Where You Disagree With Congrove...

[billybud]

And consider this, of the 6 computer systems that helped choose Florida and Oklahoma, 5 of the 6 picked the winner in less than 50% of the bowl games. Massey’s system was the only one to have the higher ranked team win more than half the time.

[..fanatic]

And consider this, of the 6 computer systems that helped choose Florida and Oklahoma, 5 of the 6 picked the winner in less than 50% of the bowl games. Massey’s system was the only one to have the higher ranked team win more than half the time.

So? What's that got to do with the topic? No one here, that I know of, is defending the specific BCS computers. Besides, bowl games are NOT a solid measurement of anyone's picking success as teams have an extended period of time to prepare for one game.

[billybud]

Just pointing out that the computers aren't always that proficient. Most of us probably had better than a 50% rate on picking bowl winners.

And, even though the teams have more time to prepare...the computers also have 12 to 13 games worth of "experience" with which to make predictions. Much better data than in game four (with three weaker OOC games in the books).

The problem is..the closer teams are matched, the tougher it gets to call more than 50%...and the teams are often more evenly matched in bowls than during the season.

[CFP Admin]

Yaeh...but Congrove purports this to be a "power poll" not a poll based on the schedule.

I disagree with his ranking of power on some of these teams...and the schedule should have nothing to do with a straight power ranking.

Perhaps you are taking too much from a casual definiton on site that POWER IS THE STRENGTH OF A TEAM. That is the short definition for the casual observer. What measures strength of one team vs. another in an unbalanced league such as the FBS? The simple answer is a comparison of schedules and results.

The point of a power ranking is an attempt to discern the number of points by which Team A is better than another. Thus, strength of schedule is not only an important ingredient in a power rating formula, it is a core value. This is why SOS was redundant when the BCS tried to ADD it to the formula for a few years. Every computer power rating already relies on it to produce a result. And every voter considers SOS to some extent.

TEAM A and the average power rating of their opponents - TEAM B and the average power rating of their opponents = spread. Once the season begins, actual on-field results change those numbers. "Their opponents" IS the schedule. Home-field advantage is learned from the schedule, factored into the picks, and ultimately into an adjusted power rating after the game is played.

That's why this was all pointed out in the conference preview to explain how Tulsa could receive such a lofty analysis by the computer. As pointed out on each conference preview page, "Our conference reports analyze what the Congrove Computer Rankings preseason forecast predicts. Team power ratings change weekly during the season to reflect the actual results of the games. For projecting results, 3 points are added to the home team."

You also can not draw the conclusion that a team rated No. 10 in power would be expected to beat the No. 15 team, unless that team was actually on the schedule and factored into assessing the power rating.

This is what creates an advantage (perhaps unfairly) to teams that play in conferences with two divisions. The conference championship adds a game against a strong team, reconfiguring their power ratings through SOS improvement and providing a chance to increase their superiority over the field.

[billybud]

A "power ranking" to most folks means that a team ranked much higher than another would be more powerful than another..and thus favored. Sagarin's power and Steele's for example.

CCR may call it a power ranking but it is more of a "expected result" ranking..IMHO

Edit...I still would take lesser "power" teams like LSU, Mississippi, Georgia, etc in a match with Tulsa...I really, I guess, am having a problem understanding the concept of a power rating.

[billybud]

I guess that I don't delve that deeply into the exact methodology each system uses....

Last year's tracking of 40 systems and how they fared:

http://www.thepredictiontracker.com/ncaaresults.php

[billybud]

Wow...I guess from what Congrove says...nothing means nothing until some games get played. sounds just like us...LOL.

Thanks, Admin...for the input and the explanation.

[Eric]

I think I saw somewhere that Temple would be 11-1. My only major beef with the ranking system.

I might give the Owls 7-5. And again, I'm no computer wiz, but it always seemed to me that the CCR preseason projections always put a little too much stock into what happened last year. I just happened to notice that many of the teams that went on to surprise were always picked to have even better years (in 2007, I think big things were projected for Wake Forest, Rice, and San Jose State but I can't be sure). Definitely not a scientific analysis; I don't know where to find past preseason projections on the main site

And, I've said this before, but if FCS teams were accounted for as individuals, they might be closer on the spread. I don't think lines are set for those game, but the score difference may be a little more accurate. Richmond is not North Carolina A&T, but they all get lumped in the same group. And saying that, I think Sagarin gives WAAAAY too much credence to FCS teams. I think last time I checked that Richmond was better than teams like USF, Tulsa, Virginia (who beat Richmond), Kentucky, Notre Dame, etc. That is bogus. Apparently, Weber State is better than Texas A&M, Fresno State, Hawaii (who beat Weber State needless to say), Louisville, and Central Michigan.

[CFP Admin]

A "power ranking" to most folks means that a team ranked much higher than another would be more powerful than another..and thus favored. Sagarin's power and Steele's for example.

CCR may call it a power ranking but it is more of a "expected result" ranking..IMHO

Edit...I still would take lesser "power" teams like LSU, Mississippi, Georgia, etc in a match with Tulsa...I really, I guess, am having a problem understanding the concept of a power rating.

If most folks believe that, then most folks misunderstand the complexities and necessities to producing a Power Rating. "Recognized" computer rankings, including Sagarin, are all based on the same pretext.

Steele is an exception to the rule. He has nine different sets of ratings based on lots of personal opinion information, then combines all 9 to make a "power poll". He explains this on page 315 of his 2009 preview, where you will also find his "Power Poll", sans any numerical data.

[CFP Admin]

Wow...I guess from what Congrove says...nothing means nothing until some games get played. sounds just like us...LOL.

Thanks, Admin...for the input and the explanation.

Positively true. Before any games are played, it's just entertainment. I'm always surprised by some of what comes out, but it's the nature of the beast and we point these out in the various conference previews. Sometimes the suprises prove true - other times they miss by a mile. But the fact is, no one can achieve 100% predictive accuracy - while the rankings and picks are perfect, the teams are simply too fallable (tongue-in-cheek humor).

[Eric]

I don't get how Steele does his power poll combination. Like Billybud said, it should be based on how good the teams are. Not historical performances, not schedules, not anything but that.

[CFP Admin]

I don't get how Steele does his power poll combination. Like Billybud said, it should be based on how good the teams are. Not historical performances, not schedules, not anything but that.

It isn't a matter of choice that makes the methodology correct. It's a matter of requirement. By process of elimination, there are few things you CAN use as concrete date, and not opinion, to produce an unopinionated result. Without comparing who played who, it all means nothing.

[billybud]

Steele does not rely completely on computers to do his preseason power poll...

He analyzes talent position by position...returners, strengths, etc. He looks at defense, offense, special teams, kicking. He maintains nine sets of power ratings for each team (and it is proprietary..he will not divulge this since it is his hallmark).

[donovan]

Sounds like what CLF does....maybe CLF is right.

[CFP Admin]

And, I've said this before, but if FCS teams were accounted for as individuals, they might be closer on the spread. I don't think lines are set for those game, but the score difference may be a little more accurate. Richmond is not North Carolina A&T, but they all get lumped in the same group. And saying that, I think Sagarin gives WAAAAY too much credence to FCS teams. I think last time I checked that Richmond was better than teams like USF, Tulsa, Virginia (who beat Richmond), Kentucky, Notre Dame, etc. That is bogus. Apparently, Weber State is better than Texas A&M, Fresno State, Hawaii (who beat Weber State needless to say), Louisville, and Central Michigan.

I'm only trying to assess FBS teams. As FBS teams win 98% of the games vs. the FCS, there is little reason for me to care about specifically rating each FCS team. Even so, I've worked at developing a reliable way of doing so, but there simply isn't enough interaction to produce reliable results.