November 1, 2014

271 Losses for Favorites Since Clemson’s Last

Clemson Upsets LSU

It could happen next week. If not then, surely there’s a chance it happens in Atlanta on November 15, or maybe in Clemson two weeks after that.

At some point Clemson is going to lose a football game in which it’s favored. It’s something that hasn’t happened since November 24, 2012. Since that day the Tigers have taken the field as a favorite against FBS teams 12 times and won each time.

Big spreads, small spreads, medium spreads, the Tigers have won them all. They haven’t all been pretty, covered the point spread or been blowouts, but they’ve all been wins.

During that time 109 teams have combined to lose 271 times as favorites. Big names and nobodies, SEC teams and AAC teams. Upsets, upsets, upsets. Bluebloods, also-rans and “who is that’s” have all been “upset”.

But not Clemson.
Losses as Fav 2014 9Virginia Tech’s demise has been well-documented and often televised. The once vaunted Hokies have lost 6 times as a favorite in this time range. They are not alone.

The PAC 12 – the conference that has become the darling of the media as the second best conference in the country – has two teams that have also lost 6 games each when favored during this time – Southern Cal and Stanford.

Stanford, the team often praised by college football media as a model of consistency has lost 3 times as a favorite THIS SEASON. Southern Cal has also lost 3 times in 2014 as a favorite – 6 times in 9 weeks these teams have been upset in 2014 (including a Southern Cal upset of Stanford).

Alabama, the gold standard for college football teams, has lost 3 times as a favorite in that time.

So, it’ll happen, Clemson fans. The Tigers will lose as a favorite, perhaps this season. When they do the college football world will chuckle, point their fingers and say, “Clemson being Clemson”.

As my buddy Todd Snider says, “Remember, when you’re pointing at someone, there’s 3 fingers pointing back at you.”

Or, in the case of Virginia Tech, Southern Cal and Stanford, 6 fingers.

Editor’s Note: Florida State and Duke are other ACC teams without a loss as a favorite in this time frame.

Tigers Protecting the Ball

CJ-Davidson-fumble-vs-Clemson

The fumble at Florida State was extremely costly, like division championship costly, but the Tigers have done a good job of protecting the ball, turning it over once every 101.3 plays and 613.5 yards gained, ranking 6th nationally (of 125 teams) in both categories.

Below are the numbers and rankings for 30 or so teams (alpha order).

The Tigers seem to have an advantage over Louisville in this area as the Cardinals rank 100th in Turnovers/Play and 104th in Yards/Turnover.

You can find the complete list by clicking right here.
TO Analysis 2014 6

Fun with Point Spreads

Random Numbers

Below are the straight up winning % and % against the spread for favored teams from 2011-2013, FBS vs. FBS only, including Bowl Games.

The straight up column contains 2,105 games – 11 “Pick ‘em” games were excluded because there was no “favorite”.  The games that ended as a “Push” are not included in the “Cover % by Spread” table.

Going by spread, the biggest upset of the last 3 years was Lousiana-Monroe’s upset of Arkansas in week 2 of the 2012 season when the Warhawks were 30 point underdogs.

Biggest cover? Florida State beat Idaho 80-14 last season to cover a 59 point spread.

The most frequent spread over the last 3 years? No surprise – 3 points – as it showed up in 133 games (6.3%).

There are some strange anomalies in the data, such as 11 point favorites win 64.5% of the time and cover only 48.4% of the time, but 11.5 point favorites are 18-0 and cover two-thirds of the time.

There’s also a weird little thing between 31.5 and 32.5 spreads where the teams are 23-0 straight up (expected) and 18-5 ATS (not so expected) which is very dissimilar to the spreads immediately preceding (3-7 ATS) and after (0-3).

 

Spread 1thru14Spread 145thru28Spread285thru42Spread425Plus

 

Geek Speak: Kmeans Clsutering with College Football Defenses

Random Numbers

Editors Note: Paul Chimenti is a marketing analyst and provides statistical analysis for Seldom Used Reserve. The tables below originated with analysis done by Paul. For more detailed information on methodology, data, assumptions, etc., please contact chimenti80@gmail.com.

The data below includes games from 2011-2013 and includes games between “Big 5″ teams only (or those that will be this season such as Louisville) and is an attempt to “cluster” defenses together using Kmeans clustering. If you’re not familiar with Kmeans clustering you may want to read this page prior to attempting to digest the data.

First, let’s clarify what the data includes (and doesn’t include):

• Includes all games between two “Big 5″ teams (ACC, SEC, Big 10, PAC 12 and Big 12), Notre Dame and teams (like Louisville) that will be in a Big 5 conference this season.
• For teams like Louisville (and Notre Dame) only games against Big 5 teams are included.
• Does not include games against FCS teams or games with teams outside of the Big 5 – i.e. Clemson vs. Citadel and Clemson vs. Troy, for example, are not included.

The data does not purport to tell you which defensive style (“cluster”) is better than another – Cluster 1 is not necessarily better than cluster 2, just different – but rather gives you an idea of defenses with similar attributes.

I’ll have to admit that I was surprised to see Michigan in Cluster 1, but 348.1 yards per game is not a bad average these days, even if the Wolverines do play in the Big 10.

D Cluster 1 2013
Cluster 2 is where the Tigers reside and that’s probably about right over the last 3 seasons.  I would classify these as “decent”, but not top tier defenses.  An interesting side note here (at least for me) is that Arizona gave up 95.4 more yards per game than Clemson, but only 1.8 more points.  Perhaps turnovers were the key as the Sun Devils averaged a half more turnover per game than Clemson.

D Cluster 2 2013

Cluster 3 introduces us to some of the more problem defenses and includes one that many see as “good” – Ohio State.  Again, we have to remember this is a 3 season window, not a look back at 2013 and the clustering is not a referendum on “good” or “bad”, but rather grouping like defenses.

D Cluster 3 2013

Each team in Cluster 4 gave up at least 411.6 yards per game and a minimum of 29.9 points per game.  Ouch.

It’s also notable that 4 ACC teams reside in this cluster.  And remember how there were 5 Big 12 teams (of 10 conference teams) in Cluster 1 on the offensive side?  Well, there are 5 Big 12 teams in Cluster 4 on defense.

D Cluster 4 2013

 

Geek Speak: Kmeans Clustering with College Football Offenses

Random Numbers

Editors Note: Paul Chimenti is a marketing analyst and provides statistical analysis for Seldom Used Reserve. The tables below originated with analysis done by Paul. For more detailed information on methodology, data, assumptions, etc., please contact chimenti80@gmail.com.

The data below includes games from 2011-2013 and includes games between “Big 5″ teams only (or those that will be this season such as Louisville) and is an attempt to “cluster” offenses together using Kmeans clustering. If you’re not familiar with Kmeans clustering you may want to read this page prior to attempting to digest the data.

First, let’s clarify what the data includes (and doesn’t include):

• Includes all games between two “Big 5″ teams (ACC, SEC, Big 10, PAC 12 and Big 12), Notre Dame and teams (like Louisville) that will be in a Big 5 conference this season.
• For teams like Louisville (and Notre Dame) only games against Big 5 teams are included.
• Does not include games against FCS teams or games with teams outside of the Big 5 – i.e. Clemson vs. Citadel and Clemson vs. Troy, for example, are not included.

The data does not purport to tell you which offensive style (“cluster”) is better than another – Cluster 1 is not necessarily better than cluster 2, just different – but rather gives you an idea of offenses with similar attributes.

Most of these are givens – you won’t get much argument about Clemson, Oregon, Oklahoma State and Texas Tech being clustered together.

But what about Indiana? The data shows they performed worse in every category than other cluster 1 teams except for turnovers.

However, the Hoosiers played fast, averaging 2.85 plays per minute of possession, which by the way was second only to Oregon’s 2.86 in cluster 1.
O Cluster 1 2013
In other words, Indiana played fast but not efficient and in that sense in makes sense they’re included in Cluster 1.

Cluster 2 also makes sense to me as it contains some very good, but not fast paced offenses like Florida State, Alabama, Georgia and Louisville.
O Cluster 2 2013

Conversely, many would question Auburn in the 3rd cluster, as I did. But remember, this is a 3 year window so the horrid Tiger offense of 2011 most likely offset the torrid Auburn offense of 2013.
O Cluster 3 2013

Cluster 1 confirms a couple of long held assumptions of mine.

First, the Big 12 has been the “fastest” offense in college football with 5 of the 11 teams in the cluster coming from that league (there are only 10 teams in the conference).  3 more come from the PAC 12. That means 8 of the 11 teams in Cluster 1 come from conferences that are known for fast paced offenses.

Secondly, Clemson is the lone ACC team in cluster 1 and that gives the Tigers an advantage over the rest of the ACC in general, Florida State’s dominance notwithstanding.

That doesn’t mean Clemson will win every game, offense is only half the battle, but absent a stout defense (i.e. Florida State) it means the Tigers have a decided advantage in most ACC games.

Geek Speak – Yards Per Play (YPP) vs. Total Yards

ch20p8

Much as I did with the total yard metric, I plotted the yard per play (YPP) for the 2,116 FBS vs FBS games for the last three years.

Not surprisingly, the curves and results are nearly the same. In fact, the YPP metric has a slight edge – 79.0% to 77.7%.

YPP Chart and Graph 2014

However, two things stand out to me:

  1. With the exception of the last range, in which both are at 100%, the total yard metric has a higher percentage than the YPP metric in each range. How then does the YPP metric have a higher overall percentage? There are many more (328 to 169) games over the last three ranges in the YPP metric thereby weighting those ranges much heavier.
  2. While the ranges go progressively higher without exception in the total yards metric, it actually goes lower (slightly) from the 3-3.49 range to the 3.50-3.99 range. The sample size is small, only an average of 26 games per year fit in this range, so it’s likely to be an anomaly and will work itself out over time.

There’s not a lot of difference in these metrics in my mind and that was part of my point in the total yard post. YPP is a simple and easy calculation, but you could easily use a metric that doesn’t even require a calculation (total yards) and get similar results.

50,000 Foot View of College Football

Random Numbers

The charts below tell the big picture story of college football from 2011-2013 and cover 2,116 games between two FBS teams.

Some things that I found within the data:

  1. Almost all categories for winners increased (far right column) over the 3 seasons.
  2. Losing teams had reduced numbers in most categories in 2013 compared to 2012.
  3. Turnovers have remained remarkably consistent for both winners – 1.3 per game across all 3 seasons – and losers (slight variation in 2013).
  4. Winning teams average more penalty yards than losers.
  5. While the losing teams yard per pass average has remained constant, the winning teams have increased their yard per pass metric 2.5% over the 3 seasons.
  6. Both have increased their yards per rush, but winners have increased at a higher rate.
  7. Average rush yards for winners has increased by 9.2% and yards per rush by 5.1% for winners from 2011 to 2013.
  8. Scoring is up for both winners (5.7%) and losers (2.9%).
  9. Both winners and losers have increased plays and total yards, but winners have increased at  a higher rate than losers.
  10. As a whole, these numbers tend to lead credence to the theory that offenses are moving faster and have the upper hand (known as the Saban/Bielema Complex)

These numbers lay the foundation for an upcoming analysis by Paul Chimenti who holds an MS in Mathematical Sciences with Statistics Concentration. Paul is using a statatistics package that will arrange offenses and defenses in “clusters” based on metrics from the 2011-2013 seasons.

Winning Teams

2011-2013 Winners

Losing Teams

Losers 2011-2013

Geek Speak: Total Yards Matter – 2014 Version

Random Numbers

While I don’t believe total yardage is the “end-all, be-all of football” it’s pretty clear to me that total yards are an important stat in college football.

Besides the obvious – it generally takes yards to score points – I have some numbers that back up this theory.

There are many guys smarter than me that say total yards mean little, are an “overrated” or “simplistic” metric and spend many hours devising complicated formulas to prove why that is.

I’m not smart enough to understand all of the mathematics behind those theories, but my general operating theory is “the simpler the better”.

It’s difficult to find a simpler metric than total yards, and this seems to give those smarter than me fits.

Specifically, out gaining your opponent is important.  The more the better.  If you think about it, out gaining your opponent takes into account many factors that occur during the game.  If you turn the ball over consistently you are likely to gain less yards, score less points and win less often, for example and using the difference between teams total yardage also means defense is factored into the equation.

So while gaining  yards is important, this analysis looks at the difference in yardage between winners and losers.  Another way to put it is, if Team A gains 600 yards and gives up 575 yards in game 1 and gains 125 yards and gives up 100 in game 2, Team A has the same odds of winning both games.

It’s not about the number of yards you gain, it’s about the difference between the number of yards you gain and the number of yards your opponent gains.

The charts and graphs below cover 2,116 games (6 games resulted in teams having exactly the same number of yards) between Division I teams from 2011 through 2013 and tell a simple story: Outgain your opponent and you will likely win. The more you outgain your opponent the higher your odds of winning.

Winning Pct by TYA Chart

Winning Pct by TYA Graph

A little further proof that yards matter? Teams with more yards than their opponents cover 64.6% of the time. And, as with the winning %, the higher the yardage differential the more likely a team is to cover, without exception.

 

Cover Pct by TYA Chart

Cover Pct by TYA Graph

Using the Pearson Coefficient I found a solid 0.606149 correlation between total yard differential and winning.

How did Clemson fare using this metric in 2013? I’ve previously posted on why I wasn’t that worried as Clemson fell behind in the Orange Bowl vs. Ohio State and the Tigers were 9-1 (lost South Carolina) when they outgained their opponent and 1-1 when being outgained (won Georgia, lost Florida State). Against the spread the Tigers were 6-5 when outgaining an opponent and 1-1 when being outgained.

No, total yards aren’t the end-all, be-all of football. But total yards, specifically when compared to your opponents total yards, matter and this simple metric can also increase the odds of picking the team that’ll not only win, but cover the spread, too.

It’s important not to confuse correlation with causation and I’m not saying having more total yards causes teams to win by itself.  Other factors (turnovers, for example) can cause a team to have more (turnovers gained) or less (turnovers lost) total yards and win or lose the game.

I’m saying total yards is an important factor in determining winners and losers, more than many want to acknowledge.

 

Figure The Odds – Final Numbers for 2013 Season

WP 3.1.1

Below are the final numbers of the “Figure The Odds” series for the 2013 season. While the numbers may not look impressive on the surface, realize that within these numbers the model managed to go 20-14 straight up and 17-17 against the spread in one of the craziest, most upset filled bowl seasons in recent memory.

WP6

During the off season I’ll continue to add to the database (currently 1,879 games) and hit the ground running in week 1 of 2014.

(The numbers above include a record of 2-4 straight up and 4-2 ATS in the first week of December).

Figure The Odds – Version 1.4

WP5

With the completion of last night’s games the model’s record now stands at 19-9 straight up and 13-15 against the spread.  I’ve noticed a potential issue with the Cover Probability portion of the algorithm (hence the .464 record) that will probably need to be addressed.  I’m not exactly positive how to best accomplish that, but for now we’ll plow ahead.

As a reminder, these probabilities are based on the results of 1,869 college football games from 2011 to 2013.

WP5

An important distinction here – I’m not predicting what will happen in these games – I’m saying that given the data that I have teams similar to Alabama have won 87.5% of the time since 2011.

To me this is a logical way to look at things. I can’t predict the future, but I do know what’s happened in the past.

Think of these less as predictions and more as a look at history of similar games.