-4 (.692) after that, losing the next week to Oregon, then late in the season to Wisky and OSU before dropping out of the top 25 and beating #9 UF in the Orange Bowl to wind up #18. UM is 962-345-36 (.730) overall, but their bowl record is 21-26 (.447). So maybe it was a bad loss in that their record that season was 0.038 worse than normal, but they won their bowl game which is better than we should probably expect. 

Bad loss? Not so bad? I don't really see that we can tell without some heavy lifting:

1. What is the seasonal variance (SD) in UM's records historically? Then how far out is the record when they suffer a "bad" loss?
2. What is a bad loss? How many points to a how much lower ranked team? Does the margin change based on the differences in ranks?
1. E.g losing to an FCS school is always the worst, losing to a G5 is next worst, and so on
2. What margin?
3. Statistics

This is deep stuff, at least to me. And really, do any of you care? If I go through all of this and come up with a game modifier for G5/P5, FCS/P5, FCS/G5, good loss, bad loss, good win, bad or ugly win (I don't even know where to start with that concept), will you be convinced that an objective system works?

There are plenty of high school systems out there that do exactly what mine is doing, with the exception of the FCS penalty. And before you say "That's high school, this is college," bite your tongues. High school systems are designed to be fair and objective, account for differences in divisions/school size and funding, and at the same time allow for playoff match ups among teams who rarely if ever play each other. Sound like CFB? Just a bit? There's a reason I based my system on the OHSAA system, and ease of use certainly isn't it. 

Where does this leave me? Contemplating a lot of off season work. Have a great week! Bowls are coming up, and do yourselves a favor and watch some cool FCS playoffs.