Writers of Pro Football Prospectus 2008

Most Recent FO Features

JohnsonKer18.jpg

» Seventh Day Adventure: Week 13

The biggest game this week is the Iron Bowl, where the playoff hopes of Alabama, Auburn, and Georgia hang in the balance.

17 Oct 2014

Varsity Numbers: F/+ vs. Reality

by Bill Connelly

The seven-week mark is a special occasion for us in terms of FO college football stats. After seven weeks, the preseason projections have been completely phased out, and we post the first iteration of the F/+ ratings that include offense, defense, and special teams breakouts.

There will still be plenty of shifts and changes over the second half of the regular season, but I thought it would be interesting to take a look at which teams are doing what F/+ thought they would do, and which teams are dancing to the beat of an unforeseen drum.

I'm going to take a look at this in three different ways that interest me: comparing current F/+ ratings to preseason projections, looking at weekly picks (to see on whom F/+ is zeroing in, and about whom it still has no idea), and retrofitting -- using the current ratings to project previous games.

Preseason projections

Here's the final projected top 10 of the preseason. Current record and F/+ ranking are in parentheses.

1. Florida State (6-0, 14th)
2. Alabama (5-1, 10th)
3. Oregon (5-1, sixth)
4. Stanford (4-2, eighth)
5. South Carolina (3-3, 42nd)
6. LSU (5-2, 35th)
7. Oklahoma (5-1, fourth)
8. Auburn (5-1, third)
9. Michigan State (5-1, 19th)
10. USC (4-2, 17th)

So basically, there have been two outright whiffs (South Carolina, LSU), three underachievers (FSU, Michigan State, USC), and five teams that were, and are still, in the top 10.

In all, 82 of 128 teams are currently within about 10 percent of their preseason projected F/+ rating. That's not bad, but there have certainly been some impressive misses. There always are.

Far above projection

  • Ole Miss (+24% -- projected 25th, currently first)
  • Mississippi State (+22% -- projected 27th, currently second)
  • TCU (+21% -- projected 39th, currently fifth)
  • Georgia Southern (+20% -- projected 123rd, currently 67th)
  • Rutgers (+19% -- projected 69th, currently 24th)
  • Virginia (+19% -- projected 64th, currently 23rd)
  • Colorado State (+19% -- projected 85th, currently 48th)
  • Marshall (+19% -- projected 58th, currently 20th)

Apparently hiring a new offensive coordinator (TCU, Rutgers) or simply residing in Mississippi was key to overachieving this year. Well, except for you, Southern Miss. Sorry.

Far below projection

  • Vanderbilt (-22% -- projected 59th, currently 111th)
  • Fresno State (-22% -- projected 56th, currently 109th)
  • North Texas (-24% -- projected 78th, currently 124th)
  • Bowling Green (-26% -- projected 53rd, currently 118th)
  • SMU (-27% -- projected 83rd, currently 127th)

If you are projected 83rd and still become the biggest underachiever of the first half of the season ... well, that's awfully impressive, SMU. Meanwhile, Fresno State and North Texas lost quite a bit of breakthrough talent from last year's teams, and Vandy and BGSU changed coaches. That at least sort of explains the difference.

Weekly picks

Right now, the win probabilities I share in each week's F/+ picks post at Football Study Hall are hovering almost exactly where they should, which is exciting.

  • Teams given a 50 to 59 percent chance of winning in a given week are currently 18-19 (49 percent).
  • Teams with a 60 to 69 percent chance are 66-40 (62 percent).
  • Teams with a 70 to 79 percent chance are 91-22 (81 percent).
  • Teams with an 80 to 89 percent chance are 88-15 (85 percent).
  • Teams with a 90 to 99 percent chance are 68-2 (97 percent).

One good week for the 50-percenters and one iffy week for the 70-percenters, and everything will be perfectly in line.

That said, F/+ still has a pretty iffy read on quite a few teams. Below are teams which weekly F/+ projections either haven't figured out or figured out a few weeks in.

Far above projection

  • Colorado State: +2.1 wins (2.9 projected wins, 5-1)
  • Mississippi State: +1.8 wins (4.2 projected wins, 6-0)
  • Minnesota: +1.7 wins (3.3 projected wins, 5-1)
  • Notre Dame: +1.7 wins (4.3 projected wins, 6-0)
  • Marshall: +1.7 wins (4.3 projected wins, 6-0)
  • Rutgers: +1.6 wins (3.4 projected wins, 5-1)
  • Washington: +1.6 wins (3.4 projected wins, 5-1)
  • Ole Miss: +1.5 wins (4.5 projected wins, 6-0)

Being that Ole Miss and Mississippi State are now the No. 1 and 2 teams in the F/+ ratings, I'd say they won't have to worry too much about being undervalued anymore. A team like Notre Dame, however, is still hovering in the teens at 6-0. The Irish have a good chance of righting wrong ratings at Florida State on Saturday.

Far below projection

  • Kent State: -2.3 wins (2.3 projected wins, 0-6)
  • Idaho: -2.0 wins (2.0 projected wins, 0-6)
  • Ball State: -1.9 wins (2.9 projected wins, 1-5)
  • Navy: -1.8 wins (4.8 projected wins, 3-4)
  • UConn: -1.7 wins (2.7 projected wins, 1-5)
  • Miami (Ohio): -1.7 wins (2.7 projected wins, 1-6)

Ball State and Navy confuse me (and the numbers) this year. The Cardinals, one of the better recent MAC programs, have played atrocious defense and have managed to lose three games by eight points or less, including losses to both Iowa and Indiana State. Navy, meanwhile, ranks a solid 56th at 3-4, ahead of 5-1 Kentucky, 4-1 Oregon State (now 4-2, I guess), and others. Losses to Air Force and Western Kentucky weren't very impressive, though competitive losses to Ohio State and Rutgers and a win over Temple look pretty decent.

Retrofitting

So what happens if we take the current ratings and fit them into the season's results? Which teams fit pretty nicely, and which are still a mystery?

Far above projection (overachieving their ratings)

  • Bowling Green: +2.4 wins (2.6 projeted wins, 5-2)
  • Wyoming: +1.8 wins (1.8 projected wins, 3-3)
  • Fresno State: +1.4 wins (1.6 projected wins, 3-4)
  • UTEP: +1.3 wins (1.7 projected wins, 3-3)
  • Ohio: +1.3 wins (1.7 projected wins, 3-4)
  • UL-Monroe: +1.3 wins (1.7 projected wins, 3-3)

Far below projection (underachieving their ratings)

  • Idaho: -2.0 wins (2.0 projected wins, 0-6)
  • Miami (Ohio): -1.6 wins (2.6 projected wins, 1-6)
  • Kent State: -1.5 wins (1.5 projected wins, 0-6)
  • Ball State: -1.4 wins (2.4 projected wins, 1-5)
  • UMass: -1.3 wins (2.3 projected wins, 1-6)

Lots of underachievers in the MAC, evidently.

Teams on the overarchieving list are a combined 12-2 in games decided by one possession, while teams on the underachieving list are a combined 1-14. That ... probably isn't a coincidence. I'd say to expect the underachievers to turn things around, but when you're in the same conference and you're playing each other, BOTH teams can't win future close games.

Actually, let's take that one step further.

Power conference teams far above projection (overachieving their ratings)

  • Oregon State: +1.1 wins (2.9 projected wins, 4-1)
  • Washington: +1.0 wins (4.0 projected wins, 5-1)
  • Arizona: +1.0 wins (4.0 projected wins, 5-1)
  • Florida State: +1.0 wins (5.0 projected wins, 6-0)
  • West Virginia: +0.9 wins (3.1 projected wins, 4-2)

Power conference teams far below projection (overachieving their ratings)

  • Pittsburgh: -1.0 wins (4.0 projected wins, 3-3)
  • Stanford: -0.9 wins (4.9 projected wins, 4-2)
  • South Carolina: -0.8 wins (3.8 projected wins, 3-3)
  • Tennessee: -0.7 wins (3.7 projected wins, 3-3)
  • Colorado: -0.7 wins (2.7 projected wins, 2-4)

Teams on the overachieving list are 12-2 in one-possession games, and teams on the underachieving list are 3-10. Hmm ... I think we may be on to something here ... we might need to pursue this a little bit further in coming weeks.

Quick note

As is customary for the eight-week mark, on the Offensive and Defensive S&P+ pages I switched from unadjusted success rate and IsoPPP numbers to the opponent-adjusted set of numbers. A lot of people have been asking me to continue posting the raw data, and I'll take that as a sign that people are paying attention and start posting two data tables on each page. Thanks!

This week at SB Nation

Another really fun (and lengthy) week of topics for me at SBN.

Monday
The 119-point TCU-Baylor game, scored as a 12-round heavyweight fight
College Football Playoff committee is BCS-esque at avoiding helpful stats

Tuesday
The Numerical, Week 7: Oklahoma defies statistics, and UCLA loves them
2014 Advanced box scores
Georgia 34, Missouri 0: Beyond the box score

Wednesday
Projecting the Big 12's 2014 conference race using advanced stats
Updated SEC conference projections through 7 weeks
Updated Big Ten projections through 7 weeks

Thursday
4 keys in the Notre Dame vs. Florida State put-up-or-shutup game
College football projections: Week 8 F/+ picks

Friday
We finally got to watch Bob Stitt's offense. It was spectacular.
One key each for Texas A&M-Alabama and the SEC's 4 other big games
Missouri at Florida preview: Tigers need passing downs success and a short memory

Posted by: Bill Connelly on 17 Oct 2014

2 comments, Last at 21 Jan 2016, 10:42am by Keshavz1

Comments

1
by Tomlin_Is_Infallible :: Sun, 10/19/2014 - 6:15am

I was really skeptical of the TCU prediction (after the coaching blunder meltdown last week, especially) - I expected a tight win at best.

But wow, you nailed that one. Absolutely dismantled OkSt, and the score was closer than the game was. Probably the 2nd most dominating beatdown of the season amongst actually supposed 'peer' level teams{no cupcakes}.

(runner up to another game this week, Bama-aTm)
Now that one... you whiffed on. as did everyone else in the country so .....

Anyways, I'm glad that if you were going to go 1 outta 3 on the ranked Big12 matchups this week, you nailed the TCU call.

/ homer
// go Frogs

--------------------------------------
The standard is the standard!

2
by Keshavz1 :: Thu, 01/21/2016 - 10:42am

how to understand it easily can anyone tell me. It is not easy for me to understand. This is great post though.
http://www.freeapkfiles.com/link2sd-plus-apk-download-latest-version-for...
http://www.freeapkfiles.com/towelroot-apk-download-latest-version-for-an...