until Texas vs Kansas

Friday, November 4

 

Under Review: Margin of Victory

How does margin of victory (MOV) affect computer rankings? The BCS did away with MOV as a factor in computer rankings two years ago after USC was left out of the BCS title game in favor of Oklahoma and LSU. The perception was that MOV was skewing the computer rankings to undesirable results. Hence, MOV as a ranking factor was abolished. Many of the people running computer rankings that were participating in the BCS formula withdrew after that change.

Today we’ll take a look to see the effect MOV has on one particular set of rankings—Jeff Sagarin’s. Sagarin offers his rankings in two data sets; the first does not take into account MOV and the second does. There are many other factors that contribute equally to both sets of rankings, but the difference between the two we’re looking at is MOV. Ideal, for this rudimentary study.

The first data set, which he calls ELO-CHESS, is the one the BCS uses, sans-MOV as a factor. The second, which he calls PURE POINTS, or PREDICTOR, does factor in MOV. Sagarin notes the difference on his website:

“In ELO-CHESS, 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. PURE POINTS is also known as PREDICTOR, BALLANTINE, RHEINGOLD, WHITE OWL and is the best single PREDICTOR of future games.”

Now that we’re introduced to the two sets of data, let’s see how they affect the rankings:



Sagarin Computer Rankings
TeamELO CHESSPREDICTOR
Texas12
USC21
VA Tech34
Alabama428
PA State56
UCLA618
OH State75
Wisconsin816
Oregon99
Tx Tech108
Miami1112
W Virgina1230
FL State1320
Michigan147
Colorado1521
Florida1619
LSU1713
GA1826
Notre Dame193
Bos College2022



What lessons can we draw? (Yes, sample size caveats apply. This is one set of data. It’s not the bible of college football rankings. We know, we know. Shut up.)

*Biggest benefactors of the no margin of victory factor: UCLA, Alabama, Wisconsin, West Virginia

*Most penalized by no margin of victory factor: Notre Dame, Michigan

*A few other teams not in the chart above had huge discrepancies their ELO-CHESS vs PREDICTOR scores: Michigan State (#10 predictor vs #26 ELO), Louisville (#11 predictor vs #28), Iowa (#17 predictor vs #36 ELO).

Okay, now things are starting to shape up a bit. What trends are you noticing? I’ll tell you what jumps out at me. I’m not entirely comfortable with the PREDICTOR rankings. Notre Dame at #3? They’re a very good football team—definitely, in my mind, a Top 10 team. But a home loss to Michigan State? That’s troubling. And Louisville at #11? I’m sorry, but that’s not okay either. And speaking of Michigan State, a system that puts them at #10 makes me uncomfortable. You simply cannot lose 49-14, at home, to Northwestern, and be #10.

My theory? Mr. Sagarin’s PREDICTOR rankings skew too heavily toward teams that can put up a lot of points. Notre Dame, Louisville, and Michigan State, each teams loaded with offense but shaky defense, are rewarded heavily. When your offense is good, if the other team struggles, you’re going to blow a lot of people out. Texas Tech fits this mold, too, and—not surprisingly—checks in at #8 in the PREDICTOR rankings. I actually think Texas Tech gets the “overrated” card too often and that they’re a strong, top 15 team. But #8? I don’t think so.


*On the other hand, the ELO-CHESS rankings look a lot better. Alabama at #4 seems a lot better to me than Alabama at #28. Can you imagine the outrage directed at the BCS if the Tide were actually ranked #28 in a computer that the BCS was using? I’m fairly certain Warren St. John would personally strangle Jeff Sagarin, and then whoever is acting BCS commissioner these days.

I haven’t studied enough data, from enough ranking systems, to make any firm conclusions about margin of victory and its place in these systems. But a quick and dirty look at Sagarin’s rankings certainly makes me very hesitant to endorse any such system. Simply put, they just don’t seem fair. Even if, over the long run and huge sets of data, the margin of victory system comes out with a slightly better average prediction, it’s not fair. It’s simply not fair to penalize Alabama that way. Or to reward Notre Dame that way.

This requires much, much more time and thought than I have right now. But it’s a start.

--PB--

Comments:
I hope Peter isn't the only one looking into these numbers. Perhaps Gameday should do a report on this and less time on what kind of jokes coaches are playing on their teams.

I didn't name any names so don't turn this into a bitchfest back and forth. I also think we could to with fewer reports about what kinds of rap music certain coaches listen to on their iPod. Who freaking cares? An SID made that up anyway, just like the well-known Halloween prank was an obvious publicity stunt.

Nice work Peter.
 
I don't understand how MOV only moved Texas Tech up two spots. Are MOV numbers also influenced by strength of schedule?
 
It's a coincidence of sorts. Tech gets a good ELO rank because they only have one loss. They get a good PREDICTOR score because they have blown a lot of people out. If their nonconference schedule weren't so weak, they'd be even higher in the PREDICTOR, I suspect.
 
Also, Tech does not get any help in their rankings by their loss to texas, becasue they got blown out, unlike ND, who gets big style points in the predictor by almost beating USC.
All that this shows is the need for a playoff. I think the guys at Student Body Right are on to something with the plus one. No one out side of the top 4 going into the bowl games has ever gotten a national championship, so the argument for #5 is not as strong. Plus it would be a lot better for my health to know that texas can still win the national championship being in the to 4 as apposed to every week hoping VA Tech blows it or Joe Blow coach or Harris poll voter does not have some bone to pick with Mac Brown.
 
So what is a ranking system then? Are we trying to figure out who has the best team, or are we rewarding a team for winning games and/or putting points up? These two are not the same. Human pollsters and the ELO Chess seem to fall more into the latter category, while the Predictor is trying to fit into the former.
Just one more reason in my opinion that we shouldn’t have polls until 7 or 8 weeks into the season.
I agree that Alabama is probably not 28th, but with the possible exception of Florida, have they beaten anyone this year? Tennessee? South Carolina? Right now, Alabama is the Texas Tech of the SEC. Nobody knows if they are good or not. I expect their ranking in ELO chess to change if they beat Auburn and LSU, which is extremely dubious without Protho.

Brandon '97
 
Sorry, that is TT before playing UT.

Brandon '97
 
For all the college football fans out there try checking out Pokerblue.com They have a free online poker tournament where the football fans, representing 70 of the Nation’s top colleges, battle their rivals at the poker table. Each of the rivalry winners gets two free tickets for their big rivalry game and $100 dollars in tailgate cash, plus they qualify for another free tournament with 2 National Championship Rose Bowl tickets and $2000 in travel cash as the grand prize. Get excited for the big game, vist Pokerblue.com/rivals or check out our blog
 
I wonder if these MOV calculations include the strength of the team being played. Is he giving more credit when Texas Tech beats Bunghole State by 40 than when Texas beats Ohio St by 3? Obviously strength of schedule factors into the entire equation, but I wonder if his formula looks at each individual game and maybe comparing how much a team "should" beat another team by. Seeing as he has a predictor formula that defines this, I wonder if his formula takes that into account.
 
I inclination not agree on it. I over polite post. Particularly the designation attracted me to study the whole story.
 
Nice dispatch and this fill someone in on helped me alot in my college assignement. Gratefulness you for your information.
 
Amiable dispatch and this enter helped me alot in my college assignement. Thank you seeking your information.
 
[url=http://sopriventontes.net/][img]http://tonoviergates.net/img-add/euro2.jpg[/img][/url]
[b]download microsoft office 2007 enterprise serial, [url=http://tonoviergates.net/]management software price comparison[/url]
[url=http://sopriventontes.net/]adobe creative suite 3 master collection[/url] ms office application software educational software downloads
iet discount software [url=http://tonoviergates.net/]autocad light[/url] pc software for sale
[url=http://tonoviergates.net/]2 Content Retail Price[/url] anti virus software for sale
[url=http://tonoviergates.net/]tamu software purchase[/url] best price adobe photoshop cs4 upgrade
Ultimate Mac [url=http://sopriventontes.net/]office 2007 software[/url][/b]
 
Easily I agree but I about the list inform should prepare more info then it has.
 
Amiable fill someone in on and this fill someone in on helped me alot in my college assignement. Say thank you you seeking your information.
 
Folks,
Help!
I'm Looking to purchase [url=http://www.milesgershon.com/tv-stands.html][b]TV Stands[/b][/url] or TV [url=http://www.milesgershon.com][b]Wall Units[/b][/url] For an apartment I'mbuying now.
Can you peoplegive me a good recommendation of where is the leadingdeal on these? I live in New York and I heard that the big thing about these [url=http://www.milesgershon.com][b]tv stands[/b][/url] is the cost of shipping and installation.
I also found this great article about wiring your entertainment center: http://www.helium.com/items/1577888-how-to-wire-your-home-entertainment-center

cheers

[url=http://www.milesgershon.com][img] data:image/jpeg;base64,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 [/img][/url]

Raimy
 
buy xanax online overnight xanax side effects long term - does military drug test xanax
 
Sіmply deѕire to say your articlе is аs amazing.
The clеarness in your pоst іs just great and i cаn
assume you arе an еxρert οn this subject.

Fine with yоuг рermission alloω mе to grаb уour RSS
feed to keеp up to date wіth foгthcomіng ρost.
Thаnκs a milliοn and ρleaѕe keeρ
uр the еnjoyable worκ.
Also visit my weblog : Mid Stakes Promotions
 
This article will assist the internet people for building up new weblog or eνen а weblоg frоm
start to end.

Feel free to visit my homeρage Americas Cardroom Poker Bonus
 
Post a Comment

<< Home |

This page is powered by Blogger. Isn't yours?

Listed on BlogShares Weblog Commenting and Trackback by HaloScan.com < ? Texas Blogs # >