Spotlight Stat: First Down Efficiency

[bostonsoxandy]

I know 3rd down efficiency exists as a statistic (generally used for teams), but I wanted to see if I could apply a version of it to individuals. I came up with the idea of first down efficiency which basically answers this question:

 

If I give this guy a carry, what is the chance he turns that carry into a new set of downs and moves the sticks?

 

 

My hope is that as I go crunch the numbers, we can see some guys who had high efficiency ratings last year that are lesser known names or not as highly valued players. I'll reflect back after I do a hundred or so running backs.

 

Ranked from most efficient to least:

Mike Tolbert - 57.4% (essentially a FB though)

Andre Brown - 45.2% (small sample size)

Brandon Bolden - 35.7% (small sample size)

Dion Lewis - 33.3% (very small sample size)

Stevan Ridley - 32.4%

Bryce Brown - 32.3% (small sample size)

Ben Tate - 31.3% (2 years)

Chris Ivory - 31.3% (2 years stats, small sample size)

Shaun Draughn - 30.5%

Evan Royster - 30.4% (2 years, small sample size)

Danny Woodhead - 30.3% (small sample size)

David Wilson - 29.6% (small sample size)

LaMichael James - 29.6% (small sample size)

Frank Gore - 29.1%

Shane Vereen - 29.0% (small sample size)

Alfred Morris - 28.7%

Ahmad Bradshaw - 28.1%

Willis McGahee - 28.1%

CJ Spiller - 28.0%

Adrian Peterson - 27.9%

Kendall Hunter - 27.8%

DeMarco Murray - 26.7%

Arian Foster - 26.5%

Isaac Redman - 26.4%

Delone Carter - 26.3% (small sample size)

Knowshon Moreno - 26.1%

Mikel Leshoure - 25.6%

Lamar Miller - 25.5% (small sample size)

Marshawn Lynch - 25.1%

Michael Bush - 25.1% (2 years)

Ronnie Brown - 25.0% (2 years, small sample size)

Vick Ballard - 24.6%

Bilal Powell - 24.5% (small sample size)

Fred Jackson - 24.3% (small sample size)

Maurice Jones-Drew - 24.2% (went back an extra year to make up for lack of stats last year)

Justin Forsett - 23.8% (small sample size)

Lance Dunbar - 23.8% (small sample size)

Armando Allen - 23.8% (small sample size, 2 years)

Rashard Mendenhall - 23.7% (year old stats)

LeSean McCoy - 23.5%

Jamaal Charles - 23.5%

Ronnie Hillman - 23.5% (small sample size)

Peyton Hillis - 23.5% (small sample size)

Felix Jones - 23.4%

Ray Rice - 23.3%

Joique Bell - 23.2% (small sample size)

Michael Turner - 22.5%

Toby Gerhart - 22.0% (small sample size)

Roy Helu - 21.9% (2 years back)

Shonn Greene - 21.7%

Doug Martin - 21.6%

BenJarvis Green-Ellis - 21.6%

Reggie Bush - 21.6%

Deangelo Williams - 21.4%

Mark Ingram - 21.2%

Daniel Thomas - 20.9%

Darren Sproles - 20.8%

Bernard Pierce - 20.4%

Matt Forte - 20.2%

Montario Hardesty - 20.0%

Steven Jackson - 19.8%

Daryl Richardson - 19.4% (small sample size)

Donald Brown - 19.4%

Ryan Mathews - 19.0%

Robert Turbin - 18.8% (small sample size)

Mike Goodson - 18.8% (2 years)

Jonathan Dwyer - 18.6%

Chris Johnson - 18.5%

Beanie Wells - 18.2%

Jacquizz Rodgers - 18.1%

Rashad Jennings - 17.8%

Jason Snelling - 17.7% (2 years)

Trent Richardson - 17.6%

LaRod Stephens-Howling - 17.2%

Darren McFadden - 16.2%

Pierre Thomas - 16.2%

Cedric Benson - 15.5%

Alex Green - 12.7%

Ryan Williams - 12.1% (small sample size)

 

 

 

So there is the first 50 or so. Although I don't have any proof (yet!) behind this hypothesis, I hypothesize that there is a direct relationship between first down efficiency and overall fantasy production. Not really that ground breaking. But considering I think this stat shows how effective a player is for the actual football team, it is reasonable to suggest they will get more or less work the subsequent year based on their effectiveness. And this measure goes behind glamour stats such as yards or TDs, as TDs can be erratic and yards can be heavily inflated due to a few long runs (see: Murray, Demarco vs. STL rookie year type games).

 

This really should be a way to get a feel for potential consistency of players and how they would grind through big carry numbers. Obviously the fewer carries the player got last season they would have slightly higher #'s due to being fresher, but I think it still is a nice little stat to atleast look at.

 

A few pragmatic take aways for this upcoming year:

• ​Andre Brown, Ben Tate, and Bryce Brown are just begging for more touches. I'm not saying they will get them automatically but I think this shows they have much higher upside and also a much higher appeal for their coaches to go to them.
  
• Chris Johnson, Run DMC, Turner, Forte, and even SJax were all more ineffective than average last year. I think SJax's change in situation will shake things up for him, but the other three -- I'd stay away.
  
• There are a ton of other conclusions you can make. Feel free to reply with your own
  

 

What does everyone think?

Edited July 10, 2013 by bostonsoxandy

[keggerz]

I think, job well done!

 

I know it isn't easy to mine data like that. As for what the metric means, that will take a bit more time to digest.

[bostonsoxandy]

Thank you. I always like to figure out stuff like this in my free time, it's some good fun going through data. I'm looking at it through a historical lens and it seems that the basic statement that this stat is related to overall running effectiveness rings pretty true.

 

Barry Sanders had a career 22.7%, AP has a career 24.0%, Emmitt Smith had a career 23.1%, Faulk had a career 23.5%, and LT had a career 21.5% So that is clearly an elite level of players.

 

Willis McGahee had a 21.2%, James Stewart had a 20.4%, Rudi Johnson had a 20.4%, Travis Henry had a 19.8%, Antowain Smith 19.2%, Duce Staley 19.1%, Michael Turner 21.2%, Deuce Mcalister 20.4% I pulled those players indiscriminately from the low-end of a top-100 list.

 

Any names of just truly mediocre backs that you can think of? Guys that just had a serviceable, non-great but non-bad career?

[keggerz]

I'd take a look at a guy like Lamar Smith to see how he grades out from 2000-2001 or 2000-2002

 

Ernest Byner too

 

Another thing would be to take a look at a guy the year prior to a breakout and compare it to the year he did

Edited July 9, 2013 by keggerz

[WashingtonD]

Wow,

What a great read and contribution! Thanks for this.

 

Like Keg, I need to puzzle over this a bit more and think it through as it's something I haven't considered or seen before, but will definitely post my thoughts soon.

 

I'd like to make sure I understand what the percentages represent.

 

Let's take CJ Spiller for example - does the 25.1% essentially mean that in 2012, 1 out of 4 times that he was given the ball on first down he moved the sticks - 10+ yards or 5 yards had their been an offsides or other penalty on play prior resulting in a first down?

 

 

EDIT - now re-reading with fresh eyes this morning and seeing Second String's helpful post I see I was mistaken. Seems like the rationale explanation is that the percentage represents when they got a 1st down after a carry on 3rd down?

Edited July 9, 2013 by WashingtonD

[SecondString]

Does this take into account all of their touches, or only for their 3rd down touches?

 

If it is the former, then theoretically shouldn't it be expected that the specialty 3rd down backs would score higher as a rule? Since a higher percentage of their touches would require shorter gains in order to achieve a first down?

 

That's just an initial thought, although looking at the list it doesn't look like that necessarily bears out.

[bostonsoxandy]

To answer Second String and WashingtonD's questions, the stat is derived simply by # of first downs gained by their rushing attempts divided by total carries. I made it total carries because 3rd down touches are something not every player gets and are more erratic than all the carries and all the first downs.

 

WashigtonD -- your Spiller example is explained almost perfectly. The only correction I would make is any time he was given the ball in any situation on the field he moved the sticks for a first down 25.1% of the time.

 

SecondString -- while it is not just 3rd down carries it is important to note that goal-line guys/fullbacks (short yardage only) should and do have a much higher % for this statistic on average. For example, FB Leach has a career 1st down eff of 58.3% and Tom Brady has a 1st down eff of 35.3%. That obviously doesn't mean they are better rushers than all RBs, but the stat has less meaning for them because of when they rush.

 

However when you compare it to just one group of players (e.g. just RBs or just QBs) you can make comparisons on running ability. Peyton Manning has a career 1st down eff of 20.3% while Rodgers has a career 36.6%. Based on ONLY this statistic you can tell Rodgers is a more effective running QB than Peyton. Culpepper had a career 39.3% and Vick has a career 39.3% as well. While a guy like Marc Bulger has a career 22.9%.

 

Keggerz - I'm going to delve into the idea of predicting breakout stars with this (if there is any actual predictive power to it) in much more detail later when I have more time. I'll make sure to post here. But just a taste for now of some breakout guys 1st down eff the year before they broke out:

 

Peyton Hillis - 34.6% career before break out season

Arian Foster - 27.8% before break out season

 

 

I'll hit a lot more later.

[flemingd]

Ok so it's not first down efficiency, it's first down conversion rate. Gotcha. I figured it had to be that, no way a guy gets a carry on first down and gains 10+ yards 25% of the time. Good info, great research, thanks!

[darin3]

That's some solid data crunching. Never really thought of that.... and it certainly bears consideration in drumming up your sleepers list, for example.

[WashingtonD]

I'd like to see if the data predicts the opposite of breakout, if it can predict fall off - would the data give a warning for the cliff a player like rudi Johnson hit a long time ago ?

[darin3]

I'd like to see if the data predicts the opposite of breakout, if it can predict fall off - would the data give a warning for the cliff a player like rudi Johnson hit a long time ago ?

 

Exactly... the counter-effect. Older players losing effectiveness - especially on 3rd downs - would give way to the younger "change of pace" kind of guys that make up the RBBC in many instances. Good call.

 

Thankfully, something to get us off the Ass Elf train.

[bostonsoxandy]

darin3 and WashingtonD -- I've been considering that just as much. I really think this stat can have legitimate value in fantasy football because of its overreaching nature. It can predict players prime to breakout and players who are ready to decline in production theoretically. You brought up Rudi Johnson, so let's see what we got for good ol Rudi:

 

 

Note: I'm also going to start counting each rushing TD the same as a FD. The initial list does not account for that which penalizes players that score a lot of touchdowns, which is quite silly. However I do not want to overvalue TDs, so they are essentially FD's for this instance.

 

2002 -- rookie year and only got 17 carries. Made them count though with a 35.3% first down conversation rate (that is a better name, thanks flemingd)

2003 - 215 carries @ 26.5% rate (could argue he got significantly more carries due to very high rate last year)

2004 - 361 carries @ 23.5% rate (again, due to fantastic rate last year he got even MORE carries)

2005 - 337 carries @ 27.0% rate (could argue that b/c his rate diped slightly he received slightly less carries but still a lot as 23.5 was a great rate)

2006 - 341 carries @ 23.2% rate (great rate last year = carries go up next year)

2007 - 170 carries @ 17.1%(injury caused him to miss five games)

2008 - 76 carries @ 15.8% (this may have caused no team to want him. career low rate with career low carries? yikes)

 

You could certainly make the argument that you could predict Rudi's rise due to the exorbitantly high rates the first two years. But to answer your question I think it would be impossible to conclude he would fall off after 2006 solely from this data. However common sense says when you ride a player for 330+ carries three straight years he is due to break at some point.

[bostonsoxandy]

I will have to recalculate the #'s for the top-50 in the initial post by accounting for rushing TDs as well when I have a moment. Should have some solid time to crunch stuff later tonight.

 

Feel free to crunch yourself if you like to do this type of stuff (I know I like it personally). All the formula is is (first downs gained+rushing TDs) / (total carries)

[WashingtonD]

So based on what you are seeing if you had to draw 4 lines in your percentages for Great, Good, Average, Bad - at approximately what breaking point would you place those four designations?

[flemingd]

I'd speculate that this is more valuable as a team performance metric, as the O line, the overall quality of the offense (how often they are in 3rd and 2 instead of 3rd and 9), and the scheme (how often they pass on 3rd and 2, do they have a wildcat/running QB?) will have significant impacts. AP might be the next best RB after God, but he's still not going to convert on 3rd and 19 very often.

 

I like this as a predictor of handcuffs and/or breakout players - rookies, getting new FA's in, cutting a vet and handing the reins to a new guy (NYG), etc. For example, Moreno and McGahee both converted almost exactly 25% last year - good news for Ball. Vick Ballard converted 23.3%, and Bradshaw showed he could get it done with a 25.3% at NYG - he'll serve well in the 4 games he's healthy in.

 

Oh, and stat of the day:

 

Cam Newton - 38.5%

[bostonsoxandy]

WashingtonD that is a great question and I need to crunch a good amount of data to answer that with some confidence. And obviously adding in TDs will slightly inflate those #'s.

 

But with very little confidence and just gut feeling at this point I would slap these numbers on as general tiers (subject to revision later):

Elite = 30%+

based on Shaun Alexander's 2008 campaign (36.2%), LT in 2006 (31.3%), CJ2K in 2009 (26.0%), and Arian Foster in '10 (32.1%)

Excellent = 25-30%

based on Arian Foster in 2012 (26.5%), Doug Martin last year ( ),

Good = 22.5-25%

Average = 20-22.5%

based on Shonn Greene last year (21.7%)

Poor = 15-20%

Awful = < 15%

 

Went be feel as I started estimating them, but I'll come back and make some more firm benchmarks,.

[bostonsoxandy]

I'd speculate that this is more valuable as a team performance metric, as the O line, the overall quality of the offense (how often they are in 3rd and 2 instead of 3rd and 9), and the scheme (how often they pass on 3rd and 2, do they have a wildcat/running QB?) will have significant impacts. AP might be the next best RB after God, but he's still not going to convert on 3rd and 19 very often.

 

I like this as a predictor of handcuffs and/or breakout players - rookies, getting new FA's in, cutting a vet and handing the reins to a new guy (NYG), etc. For example, Moreno and McGahee both converted almost exactly 25% last year - good news for Ball. Vick Ballard converted 23.3%, and Bradshaw showed he could get it done with a 25.3% at NYG - he'll serve well in the 4 games he's healthy in.

 

Oh, and stat of the day:

 

Cam Newton - 38.5%

 

 

I'm glad you brought that up because I agree. When a player goes to a new team I think it best to ignore this stat basically completely. New situation makes this useless. Same as MAJOR changes in the team's offense.

 

And as for Newton - QB's tend to have much higher #'s (not to burst your bubble - that still is an elite running QB #) so they would have a different scale probably. Vick has a career 39.9% so the #'s are inflated and Brady has a 38.8% - more of a testament to how the usual time Brady runs is when he needs a first down or TD a yard away.

[bostonsoxandy]

Rankings updated to add-in TDs. Holy cow, Andre Brown is up there. His % is quite inflated due to the fact that he got a touchdown five straight games while getting virtually no carries (single digits four of those fives games). He is begging for more opportunity.

[Axe Elf]

Seems like this is basically a measure of a player's ability to imitate Leroy Hoard, of whom it was once said, "If you need three yards, he'll get you four. If you need five yards, he'll get you four."

 

Maybe you could call it the Hoard Ratio, as in, "Trent Richardson's HR of 17.6 is the lowest among all first round RBs."

[Axe Elf]

Rankings updated to add-in TDs. Holy cow, Andre Brown is up there. His % is quite inflated due to the fact that he got a touchdown five straight games while getting virtually no carries (single digits four of those fives games). He is begging for more opportunity.

 

 

Yeah, I've tried to pass along the word that David Wilson is a scrub. Looks like it just takes a healthy Hoard Ratio for Andre Brown, and people start paying attention.

[bostonsoxandy]

Yeah, I've tried to pass along the word that David Wilson is a scrub. Looks like it just takes a healthy Hoard Ratio for Andre Brown, and people start paying attention.

 

 

Well, Wilson is well up there too with a 29.6%. Of course both Brown and Wilson have small sample sizes to go off of, so have to put a dash of caution before a glowing endorsement to take either at whatever the cost. But I think it does show the Giants have two effective runners in their current situation.

[Axe Elf]

Throw in Wilson's fumble percentage of what, 73.9, and Brown is the one to own.

[bostonsoxandy]

Throw in Wilson's fumble percentage of what, 73.9, and Brown is the one to own.

 

 

He had one fumble last year FWIW

[bostonsoxandy]

The top-100 RBs are now on the list. Will add darkhorses as they come along. Obviously rookies are excluded as they have no previous stats to use. I think it is very useful to view by team - that way the situation is more constant and the % has more value to determining how carries are going to be split potentially.

 

For instance the Patriots are clearly one of the most efficient teams in the league, so running backs in NE need high %'s to keep being used. Take the guys from last year:

Ridley (32.4%), Vereen (29.0%), Bolden (35.7%), and Woodhead (30.3%) are all very efficient runners by this metric leaguewide. But now that Woodhead is going to San Diego, it is very reasonable to expect that number to be near meaningless and for it to drop off. This also tells you that Bolden can be expected to have more carries this year than last year (which I expect, as a Pats fan, will be the case)

 

I think it can also help predict how carries might progress throughout the year. In Detroit, for instance, Bush is the new FA so he will probably get more carries than he deserves early to justify his purchase by the team. But as the season progresses, I think we may see a shift to Leshoure (25.6%) over Bush (21.6%) and Bell (23.2%). Of course Bush's % may be meaningless as he changed teams, but after a few weeks a new number can be generated - and it could be possible that Leshoure is more efficient and can be had at a great value price.

 

Stuff like that is why I think it is a valuable metric.

[Swiss Cheezhead]

First of all, props for doing this data research. Second, I was also confused by the metric name and description in the first post, but I understand it now. I'd suggest you come up with a new name for it, at least.

 

I think the easy way to determine the value of this metric is to make a list of all the RBs from 2011 and 2012 whose end-of-year performance was drastically different from their preseason ADP (both breakouts and busts). Then, find their FD/TD efficiency metric from the year before their "surprise" season. That will either define this metric as a solid predictor or a meaningless indicator of O-line/scheme/etc.