clock menu more-arrow no yes

Filed under:

The Numbers Game: 40 Days and 40 Nights of BCS Posts

New, comments

So we've seen the new BCS poll, Peter's taken us through the various scenarios in which Texas can make a BCS game and/or the national championship game, and HornBrain has fired his perpetually delayed weekly salvo of hate in the war against Richard Billingsley's awful computer ranking system.  But what I want to focus on here is the nitty gritty of the numbers  that constitute the BCS rankings to see what we're looking at going forward and how likely things are to change given the various scenarios.  That's right, it's the 84th post in the last 2 days on the BCS!  Now with even more numbers!

Some might call this a useless exercise in "how can Texas back into the Big 12 and national championship games" and that's true in a sense.  But what's also true is that just because a team has 1 loss doesn't mean it's backing into anything.  A 2-loss team won the national championship last year.  This is the reality we're living with.  You're not out of anything until you're mathematically eliminated.  And Texas' math is very much alive for both the Big 12 and National Championship games.

So let's take a look at the numbers, shall we?

An Explanatory Prelude That Seems Simple But is Oh So Important:  It's important to realize that rankings in the BCS component polls (Harris, Coaches and Computers) mean very little.  It's the scores that establish those rankings that matter, not the rankings themselves.  In the Harris Poll, there are 114 voters for every #1 ranking, a team gets 25 points, for every #2 ranking, a team gets 24 points, and so on.  Thus the greatest number of points a team can get if it gets all 114 first place votes is 2,850, which is 25 times 114.  Alabama has 2,808 out of a possible 2,850, which gives them their Harris poll score of .9853 (which is 2808 divided by 2850).  If Alabama had, say, only 2770 points out of 2850, they would still be ranked #1, but their Harris Poll score for BCS purposes would only be .9719, significantly lower than it is now.  Thus rank doesn't matter, only points.

And because of that, you can have a "strong" ranking or a "weak" ranking.  Therefore, if the #4 team is just 1 point behind the #3 team in the Harris poll (thus #3 is weak and #4 is strong), and then if the #4 team then passes the #3 team the next week and now leads by 1 point, that means virtually nothing for the BCS because even though the rankings changed, the points stayed almost exactly the same.  It's not absolute rankings that matter, it's the points.

That being established, let's take a look at the Harris and Coaches Polls, points-wise:

Harris Poll
Rank Expected Pts Actual Pts Difference Team
1 2850 2808 -42 Alabama
2 2736 2768 32 Texas Tech
3 2622 2531 -91 Florida
4 2508 2471 -37 Texas
5 2394 2427 33 Oklahoma
6 2280 2329 49 USC
7 2166 2073 -93 PSU
8 2052 2034 -18 Utah
USA Today Coaches Poll
Rank Expected Pts Actual Pts Difference Team
1 1525 1508 -17 Alabama
2 1464 1469 5 Texas Tech
3 1403 1348 -55 Florida
4 1342 1314 -28 Oklahoma
5 1281 1300 19 Texas
6 1220 1268 48 USC
7 1159 1104 -55 Utah
8 1098 1074 -24 PSU

So now we've got something to work with.  Here's what we see: Bama and Tech are clearly 1 and 2, though a few brave souls have some other team (likely Florida) ahead of one or both of them.  Then there's a clumping from 3-6 as numbers 3 and 4 are both below what would be expected of their rankings, points-wise, while 5 and 6 have more points than would be expected for their rankings.  This seems to show that these 4 teams are clumped together and voters disagree on where to rank them relative to each other.  Only 80 points separates these four in the Coaches Poll rather than the expected 183 and only 202 points separate them in the Harris Poll rather than the expected 342.

It seems that barring a catastrophe of epic proportions, the national championship game contestants will likely be either Bama, Tech, or (should either or both of those two lose) one of these four clumped teams.  So let's unpack this a bit by establishing Five Facts.

Opinion Masquerading as Fact 1: USC's remaining schedule will not allow them togain many points on any team ahead of them unless those teams lose or play incredibly poorly.  Crushing victories over Stanford, Notre Dame and UCLA are not going to impress anyone.  USC needs several losses to make the national championship game. Only Florida, Alabama and Texas Tech control their own destinies.  Win out and they're into the national championship game.

Actual Fact 2: Texas' lead over OU in the Harris Poll is greater than OU's lead over Texas in the Coaches Poll.  Texas leads OU by 44 points in the Harris, where the expected lead for #4 over #5 is 114 points (or 37.61% of expected).  OU's lead over Texas in the Coaches poll is 14 out of an expected 61 (or 22.95% of expected).  This means that Texas is currently leading Oklahoma in the Human polls.

Opinion Bordering on Fact 3: OU has the most daunting schedule remaining in the regular season.  Florida plays South Carolina, The Citadel, and at Florida State, while Texas plays at Kansas and vs. Texas A&M.  OU, on the other hand, has Tech at home and Oklahoma State in Stillwater.  That's rough, but if they pull it off, they stand to gain some ground points-wise, if not rankings-wise.

Actual Fact 4 Followed by Reasoned Speculation: Texas plays Kansas in Lawrence this weekend while OU is idle.  A dominating win over a pretty decent Kansas team on the road while OU is idle might provide incentive for Texas to increase its points lead in the Harris poll while perhaps overtaking OU in the Coaches poll.  By which I mean that more voters in both polls would rank Texas ahead of OU than currently do.

Actual Fact 5 Followed by Somewhat Specious Speculation: Texas beat OU on a neutral field.  Head-to-head tends to resonate with human voters.  If Texas is ranked ahead of OU on most voters' ballots going into OU's final two games, then the head-to-head win by Texas on a neutral field might prevent some voters from moving OU ahead of Texas if OU wins out because it takes it takes a certain amount of audacity (by which I mean to connote "fearless daring" and not "insolence") to take the plunge and move a team like OU above a team like Texas when Texas beat OU on a neutral field.  However, it takes no audacity (of either connotation) to keep OU above Texas if they are already there.  Thus, if most voters have OU ahead of Texas to begin with, anything short of a loss by OU will not move Texas ahead of the Sooners on those particular ballots.

A Calming Word to Texas Tech Fans on Scenarios Leading to the Big 12 Championship Game: We know for a fact that Texas cannot make the Big 12 Championship game unless Texas wins out and Oklahoma beats Texas Tech.  In speaking at all about the possibility of Texas going to the Big 12 Championship game, we are implicitly assuming that these things will happen, not because we assume that they will, but because we know that they must under this hypothetical.  Okay, Tech fans?  So, if those things occur, then 1 of 2 things (or both) needs to happen: either Tech loses to Baylor and/or OU beats OSU.  If Tech loses to Baylor, Texas goes to Kansas City regardless of the outcome of the Bedlam game.  If Tech beats Baylor and OU beats OSU, then the team that's ahead in the BCS wins the South division and goes to Kansas City.

I think it's a fair assumption that if Tech loses to OU, they will be ranked below both Texas and OU in the BCS.  It's not necessarily fair that this will be the case, but as a matter of name recognition above anything, humans will put them below while I think the computers will as well (at least behind UT if not OU) given their gimpy early-season schedule.  Thus in this limited scenario, the question becomes who is ahead in the BCS, Texas or OU?  Fact 3 above states that OU has the possibility to impress the most, but Facts 4 and 5 give me some hope that a majority of Humans (or at least around half) will still have Texas on top of OU in their polls come season's end.  The important thing is for Texas to gain ground this weekend to offset to the greatest extent possible the ground that OU will gain back in the last 2 weeks.

If OU and Texas are approximately tied in the human polls or if OU has a slight advantage (i.e. something close to the advantage that Texas currently has), then it will come down to the computers.  Let's take a quick look at those.

Computers Poll

Rank A&H Billingsley Colley Massey Sagarin Wolfe Average Team
1 24 25 24 25 25 24 .980 Texas Tech
2 25 24 23 24 24 25 .970 Alabama
3 22 23 25 23 23 23 .920 Texas
4 23 19 21 22 22 22 .870 Utah
T-5 19 20 20 21 21 21 .820 Oklahoma
T-5 21 21 22 20 20 20 .820 Florida
7 18 18 17 19 19 19 .740 Georgia
8 17 22 19 18 16 18 .720 USC


Remember that the top ranking and the bottom ranking get thrown out (for some reason that I don't understand at all.....just throw out Billingsley's every time and be done with it) so only the middle 4 rankings get figured into the BCS score.  Notice also that Texas is the consensus #3 team.  The #3 team should have 92% of the possible points (23 times 4) and Texas has exactly that.  perhaps more importantly, though, note that Utah is ranked in between Texas and OU in 4 polls and that Florida is ranked in between Texas and OU in 3 polls and USC, Texas Tech and Alabama are ranked in between Texas and OU in 1 poll each (as is Boise St., which is not shown in the chart).  For Texas to have the best chance to be ranked ahead of OU at the end of the regular season, root for each and every one of these teams (particularly Utah and Florida) to win their remaining games in the regular season (except for Tech against OU of course) to keep as much buffer between Texas and OU in the computers as possible.  Also keep in mind that while humans tend to forget or not pay attention to what happened earlier in the season, computers don't.  Root for every one of OUs previous non-mutual opponents to lose (particularly TCU) and every one of Texas' non-mutual opponents to keep on winning (particularly Mizzou).

OU is ranked fairly low by the computers (they are a concensus 5.5, as is Florida), which is good for now, but it also means that the Sooners have plenty of room to move up with wins over Tech and OSU.  They have a lot of ground to make up to catch Texas though and I'm not sure they can do it.

Speculation Evincing My Poor Judgment Not Necessarily Due to the Substance of the Speculation but Rather Simply by Virtue of My Speculation on Such Matters at All: But let's say that Texas moves up to a consensus #2 after the necessarily-assumed loss by Tech to OU and then OU moves up to a consensus #3 due to their presumed wins over Tech and OSU.  Let's also say then that OU has an advantage over Texas in the human polls to the same degree as Texas currently has an advantage over OU.  Who leads in the BCS then?  Texas.  And it's not really all that close either.  

Granted, Texas current lead over OU in the human polls is quite small, so OU could indeed develop a bigger lead, but I think that there will be enough confusion among voters between Texas, OU, Florida and USC to keep things relatively close and prevent a consensus from arising.  That is, I think the top 2 among those four teams will have fewer points than expected for their rankings and the bottom 2 will have more points than expected, just as they do now.  And if things remain relatively close in the human polls, even if OU has a lead in them, they will have to gain a LOT of ground in the computers to catch Texas in the BCS.  They have the best schedule they can ask for to do it, but are there enough games left?  I'm just not sure.  

Paradox: But one thing we can be sure of is that nothing is certain.

Quick Note on the BCS if Texas Does Not Make the Conference Championship Game and More Vagarious Speculation:  Obviously, the best way to get into the national championship game for Texas is to win the conference.  It's practically a guaranteed invite for either Texas, Texas Tech or Oklahoma.  But various things can happen to send Texas to Miami without a detour in Kansas City, as Peter has outlined previously.  However, in none of these scenarios is BCS number crunching really necessary as far as Texas goes, at least at this point, because it relies too much on chaos and no one knows how anything is going to react in chaos.  We should revisist those issues if something unforeseen is to happen in the future.  

A few things are safe bets though.  Texas will go to the BCS championship game before USC and before any 2-loss team (even Florida), and likely before a 1-loss Texas Tech and a 1-loss Alabama.  They will not go to the BCS championship game over a 1-loss Florida.  So we're essentially down to one team: Oklahoma.  If Texas is fighting for 1 spot in the championship game with 1-loss Oklahoma, who goes?  I tend to think that if OU wins the Big 12 championship, they are ranked higher, and if neither 1-loss Texas nor 1-loss OU wins it, Texas is ranked higher.  But this is all speculation.  We just don't know what voters and computers will do with chaos.  It's a fool's errand to predict the reactions in times of chaos and tumult.