January 21, 2019

Geek Speak: Why Every Yard Matters, The Relationship Between Yards & Points

One of the most important statistics in football is also one of the most basic – total yards. If you knew nothing else about two teams,  be it records, point spread, who was favored and who was the underdog or any other in game stat, but you knew who had the most total yards you would have a 75% chance of picking the winner.

Closer to home Clemson went 9-1 in games where they out gained opponents and 1-2 when being out gained.

It’s usually at this point that football savants remind me of the teams who gained more yards than their opponents and lost as proof that I’m “wrong”.  Maybe they feel like I’m taking the physicality and strategy out of football by assigning a value to each yard gained (more on this below), but really the point of this is to reinforce the importance of each and every yard gained or lost, emphasizing the importance of the physical nature and accompanying strategic moves that are part of the game.

Many prefer black and white, yes or no and disdain odds and/or probabilities.  The only metric they are interested in is points.  Getting more than the other team guarantees a win 100% of the time. Nothing else matters.

Yet if a team starts at its own 10, drives 40 yards, punts and their opponent is backed up inside the 10 the team gets 0 points for that drive – but those 40 yards have a value.  Field position has been changed, and so have the odds of winning because of those 40 yards that yielded 0 points.

Every yard is important, at least while the game is in doubt.

Clemson scored on offense, defense, and special teams in 2014.  However, the vast majority of touchdowns (and therefore points) came on offense (89.6% of Clemson touchdowns came on offense) and involved gaining some amount of yards.  Sometimes it takes a lot of yards, sometimes just a few, but by and large you score by gaining yards.

The graph below plots points and yards of every game over the last four years (between 2 FBS teams). The slope should tell you all you need to know.

Yards & PointsYes, there are outliers, but the picture tells a story in 3 words – yards equal points.

Better than that generic “yards equal points” phrase, we are able to determine exactly how many points a team can expect to score based on yards gained.  Even more intriguing than that is the close to perfect symmetry of the numbers below.  For almost every 11 yard increment one additional point can be expected.

Expected Points Per Total Offense

There are no exceptions, meaning there is no instance where gaining more yards means you should expect less points.  It sounds amazingly obvious, but you’d be surprised at how many football fans believe total yards is an irrelevant metric.

Of course, the yards you gain are only part of the game and are therefore relative.  You can expect to score 36 points if you gain 490 yards, but if your defense gives up 510 yards you will most likely lose.

That doesn’t mitigate the overall point, which remains valid – yards are important because the more yards you gain the more points you are likely score – without exception, statistically speaking – and that means the more likely you are to win.

Total yards is certainly not the only metric I use when determining win probabilities, but its an important one that I give significant weight in my calculation.


Projected Stats, Third Downs and Expected Points

The projected stats for the Wofford game are listed below:


Name Complete Attempts Yards   % TD Int Y/A
Boyd       19       33   214  57.6  2   1  6.5



Name Attempts Yards Average TD 10+ 20+
Ellington      14    78     5.6  1   2   1
Bellamy       3    17     5.7  1   1   1
Boyd       4     4     1.0  0   0   0
Watkins       2     5     2.5  0   1   0


Name Catches Yards Average TD 20+
Watkins      6    72 12.0  1   1
Hopkins      5    55 11.0  0   1
J. Brown      3    35 11.7  0   0
Allen      3    34 11.3  0   0
Ellington      1    11 11.0  0   0
Ford      1     6   6.0  0   0

We’ve harped endlessly here about third down conversions and while it was only one game, last week was another example. 0 for 8 in the first half led to a 16-13 Troy lead at intermission. 5-8 in the second half led to a 30 point outburst and 43-19 victory. Too many times (11 of 16) the Tigers found themselves in third and long, a situation with a low conversion rate.

The graphs below are a visual representation of the information above, comparing last years third down conversion rate with 2011 and showing the high number of third and longs. 

Expected points is a stat that measures how many points a team scores on a drive on average when they have a first down at any given point on the field. Professional statisticians, which I am not, will tell you that a large number of data points are needed in order for the averages to be statistically valid. Often times they consolidate data from entire leagues in order to have statistically viable results.

There are several reasons why I have only included Clemson data and only data from this year (though it would be interesting to compare the Morris offense to its predecessor or Clemson to other ACC teams). At the top of that list is time. Mine is limited. Secondly, other than the passing interest expressed above in comparing Clemson to other teams and to previous Clemson teams, the ultimate goal is to determine the average for this team and perhaps future Clemson teams.

The data below only includes drives from the Troy game (we’ll add as the season moves forward). Theoretically, as more data is added you should see a line from the bottom left to the top right of the graph – a team typically scores more often as they move closer to their opponent’s goal. With such a small amount of data available we see huge peaks and valleys in the averages which will tend to smooth out over the season. But even with this small sample size you can see that there are many more 6 and 7s on the right side of the graph (first downs in opponent’s territory) than on the left side (first downs in Clemson territory).

What can this data tell you? Let’s look at the drives starting at the Clemson 20 as an example. Three times the Tigers had first downs at their 20. They scored touchdowns on two of those drives (with one missed extra point) for a total of 13 points on 3 drives, a 4.33 point per drive average. For illustration purposes lets say that this average holds for the entire year. Fast forward to the bowl game. Clemson is trailing by 5 with a first and 10 on their 20. What are the odds of Clemson scoring 6 to win the game from this point in the field? Is it 2 in 3? Or is it lower because the Tigers averaged 4.33 points per drive when they have a first down at this spot on the field?

The great thing about expected points is that the odds change as the team moves down the field on the same drive. Again, assuming the averages on the chart above stay the same, let’s say Tigers move down the field and now have a first down at the opponents 48. All of a sudden the Tigers average 6 points per drive (unlikely over the course of a season) with a first down with this field position and things are looking up.