October 8, 2015

SUR Algo Picks – Week 6

A 16-10 week 5 gives me reason for optimism, but another losing week against the FPI adds additional doubt.  Looking at the bigger picture, I’m really only disappointed at the 60%-69.9% range where a couple more correct would help validate the assigned probabilities, but there are few opportunities to do so this week.

The small sample size in the 80-89% range will be doubled this week and I need all 5 correct to reach the 80% mark.

SUR Week 5 Results


Podcast: Irish rewind, Georgia Tech overview, picks and score

I’m joined by Brandon Rink and Brad Senkiw of orangeandwhite.com to discuss takeaways from the Notre Dame game, preview Georgia Tech, go around the ACC and pick the Clemson/Georgia Tech winner and score.

You can download and subscribe via iTunes here or at the bottom of this page for iTunes or Android, play in this window or play in a new window.

Podcast artwork courtesy Eargle Design.

Receiver Metrics – 4 Games

Artavis Scott continues to lead the receivers and Hunter Renfrow catches everything thrown to him (and almost one that wasn’t).  Two veteran receivers, Charone Peake and Germone Hopper were all but invisible against Notre Dame.

It almost seems like the “rotation” has come down to Scott,  and an occasional throw to Leggett and McCloud, with one a game in Renfrow’s direction.

Running backs were targeted 61 times in 2014 (4.7 times per game) and have only been targeted 6 times (all Gallman) in 4 games.Receiver Detail 15 4

Plays, tempo, explosives all down

Your eyes are not playing tricks on you Clemson fans.  If it looks like the Tigers are running fewer offensive plays, at a slower pace and aren’t as explosive as last year, it’s because….they’re running  fewer offensive plays at a slower pace and aren’t explosive as last year.

The graphs and charts below compare the first four games of 2014 and 2015.  Obviously the opponents are difference, but you can make an argument that both years consists of 2 good defenses (Georgia and FSU in 2014, Louisville and Notre Dame in 2015) and 2 not so great defenses (South Carolina State and North Carolina in 2014 and Appalachian State and Wofford in 2015).

Jeff Scott intimated Monday that any issues with the offense were related to: a) the first two opponents, b) the short week before Louisville, and c) the weather vs. Notre Dame.  While I can certainly see some of that, it’s plain to see that the offense is not as explosive as last season.

Through 4 games the Tigers have run 36 less plays and have 10 less explosive plays than they had last season at this point.  The percentage of explosive plays is only 1.8% below last season, but because of the slowing of the offense Clemson is on pace to have 30 less explosive plays than in 2014.  The loss of Mike Williams 11 offensive plays into the 2015 season is an obvious reason for a decrease in explosiveness, but is it the only reason or does Scott’s explanation make more sense?1415 Comps 4                                                                                                            * Pass plays include sacks

2015 4 Tempo

There’s 8 games left in the regular season and plenty of time to ramp up the tempo and break out the explosives we are used to while watching the Tigers and the coming weeks will tell us if Scott’s explanation makes sense or if there’s another reason for the decreases.

Third Down D – Conversion rates, averages and who makes stops

Below are the numbers for third down defense on the season.  Notre Dame faced an average distance to go on 3rd down of 10.8 yards and for the season Clemson opponents have averaged 3rd and 8.6 yards to go.3rd Down D 2015 4

Here’s who’s making those stops on third downs to prevent first downs.
3rd Down D Tackles 15 4

SUR 10 Week 5

Our panel –

Brandon Rink – OrangeandWhite.com multimedia editor for the Anderson Independent-Mail.
Chris Cox – SUR Recruiting Analyst and contributor.
Nic Mills – Intern in the Clemson Athletic Communications Office and SUR contributor.
Marty Coleman – SUR editor.

The voting is done independently and without the knowledge or influence of how the others voted, tabulated by our accounting firm and hermitically sealed in a #10 envelope until unveiled each week.

The Results:

SUR 10 2015 5

Also receiving votes: Oklahoma (3), Stanford (2), Florida (1), Notre Dame (1)

Deshaun Watson by Distance – 4 Games

Complete through Notre Dame.

Watson ND and Season 2015-4

SUR Week 5 Algo Picks

While the results below may not look stunning, remember one measure of a model such as this is to have the actual results fall in line with the projected results. To that extent the algorithm has succeeded, with the exception of the 80-89% range (tiny sample size).

Head to head against the ESPN FPI? Not so much.SUR W4 ResultsSUR W5

Podcast: Clemson vs. Notre Dame

I’m joined by the Killer B’s, Brandon Rink and Brad Senkiw of orangeandwhite.com to talk Clemson football. With no previous week’s game we focus on Clemson/Notre Dame with a side trip to our ACC picks of the week including the undefeated Miami Hurricanes.

Some of the topics we traverse:

  • Clemson defense vs. Notre Dame offense
  • Clemson offense vs. Notre Dame defense
  • Special Teams
  • Pick and score

You can download and subscribe via iTunes here or at the bottom of this page for iTunes or Android, play in this window or play in a new window.

Podcast artwork courtesy Eargle Design.

Figure The Odds: Clemson/Notre Dame

The probabilities below are derived from an algorithm based on 2,967 college football games between two Football Bowl Subdivision teams and includes 18 metrics for both the winners and losers of those games.

The algorithm endeavors to identify metrics that are statistically relevant to winning and losing (a few of which are included in the graphic below), estimate the statistics each team will accumulate in a game (based on opponent) and use those estimates to derive a win probability.

Over the course of those nearly 3,000 games a few statistics stand out, namely yards per play, total yards differential and turnovers. 78.9% of the time the winning team averages more yards per play than the loser and 77.5% of the time the winner has more total yards than the loser.

The turnover stat is a little more nuanced in that turnovers are often equal (22% of games), which means instead of saying “We have to win the turnover battle” a coach might be more accurate saying, “We can’t lose the turnover battle.” Winners are even or + in turnovers 80.2% of the time.

CLEMvsND Prob 15 4

The estimates show that Clemson is likely to gain more yards than Notre Dame and have a tiny advantage in yards per play. Given the statements above why does Notre Dame have a win probability of 52.1%? In a word, turnovers. The Tigers are projected to have 2 and Notre Dame 1, leaving the Tigers -1 in this crucial metric.

As currently constructed, the algorithm reduces a team’s win probability by 8.8% for each turnover. If the Tigers were able to stay even in the turnover battle the win probability would flip to 56.7% in Clemson’s favor.

Footballs take funny bounces sometimes and it’s a tough proposition to realize that a year (or more) of preparation, work, sweat and tears could come down to a tipped pass or whether a ball rolls out of bounds instead of into the arms of the opponent, but that very well could end up being the case Saturday.

The model has worked decently in 2015 to date. For the games with a win probability between 50.1 to 59.9% the algorithm has picked the winner 54.8% of the time. One measure of a model is getting results that align with the probabilities generated and that has happened for this range over the first 4 weeks of the season.

Take heart Clemson fans, both the ESPN FPI and FEI Game Projections predict more favorable outcomes for the Tigers.