The Real Story About Strength of Schedule

Wow good work London Raider! As usual you continue to amaze with the insightful and cogent review of the "way it is". Good work. SOS is by its nature a beast invented by journalist (IMO) who want to use computer rankings to justify their opinions. The deal here is that we all know good coaching , good recruiting and an organized approach to balancing the student athlete produces a good team. (Ie not cheating/breaking NCAA rules, making the talent successful in your schools setting) That being said the intangible come into play cohesion, leadership, skill positions scheme etc. So in the SOS category everybody needs a fired up opponent to hone your skills against while not exposing your key folks to injury (like we did to OU’s Bradford in 07). It’s essentially a no hold barred slug fest with the eye on repetitions for the starters and their backups. Think dress rehearsal. We have recently had a whole litany of comments on who has played who of consequence. 12th man blogger in particular. Based on air travel any school can play any other. We used to be limited by the bus travel i.e. regional opponents coupled with a 10 game schedule -to large degree dictated who you really could play. So IOT get a set of starters fully trained against a full speed opponent you need a mid-major or lower division team to practice (play) at game speed. Very few teams in BCS division are scheduling a slew of FSU, Penn State, USC and LSU’s. It’s as you said "why do it?" More downside than upside based on the rest of season in our league. Imagine losing 2 x Rajon Henley’s (quality starters) for a season because somebody thought it was cute to play Florida in week 2. Besides the mid majors ie Sun Belt, MAC, MWC etc need the opportunity to gain a rep like Appalachian State did in 06 vs Michigan. I for one don’t have much of a problem playing NM or Marshall or even SD State. BTW does anyone else think Tulsa even blinked when dropping us last year?

Clearly, the top four of the Big-12 (OU, UT, OSU, TTU), besides being at the top now, have shown a habit of playing the tougher out-of-conference foes for six years now. Consider if we had not been dropped by Tulsa (whom I now root against all the time, regardless of opponent), we would show up slightly better, yet. Still, this is a good chart showing that the elite of the conference are there due, in part, to tougher non-conference scheduling. It obviously makes the four atop the list better prepared for the conference challenges. Way to go again, LR. Thanks.

Yet another great post. Where’d you pull these numbers?

and why did you start your ranking with the 2003 season?

this guy has Baylor with the toughest schedule in 2008, anf 10th toughest in the nation

However, rather than repeat myself:

My only criticism…

Hi everyone. Thanks again for the great comments.

I sourced my numbers from www.teamrankings.com. They have a link called Strength of Schedule Power rankings and their data goes back to 2003.

Based on the posts here, I looked at the CBS Sportsline rankings which led me to the Sagarin ratings and a host of other sites that focus on this sort of thing. The results in each case, its safe to say, were not the same. This surprised me. I thought there was a universally accepted approach to calculating strength of schedule.

As it turns out, different services and ranking systems utilize their own proprietary algorithms to calculate strength of schedule and other various power ratings. Some of these rating services use more sophisticated weightings than others. Some weigh certain variables slightly differently.

TeamRankings.com say they completely ignore the Coaches Poll and the AP Poll and look purely at the numbers.

From their website:

“Our algorithmic game predictions provide a 100% data-driven foundation for making educated decisions and sports picks. The sports predictions models that we have developed use a variety of statistical methods, such as power ratings, scoring margins, historically similar matchups, and game simulations. While no data-driven forecasting system is perfect, an intelligently derived, objective prediction system is a key weapon for cutting through biased media and message board hype.”

The website www.nationalchamps.net/ncaa/bcs/strength_of_schedule_explain.htm goes into more detail about how the SOS formula is calculated. This is the approach used by the BCS.

(The SOS) component is calculated by determining the cumulative won/loss records of the team’s opponents and the cumulative won/loss records of the teams’ opponents’ opponents. The formula shall be weighted two-third (66 2/3%) for the opponent’s record and one-third (33 1/3%) for the opponents’ opponents record. The team’s schedule strength shall be calculated to determine in which quartile it will rank: 1-25, 26-50; 51-75; 76-100 and shall be further quantified by its ranking within each quartile (divided by 25). For example, if a team’s schedule strength rating is 28th in the nation, that team would receive 1.12 points (28/25 = 1.12). Should a team play a Division I-AA opponent, only the losses of the Division I-AA team shall be used in determining the opponent’s record or the opponent’s opponents’ record."

Let’s compare the SOS for a few Big 12 between TeamRankings.com and the BCS System for the 2008 season.

TTU
TeamRankings.com: 3
BCS System: 27
Differential: -24

UT
TeamRankings.com: 4
BCS System: 14
Differential: -10

OU
Team Rankings.com: 1
BCS System: 7
Differential: -6

TAMU
TeamRankings.com: 62
BCS System: 30
Differential: +32

Jeff Sagarin has this to say about the BCS approach.

In (the BCS Formula) only winning and losing matters; the score margin is of no consequence, which makes it very “politically correct”. However it is less accurate in its predictions for upcoming games than is the PURE POINTS, in which the score margin is the only thing that matters.

Sagarin publishes the formula used by the BCS as well as a measure called Pure Points as described above.

I don’t know about you (that is if you’ve managed to stick with this post), but I find the win loss approach employed by the BCS overly simplistic. Maybe the BCS approach is the most straightforward way to measure SOS for the purposes of a national poll. Its also is a measure which most sports fans, journalists and coaches best understand. Fair enough.

But let’s say you were going to make a bet based solely on the data provided by both formulas.

Which would give you more confidence as a predictive measure?

Intuitively, I would want more information to be reflected other than just wins and losses (i.e. home/away games, margin of victory, SOS of a conference etc). Still, before answering that question, I would still want a better idea of how TeamRankings.com or similar services calculate its SOS formula. But, I think you probably get the idea.

Thanks again for the great comments and questions. Its helped to make me a more knowledgeable fan, and I hope this discussion does the same for you. Apologies as well if anyone thought the previous post was misleading. That was not my intent.

Go Red Raiders!!

I found out why TTU got a better SOS than UT in 2008: The website apparently doesn’t count FCS games. If you look at their website here: http://www.teamrankings.com/ncf/rankings/?tt=3 it shows TTU as 9-2, which on first examination did not make any sense, but then I figured out that all teams only had their records for FBS games shown. That then allowed TTU’s SoS to get more weight out of Big XII opponents such as UT and OU, simply because instead of there being 61.5% of the games in conference, in conference games jump to 72.7% of the schedule.

TTU4Life,

You are correct that Teamrankings.com does not include games FCS games in determining SOS, but neither does the BCS system. According to the BCS:

Should a team play an FCS opponent, only the losses to the FCS team shall be used in determining the opponent’s record or the opponent’s opponents’ record."

Teamrankings.com seems to ignore both wins and losses to FCS team because their formula is based on the outcomes of the FBS teams (who beat who, who beat who and so on). The possible number of combinations amongst those teams is 6.68950291344913e+198 (in otherwords one hell of a big number).

If you were to add in FCS teams to the mix that would include 143 additional teams – most of which never end up playing an FBS opponent – and needlessly complicate the formula. We can assume that in most cases losses against FCS opponents rarely happen (that’s why the Appalachian State victory of Michigan was such a big deal).

My guess is that when you are trying to create a mathematical formula to determine these sorts of things, there is little value add to including these games in a formula. Nevertheless, its a very good observation.

As for Beergut, I can’t understand why anyone would accept the BCS formula as the best possibility, especially when every professional you read about or speak to considers it to be lowest common denominator for assessing SOS?

Let’s just look at a couple of (major) things:

The formula shall be weighted two-third (66 2/3%) for the opponent’s record and one-third (33 1/3%) for the opponents’ opponents record.

Why assign a two-third weighing for the opponent’s record? Why not 75%? Why not 42.85646%? Its a totally arbitrary weighting.

Why should the opponent’s opponent receive a one-third weighting? Why not 2%? Why not 29.6%?

The team’s schedule strength shall be calculated to determine in which quartile it will rank: 1-25, 26-50; 51-75; 76-100 and shall be further quantified by its ranking within each quartile (divided by 25). For example, if a team’s schedule strength rating is 28th in the nation, that team would receive 1.12 points (28/25 = 1.12).

What’s the value of dividing the team in a particular quartile and dividing it by 25 increments? Can you explain that at all or why it really matters? Why limit it to 25? Why not divide it by 50? Why not divide it by 120?

Switching gears for a minute, I thought I would also weigh in on some of your ‘contributions’ to this blog to which I have only recently become aware. It appears that some of the other bloggers on the site have a longer history with your work.

I’m all for free speech. However, to quote Samuel Johnson, “Every man has a right to utter what he thinks truth, and every other man has a right to knock him down for it.”

Consider me the ‘Other Man.’

This is the thinking fan’s sports blog. You need to raise your game before offering outright dismissals or proclomations.

If you’re going to make sweeping pronouncements, then you need to back them up with something.

I’ve taken the suggestions of some of my fellow bloggers to look at your site. Their comments are accurate. Your site gets almost no traffic or comments despite your being an alumnus of a school that is twice the size of Texas Tech and which has one of the best fan bases with one of the deepest traditions in all of college sports. Something, my friend, does not add up.

I would suggest you spend your team looking at this blog and a few others for motivation and best practices. Rather than waste your time here, you should try to apply those lessons to your own site and try to build something that rivals what Seth C has done with this forum.

Otherwise, my second suggestion would be to turn over the keys to someone who has the wherewithal to build a blog worthy of Texas A&M, its great traditions and its passionate fans.

At the end of the day its Seth C’s site. Like one of the other bloggers noted in another post, making comments on this site is akin to being invited into someone’s living room. There is a certain respect you should show to Seth C and his other guests. You, sir, repeatedly seem to violate this unspoken code.

I won’t speak on Seth C’s behalf, but I will say that you should demonstrate to the contributors and Seth C more courtesy than you have shown.

Having our posts and comments challenged is one thing. None of us seem to have a real problem with that, but having to constantly endure your boorish tone and unsubstantiated comments is tiresome and unwelcome.

Go Red Raiders!!