Comparing the final Football Outsiders F/+ projections to the AP poll and conventional wisdom.
22 Oct 2010
by Bill Connelly
Heading into last night's Oregon-UCLA game, I hoped that the Ducks would either lose to the Bruins or humiliate them. A loss would verify the Ducks' oddly low S&P+ ratings, and a blowout win would begin to rehabilitate the rankings. Obviously the latter took place, and I'm happy about that. As I mentioned last week, having odd S&P+ rankings at the end of October is not a rarity -- it typically portends future struggle/success, but that isn't always the case. Oregon's tough remaining schedule would tell us all we needed to know about the Ducks. All we need to know now is that they passed their first test.
With the first 2010 iteration of Brian Fremeau's Field Position Advantage getting released this week, I felt it was a good time to follow up on the "Four Truths" column I wrote a few weeks ago. One of my favorite aspects of basketball stats is that the Four Factors concept has begun to spread. It makes intuitive sense, and it is easy to follow. Obviously things like "leverage rate" have not made it into the typical college fan's lexicon, but while I look at the best way to weight each of these factors (and determine whether these are the four "truths" most worth following), let's take a look at how to rank teams according to each of the four truths.
I ran into some disagreement with this one when I discussed it, but when you look at run-pass splits on standard and passing downs, you quickly begin to get a grasp for how much a game is defined in the first half. As I mentioned in the original Four Truths column, we remember the games that are won in the fourth quarter; but your odds of pulling off a fantastic finish are drastically diminished if you lay an egg in the first half. That is, unless you are a good deal more talented than your opponent. You certainly can win a game in which you are terrible in the first half, but your overall odds decrease considerably.
The easiest way to measure first-half performance with our FO advanced stats would be to derive a 1st Half S&P+ figure.
Top Five Teams According to 1st Half S&P+
1. South Carolina
2. Ohio State
4. Boise State
This one is much easier to accept. You can win games with a grind-it-out, three-plays, 10-yards attack but it is difficult.
An obvious measure for explosiveness is the opponent-adjusted version of Points Per Play (PPP). Add together a team's Offensive and Defensive PPP+, and voila!
Top Five Teams According to PPP+
1. Ohio State
3. South Carolina
Last week in College Station, Missouri went 3-for-12 on third downs (25 percent). They held Texas A&M to just 6-for-19 (32 percent), but they still lost the third down battle. They also won by 21. Third downs didn't cost them because they didn't face many third downs. They were staying in standard downs, getting good yardage on first downs, and staying out of awkward down-and-distance situations. Their leverage rate was 73.6 percent, and A&M's was 61.7 percent, and they dominated. Obviously one can find anecdotal evidence to back up any claim they want to make, but leverage is so closely tied to wins and losses that it is hard to ignore.
The best way to rank teams according to leverage would be to subtract their defensive Leverage Rate allowed from their offensive Leverage Rate, creating a Leverage Margin.
Top Five Teams According to Leverage Margin
3. Boise State
As mentioned above, Fremeau's FPA is the best available tool for measuring the impact of field position on a game. It's one of my favorites.
Top Five Teams According to FPA
Army shows up in the top five of two of these lists, which potentially explains why the Cadets are 4-3 despite ranking 62nd in F/+. Are the overall ratings not giving them enough credit? Is their field position and leverage standing a temporary mirage? We shall see. They play three nice "barometer" teams in the coming weeks (Air Force on Nov. 6, Notre Dame on Nov. 13, and Navy on Dec. 11).
So what if we ranked teams according to the sum of their rankings in these four categories? We get a list with a handful of peculiarities (Central Florida is quite high, though they did throw a major scare into Kansas State not very long ago) and some potential.
|Top 25 Teams According to Four Truths Rankings
|2||Boise State (6-0)||285.9||4||269.2||6||+.138||3||.536||21||34|
|5||Ohio State (6-1)||317.2||2||333.7||1||+.066||20||.535||23||46|
|8||Virginia Tech (5-2)||249.9||29||243.8||19||+.118||6||.547||12||66|
|20||Central Florida (4-2)||225.4||48||208.6||51||+.108||7||.560||6||112|
|21||Mississippi State (5-2)||245.4||35||216.6||45||+.093||15||.537||19||114|
|23||South Carolina (4-2)||324.2||1||278.4||3||+.027||48||.495||69||121|
|24||Michigan State (7-0)||254.3||22||257.9||12||+.058||25||.499||64||123|
|25||Florida State (6-1)||250.4||28||227.1||31||+.060||24||.515||45||128|
|32||Oklahoma State (6-0)||252.4||23||222.9||37||+.043||36||.512||50||146|
And no, Oregon doesn't fare well on this list either. Their low S&P+ rankings continue to baffle, especially in light of their extraordinary FEI rankings. Here's where I continue to tell myself it will all work out correctly in the end.
(One thing to note: If I expanded the definitions of what constitutes a "close" game, Oregon's S&P+ ratings would rise. As soon as I see what happens to overall ratings and win correlations if I do that for previous years, I might make a change to the overall formulas in that regard. But it has to make sense for all data, not just this odd year.)
In all, I love the idea of giving people a "four factors" way of looking at games, and though we might end up moving toward more obvious factors when all is said and done (turnover points, for one, will probably play a role), I look forward to exploring this more in the offseason.
Now, to move to a list that continues to frustrate. This week's F/+ Top 25 is below, and we have quite the confounding top team. With the two systems -- S&P+ and FEI -- disagreeing on so many teams to date, one two-loss team both systems like moved to the top.
Included this week: teams' change in F/+ rating from last week.
|F/+ Top 25 (After Seven Weeks)|
|1||South Carolina (4-2)||+28.7%||+1.4%||274.2||3||.253||5|
|2||Ohio State (6-1)||+28.6%||-7.5%||288.5||1||.181||19|
|3||Boise State (6-0)||+28.2%||+1.1%||270.9||4||.258||4|
|9||Michigan State (7-0)||+23.7%||+4.8%||255.4||9||.242||7|
|F/+ Top 25 (After Seven Weeks)|
|12||Virginia Tech (5-2)||+22.1%||+1.6%||251.1||14||.232||9|
|16||Oregon State (3-3)||+19.8%||-0.9%||245.6||17||.212||14|
|F/+ Top 25 (After Seven Weeks)|
|21||North Carolina (4-2)||+14.4%||+3.3%||236.0||29||.151||23|
|22||Oklahoma State (6-0)||+14.0%||+2.3%||232.6||31||.158||22|
|23||N.C. State (5-2)||+13.4%||+0.1%||223.5||42||.191||18|
|24||Arizona State (3-3)||+13.3%||+4.5%||217.3||48||.219||13|
26. Notre Dame (4-3), 27. Pittsburgh (3-3), 28. Florida State (6-1), 29. Michigan (5-2), 30. Illinois (3-3), 31. Arkansas (4-2), 32. Florida (4-3), 33. Hawaii (5-2), 34. Kentucky (4-3), 35. West Virginia (5-1), 36. Clemson (3-3), 37. Utah (6-0), 38. Mississippi State (5-2), 39. Georgia (3-4), 40. Central Florida (4-2), 41. San Diego State (4-2), 42. Washington (3-3), 43. Ole Miss (3-3), 44. Texas A&M (3-3), 45. East Carolina (4-2), 46. California (3-3), 47. Air Force (5-2), 48. Nevada (6-1), 49. Cincinnati (3-3), 50. Kansas State (5-1).
Biggest Positive Change From Last Week: San Diego State (+8.0%), East Carolina (+7.9%), Texas (+7.5%), Missouri (+6.7%), Louisiana Tech (+6.4%).
Biggest Negative Change From Last Week: California (-12.8%), Kansas (-10.0%), Ohio State (-7.5%), Vanderbilt (-7.2%), UNLV (-7.2%).
Ratings Differences Between FEI and S&P+:
In some seasons, everything plays out according to plan. The top teams are the top teams, and there is little debate. In other seasons, like 2007 and 2010, there is chaos. S&P+ and FEI disagree by 10 spots or more for six of 15 teams in the F/+ top 15. It has possibly never been more difficult to figure out who the best team in the country truly is.
Arizona State managed to improve by 4.5 percent and get into the F/+ Top 25 without playing a game last weekend. It was the perfect confluence of events, really. Two recent opponents (Washington and Oregon State) played each other, but while Oregon State's F/+ ratings regressed by 0.9 percent by losing, Washington's went up by 2.5 percent. That helped the Sun Devils' rating a bit, and Wisconsin's upset of Ohio State helped out a lot. Plus, it appears that the official phasing out of preseason projections helped them as well. They should not count on getting this much help in the future, but it was a nice regulation of their overall numbers.
Again, just for fun, here is what the BCS standings would look like if derived from a combination of 60 percent AP poll rankings, 40 percent F/+ rankings. Neither are part of the BCS formulas, so it is a perfect fit.
1. Boise State (Real BCS Rank: 3)
2. Auburn (4)
3. TCU (5)
4. Alabama (8)
5. Oklahoma (1)
6. Oregon (2)
7. Ohio State (10)
8. Michigan State (7)
9. LSU (6)
10. Stanford (12)
11. South Carolina (21)
12. Nebraska (16)
13. Wisconsin (13)
14. Missouri (11)
15. Iowa (15)
Kentucky over Georgia. Spread: Georgia -3.5 | F/+ Projection: Kentucky by 4.8. I have begun tinkering with the best way to incorporate momentum into the picks formulas, and if I were to figure out a good way to do so, it would likely help Georgia. The Bulldogs have beaten two SEC East opponents (Tennessee and Vanderbilt) by a combined score of 84-14 and have emerged, somehow, as a potential East favorite despite that inconvenient 3-4 record. But by judging both of these teams on their entire bodies of work, the home Wildcats get the nod. With a win, Kentucky would be, at worst, a game out of the East lead themselves.
Missouri over Oklahoma. Spread: Oklahoma -3 | F/+ Projection: Missouri by 6.5. The S&P+ rankings love Missouri (I swear that was not intentional); that will be put to the test when the Tigers host the Sooners Saturday night. You never bet against a streak, and Oklahoma has quite an impressive one going against Mizzou at the moment (last Mizzou win: 1998), but the numbers know none of that.
Arizona State over California. Spread: California -3.5 | F/+ Projection: Arizona State by 3.9. The "What the ...?" team of the week faces on last week's "What the hell was that?" team. California has had trouble with bad momentum turning into an avalanche. If they are reeling too much from their destruction at the hands of USC last weekend, they could easily fall to the Sun Devils.
Colorado over Texas Tech. Spread: Texas Tech -2.5 | F/+ Projection: Colorado by 5.8. This pick has more to do with how far Texas Tech has fallen than anything Colorado has accomplished lately.
As expected, Miami fell back to the pack a bit this week following a win over Duke that was never in doubt, but never really impressive. They fell from sixth to 14th in this week's F/+ rankings, and their rating might be in the process of regulating a bit after seeming artificially high the last couple of weeks.
I asked our friend Josh Darrow, pre-game host and sideline reporter for the University of Miami, a few Hurricane-related questions this week.
Bill Connelly: Miami fell to 14th in our F/+ rankings this week after their defeat of Duke. You've seen them up close and personal -- does 14th sound about right, or do you think they still have more to prove? Because of the defense, Miami's S&P+ ranking has been higher than it seems they deserve, but is this still one of the 15-20 best teams in the country?
Josh Darrow: I think they are closer to a top 25 team than a top 15 team so far this season. But, with the opportunity that lies ahead, they certainly can move up with games left against UNC, Georgia Tech and Virginia Tech, and a possible ACC Championship game as well. Run the table, and Miami looks more like a top 15 team.
I think Miami is doing a lot of things well this year, especially on defense. They have 18 takeaways to go along with 20 sacks, and UM leads the country in tackles for loss. They are getting great play from linebacker Sean Spence, conerback Brandon Harris and safety Vaughn Telemaque, and they have depth along the defensive line.
The offense has been inconsistent, and clearly was plagued by turnovers early in the season. Also hurting them were dropped passes and penalties, especially pre-snap yellow flags that put the offense in some compromised down-and-distance situations. On the other hand, a young and inexperienced offensive line has held up pretty well, and Miami is getting great play from Leonard Hankerson and Damien Berry. What's missing is the home-run threat from the running back spot. Craig Cooper is just getting back on the field after a devastating knee injury in the Champs Bowl and Lamar Miller has missed the last two games with an injury.
BC: The Hurricanes' maintain a strong ranking because of their defense -- second in overall Defensive S&P+, fourth against the run, sixth against the pass, first on standard downs. Last year, they ranked 24th. What do you think has been the major cause for the defense's improvement?
JD: First, you have to start with continuity in the coaching staff. Until this year, Miami has had a different coordinator in each of Randy Shannon's first three years. Now, Miami is in Year No. 2 under defensive coordinator John Lovett. Second, Miami has quite a few players who have been in the program two to three years as well.
With Miami leading the country in tackles for loss, it would seem that Miami's opponents are facing more third and long type situations, and the Canes secondary has responded with 12 interceptions, which is tied for third behind Alabama, Florida and Bowling Green. Last year, Miami had a total of nine picks for the entire season.
BC: Once again, passing downs performance appears to be dragging down Miami's offensive ranking. The Hurricanes rank 28th in Passing S&P+ and 29th on standard downs, so the play-calling and early execution still seem decent. But they rank 69th on passing downs. What tends to trip up Jacory Harris and company on the second- and third-and-long situations?
JD: I think a few things have contributed to this. One is the penalties -- Miami is putting themselves in too many third-and-long-type scenarios. I also think Miami seeks out the big play, but as we all know, those are high reward but not high percentage plays. Miami can be impatient at times, and I think that throws off the consistency in execution. In addition, they have been plagued by drops from receivers, tight ends, and running backs. The number is somewhere in the 30s through six games. That is a high number that has negatively impacted Miami's performance.
In honor of the Four Truths, of course ...
"Four Corners," by The Sea and Cake
"Four Day Worry Blues," by Leadbelly
"Four Days," by Counting Crows
"Four Horsemen," by The Clash
"Four in One," by Thelonious Monk
"Four Night Rider," by The Rural Alberta Advantage
"Four Sticks," by Led Zeppelin
"Four Winds," by Bright Eyes
"Four Women," by Nina Simone
"Fourth Time Around," by Bob Dylan
And, of course, all of Blues Traveler's Four and everything the Gang of Four ever wrote.
As much as we tend to complain about certain aspects of the sport we love, it is important to also recognize when somebody gets something just right. And the Pac-10 (soon to be Pac-12) got its six-team divisions just right. They didn't overthink the issue with the complicated "zipper" method (Washington in one division, Washington State in the other; USC in one, UCLA in the other), they just broke things into North and South. Plus they got creative: the conference title game will take place on the home field of the "No. 1 seed" in the title game. No rotating host sites, no Stanford-versus-USC matchups in Seattle (or Oregon-versus-Colorado matchups in Los Angeles) -- no fuss no muss.
It appeared the conference was going to go this way for a while, of course. The only question mark was what to do with the four "central" teams -- Cal/Stanford and Utah/Colorado. In the end, they sent Utah and Colorado to the south, and that was that.
I thought I would look at what potential title game matchups would have looked like in previous years had this conference setup existed. These matchups were in no way derived by numbers or projections -- they are just rough guesses. Maybe Colorado makes the title game in 2002, maybe Oregon makes it over Oregon State in 2008. The point was not to nail these.
2002: No. 6 Washington State at No. 4 USC
2003: No. 15 Washington State at No. 3 USC
2004: No. 4 California at No. 1 USC
2005: No. 6 Oregon at No. 1 USC
2006: No. 18 California at No. 5 USC
2007: Oregon State at No. 7 USC
2008: Oregon State at No. 6 Utah
2009: Arizona at No. 7 Oregon
USC would have had a monopoly on the conference from 2002 to 2007 or 2008 regardless of the alignment, but it will be interesting to see how these title games take shape in upcoming years, where Oregon is the power program instead of USC. Regardless, I love these divisions, and I love their title-game choice. As Yahoo!'s Dan Wetzel put it on Twitter yesterday, Pac-10 commissioner Larry Scott has had the best freshman year since Carmelo Anthony.
18 comments, Last at 25 Oct 2010, 1:10pm by Jeff Fogle