Math Wednesday: Measuring Efficient Offenses

By Chad Peltier on July 17, 2013 at 7:00p

The search for the One, True, Most Perfect Single Statistic to analyze the strength and efficiency of offenses and defenses has recently come down to a debate between two YPPs: yards per point and yards per play. 

Even I could keep the offense running efficiently through holes like thatHyde was a big reason for the offense's efficiency 

Ed Feng describes the latest volley in the fight by Tim Chou at the Sloan Sports Analytics Conference:

football coach Tim Chou proposed Yards Per Point as this essential metric for football. For offense, take yards gained and divide by points scored. On offense, lower values for Yards Per Point mean a more efficient offense. This unit tends scores touchdowns instead of field goals after long drives.

...In his talk, Chou showed that the difference in Yards Per Point (defense minus offense) correlated strongly with winning percentage in college football. Hence, Yards Per Point is a good efficiency metric to judge a football team.

Yards per point and yards per play aren't the only two big measures of efficiency. Points per play is also a widely-used metric by both Football Outsiders' S&P+ and Mr. SEC, who calls the statistics "quick strike" and "slow grind" for offense and defense. 

These hopefully sound familiar to you by now, as I've tracked the Buckeyes' quick strike and slow grind numbers for the past few seasons.

Feng argues that the correlation between early and late season yards per point values is weak, random, and doesn't isolate a single unit, indicating that random events like turnovers play affect yards per point throughout the season. Instead, Feng argues that yards per play is a better efficiency metric for both offense and defense.

Each statistic has its strengths and weaknesses, so it's worth taking a look at which best described the Buckeyes' offense in 2012. 

If you take these three variables – yards, points, and plays – and combine them, you are left with the metrics described above. The table below uses these three metrics to describe the Buckeyes' 2012 offense. The number in parentheses is the Buckeye offense's rank.

  Yards Per Point yards per play Points per play
2012 Buckeyes 11.4 (4th) 6.08 (33rd) .516 (9th) 

There’s obviously a huge discrepancy in the rankings between the yards per point and points per play on the one hand and the yards per play on the other. While the Buckeye offense was near the top of the pile in the former two metrics, Braxton and company lagged behind on yards per play.

While initially confusing, this actually tells a fairly coherent story about the Buckeye offense.

Yards per point is closely related to red zone efficiency, meaning that teams with low yards per point scores generally score touchdowns more often than field goals. These teams minimize the number of yards it takes to get each point.

Herman can probably tell you these statistics in his sleepHerman is probably calculating these stats in his head

Similarly, a team with a high points per play has few negative plays, generally gaining yards with each down and getting touchdowns from drives instead of turnovers, field goals, or punts.

On the other hand, yards per play measures efficiency, but does not take turnovers into account. A team might rank near the top of the country in yards per play but still not be an effective offense due to turnovers, field goals, or poor red zone percentage. After all, points win games – not yards.

However, which metric (or two metrics) best describe the Buckeyes last season?

The rankings above seem to indicate that the Buckeyes were very efficient with their drives without necessarily racking up yards.

The Buckeye offense was fourth and ninth in the country in the two points related metrics, but 33rd in yards per play. This view is corroborated by the Buckeyes’ stellar red zone percentage – 88% – which was eighteenth in the country.

In my opinion, it is necessary to see all three of these metrics side by side to get the full picture of the Buckeye offense from last season. The offense was generally efficient in translating long drives into touchdowns, but it also had a fair number of three and outs, which drove down its yards per play values. 

What do you think? Does one measure – YPPoint, YPPlay, or PPP – make more sense to you? Next week I'll take a look at the defense numbers for each metric to try and better understand the defense's performance. 


Comments Show All Comments

buckguyfan1's picture

I think the yards per play stat gets a whole lot better this year.

EvanstonBuckeye's picture

Hyde is the key. When he was running like a horse last year, it just looked so easy for us. I really think the Brax for Heisman thing and our success run counter; the more effective our running game is, the less Brax will have to carry the load. His stats might go down slightly (though maybe not), but our efficiency goes way up. Let him have his big games in big places and spread the wealth otherwise.

QBYBuckeye's picture

I prefer points per possession as a metric.
If you score a TD on every possession you can't be beat, no matter how many plays or how much time it takes, as long as your Defense can make just one stop.

New York Buckeye

Jabba the Hoke's picture

This is the true measure of efficiency for me. The other stats put a lot of importance on big plays, but a team that can grind out 10 to 15 play drives and score touchdowns that way is equally as dangerous, plus it takes a lot of pressure off of the defense. In the end, it doesn't matter how you score, just that you do it consistently; points per possession is a great way of measuring that.

Chad Peltier's picture

Great suggestion - I think this is something I'll have to track next season 

Hovenaut's picture

This is the stuff.

Since we're not watching fuhbah yet, and I'm probably not breaking down film anytime soon, these are the details tiding me over.

We saw Brax get electric at times, El Guapo move piles at times and in between we saw some contributions from the receiving corps.

I think things crank up a notch, not only in effectiveness and efficiency, but also with explosiveness with some of the young jets ready for lift-off.

I can hardly remember what exactly constituted Tresselball....pick your poison defenses.

Toilrt Paper's picture

All I need to know is PPG. Ohio State averaged 37.2 last year. With almost everyone back, the 2nd year in the system, with several speed guys coming in. How many PPG over 40 will this team average?

MN Buckeye's picture

Yes.  PPG over 40? I am guessing 9 games over 40, overall 44ppg.

AndyVance's picture

What this article illustrates really well is that striving for a single metric to measure anything is generally a bad idea. I'm currently reading Nate Silver's treatise on forecasting, The Signal and the Noise, and he discusses the various economic indicators Wall Street wizards use in attempting to forecast the economy - and why most of their elaborate systems fail because they've tried to focus on too many, too few, or simply the wrong indicators.
Likewise, without looking at all three metrics discussed, one misses the bigger picture of the 2012 season for the Good Guys. Great stuff, Chad, keep it up.

ellspar's picture

I highly enjoyed that book. A very good read, especially about the dangers of prognosticating.

Chad Peltier's picture

Nate Silver is like a rockstar. 

pjtobin's picture

I'm down with opp. Ya. You know me. 

Bury me in my away jersey, with my buckeye blanket. A diehard who died young. Rip dad. 

MassiveAttack's picture

Ha-ha, that was awesome, PJTOBIN!

The Ohio State University - "Haters love us!"

gravey's picture

It would be interesting to see these stats for various great Buckeye squads over the past 15-20 years.   Where was Bama last year on these?
It would seem that some of Tressel's and Coop's teams, both had great records, but very-very-very different offensive styles.

Catch 5's picture

Bama's Offensive stats were as follows:
Yards per point:  11.5
Yards per play:  6.95
Points per Play:  0.60
Defensive Stats were:
Yds/pt:  22.9
Yds/play:  4.18
Pts/play:  0.18
Roll Bama Roll did a breakdown of their stats (through Saban's years) here if you are interested:

Make their asses quit! - Nick Saban

Chad Peltier's picture

Thanks for linking these articles/stats, Catch 5. 

darbnurb's picture

Thanks for putting statistics to what I was just thinking about yesterday (as I was re-watching the OSU - Cal game).  Buckeyes seemed to score a lot of points, but not necessarily rack up a lot of yards and had many stalled drives (often times a whole quarter of 3 & outs).  

CentralFloridaBuckeye's picture

I think our points per game is going to be crazy high this year.  I look for it to be in the 40s and possibly higher in many games.  going to be a fun year!

southernstatesbuckeye's picture

I wouldn't think our ppg would be any less than 50 once the year is said and done.  Scoring 37 per game last year with the flaws this team had was incredible considering that 2011 (minus the bowl game which would've brought it even lower) only produced 25 points per game with the same players, basically.
That's a difference of 13 points for a squad with 1) No incentive for a bowl game, 2) New coaches, and 3) A whole new system/scheme.  That is significant. 
With the addition of an eye-popping slate of new arrivals, and proven performers at key positions already, how could we NOT score an average of two more touchdowns per game?  It seems to me to be a no-brainer!

I like cookies.

Catch 5's picture

Because your defense is going to be better.  The question is not how many points COULD they score but how many WILL they score.  When your defense is better, your opponents fail to score as many points - and as such you don't have to keep scoring into the 4th quarter at as high a rate.  You can slow down and eat the clock and run more predictable plays in attempts to not run up the score.  If/when OSU fails to make huge increases in points per game, don't lose heart - look instead at the difference in ppg between OSU and their opponents.  That will be a better measure that should increase this year.

Make their asses quit! - Nick Saban

Toilrt Paper's picture

I would think Urban understands that to make the BCS Title game, it could take scoring as many points as possible. If there are there 3 undefeated teams it could take scoring 45+ PPG to beat out a school with a more difficult schedule.

NEWBrutus's picture

The issue I have with Yards per play or points per play is it somewhat misses the point of the overall objective of the contest.  As on offense your goal is to score points.  As a defense, your job is to prevent your opponent from scoring.  The number of plays is not entirely relevant to the outcome. 
Isn't the better measure (and perhaps more complicated to track due to the way stats are provided) are the points per possession or drive as opposed to points per play? 
A one play drive resulting in a TD has the same value as a 14 play drive which results in a TD.  The outcome is 7 points either way.  Yes the 14 play drive has some difficult, if not impossible, to measure cumulative effects over the course of a game, but in the end they still have accomplished the primary goal of an offense.  The number of plays is hardly relevant to the primary goal, ie scoring points. 
For example, Team A is known for their "explosiveness" and Team B is known for it's grind it out.  Both teams have 10 posesssions or drives in the game, and the game ends in a 42-all tie before overtime. 
Team A runs a total of 5 plays per drive (For whatever reason) or 50 plays.  Team B averages 8 plays per drive or 80 plays.  Team A's points per play is .84, team B's is .525.  Is one really better than the other?  Both offenses accomplished the primary objective, finish your drives with points, preferably touchdowns.  When it is 42-42 does it really matter how many plays were executed for each team?  Aren't both teams being equally efficient by converting 60% of their opportunities into touchdowns?  (Likewise each team's defenses are equally inefficient by ending only four possessions each without scoring!)