Tuesday, December 04, 2007

Something About Ranked Opponents

The M Zone calls a team's record against ranked opponents the "dumbest stat in college football". I, for one, think punting statistics are the dumbest in college football, because they don't take into account the intent of a particular punt (if punters are pooching or cornering kicks frequently, the average length of a punt is a moronic stat).

But the topic of record against ranked opponents is an interesting one, and I think The M Zone's analysis is incomplete.

Their view is that when people cite a particular record against ranked opponents, they normally cite the record against opponents who were ranked at the time they played, and this is erroneous because teams can very well be ranked inaccurately at the point they play. As examples, they cite LSU's schedule, and credit for wins against South Carolina and Alabama, who were ranked at the time LSU played each. South Carolina and Alabama both went in the tank at the end of the year, losing their last 5 and 4 games, respectively.

The M Zone says that "record against ranked opponents" should take into account the rankings of those opponents at the end of the year, when the body of work is complete.

My view: yes, simply citing ranked opponents as of the time played is incomplete analysis. But citing ranked opponents as of the end of the regular season is also incomplete analysis.

Rankings are fluid over the course of the year, but so are teams. It is possible that an inflated ranking early in the year isn't an accurate picture of a particular team (such as, say, an early win over a top-10 preseason ranked Louisville team). But it's also possible that an inflated ranking early in the year is an accurate picture of a particular team at that point in time - just the team changed significantly over the course of time, as a result of injuries or personnel changes.

The M Zone cites South Carolina as it relates to LSU (and dozens of others have pointed to Georgia's loss to South Carolina as a reason for excluding the Dawgs from the BCS NCG). USC lost their last 5 games, dropping their overall record to a thoroughly mediocre 6-6. The week LSU played USC, USC was 3-0 and ranked 12th in the country. One could look at their final record and say they were significantly overrated at the time. Or one could look further. USC's best player on the defense, Jasper Brinkley (an All-SEC linebacker) injured his knee in the LSU game, changing the USC defense entirely. Before the injury, USC had averaged allowing fewer than 10 points per game. After the LSU game, with Brinkley out, USC conceded 28 points per game. Maybe they started playing better opponents, maybe they simply weren't that good overall. Or maybe the loss of key personnel (who played in the games against LSU and Georgia) made a massive difference to the season. The "at the time of" ranking might be an accurate depiction of the team LSU (or Georgia) played, AND the "end of season" ranking might also be an accurate depiction of the complete season USC had.

The thing is that temporal rankings can provide some context as to a game's value, but it takes more than a cursory citation. The problem is that the citation of such a stat is a simplification - but then again, the citation of any stat can be a simplification.

And at the end of their post, The M Zone is definitely right about how crazy it is that this sport uses opinions to decide things, and those opinions rely on incomplete information.


peacedog said...

They were also without Captain Munnerlyn for some of that stretch, yes?

I think another factor it underscores is how depth is key. It'll take Spurrier more than 2 years of recruiting to build better depth there. There aren't really any teams like the FSUs of old who could lose a key player here or there and not miss a beat. However, a team like UGA has better depth than South Carolina right now (though that wasn't true for some positions at times during the season, e.g. with Brown and Lump out), for most positions.

That's not to excuse Carolina's losing. But I think it's important to examine why a team had radically different halves of the season, and that this analysis is important when evaluating team's Ws and Ls against that opponent.

Andrew said...

I think wins and losses are the dumbest stats in football.

Kanu said...

Exhibit B is Cal.

Started out 5-0 and ranked #2, including a 14 pt win over Tennessee and going to Autzen and beating a full strength Oregon team. Whether or not they were a top 2 team doesn't matter, back then they were most certainly a top 10 team and playing really good football and thinking Rose/MNCG even.

QB Longshore hurt & 2 disappointing losses later, then a road loss to #4 at the time ASU, and at 5-3 they basically quit on the season, playing shitty football in their last 4 games going 1-3 with the lone win a hold on for dear life 20-17 over Wazzu even thought they {read: Longshore} was back to health.

Same players who were playing great football early (5-0 with 2 quality wins) are now playing shit football late (1-5 with one close shave against a shit team).

But as you said, you can't look back and say that they were a crappy team when they played UT and UO, and vicey versey. Like many things in CFB, it's more nuanced than that.

asimperson said...

If exhibit B is Cal, then our next look should be at Oregon. Has any team been hurt by the loss of one player more than them? With Dennis Dixon, they're one of the top 5 teams in the country, without, well, we saw what happened.

Thus, taking into account both ranking at the time and end-season ranking may be a good idea. But how much is each one worth? How do you weigh some rankings which were probably correct at the time (Cal, Oregon) against teams that were obviously overrated (Louisville, South Carolina)?

I think that's why people just pick one.

LD said...

South Carolina wasn't obviously overrated - it's not clear. I think they were a much better team when they had Captain Munnerlyn and Jasper Brinkley - and beat UGA and were a good win for LSU. But that's just my opinion. It's impossible to quantify exactly how good a team actually is on a particular day, even in reference to the overall picture of that team.

Point is that people just pick one - and there's not enough thought or analysis put into what people do. And that's why subjective opinions are stupid.

The Wrangler said...

Probably noted elsewhere, but it's amazing how the coverage by CFB's most well-known "analyst" is shadowing the last two Presidential elections:

2000: Herbstreit "calling" the Miles to UM hire, then it being too close to call, then 4 (days) later Miles getting a second term at LSU.

2004: Herbstreit "flip-flops" from the UM promoting - "you need to decide who the 2 best teams are" - to the UGA slighting - "didn't even win their division" and "it's not who the two best or hottest teams are."

And doesn't Les Miles just remind you a little bit of a Forrest Gumpier "W?"

Anonymous said...