≡ Menu

Previous post:

Next post:

Some Initial Thoughts Running Backs

Peterson and Foster each finished in the top four in rushing first downs

Peterson and Foster each finished in the top four in rushing first downs.

I’ve spent a lot of time this offseason looking at how to grade wide receivers; today I wanted to get some initial thoughts down on paper on running backs.

I’m short on time today, which means a lot of data and not so much theory. One of the more underrated statistics to measure running backs is the number of first downs they produce. I don’t like using Yards per Carry for running backs because that metric is pretty sensitive to outliers. But by using rushing first downs, perhaps we can smooth things out.

We know that the value of a touchdown is about 20 yards, but what is the value of a rushing first down? Being short on time, I took the easy way out. Pro-Football-Reference has produced Expected Points Added for each team’s running game going back to 2000. I decided to run a regression on the team level to best predict rushing EPA based on four rushing statistics. The R^2 was 0.77, but more importantly, here was the best fit formula:

EPA = -16.6 -0.58*Rush + 0.067*Rush_Yd + 1.43*Rush_TD + 1.08*Rush_FD

What interests me is the relationships between the variables. Rushing touchdowns are considered 21.3 times as valuable as rushing yards, which happens to fit in well with our previous assumptions. But more importantly, this tells us that a rushing first down is worth 16.1 rushing yards. That seems pretty high to me, and I reserve the right to adjust this later, but for now, let’s adjust down and say the value of a rushing first down is 15 yards. Now what?

There were about 12,000 rushes by running backs last season. If we give each running back 20 yards for each rushing touchdown and 15 yards for each rushing first down, then the average running back gained 8.05 Adjusted Yards per Carry last year. From there, we can — by making some broad and some inappropriate assumptions — come up with a value over average statistic.

Let’s start with Adrian Peterson. He rushed 348 times for 2,097 yards last year, and also picked up 88 rushing first downs and 12 rushing touchdowns. That gives him an average of 10.5 AY/C last year, nearly 2.5 AY/C above league average. That means over the course of his 348 carries, he is credited with 856 Adjusted Yards1 above average. If we do that for every running back, we get the following table:

1Adrian PetersonMIN34820976.03128810.51856
2C.J. SpillerBUF20712446.0165310.43493
3Alfred MorrisWAS33516134.8113869.44467
4Stevan RidleyNWE29012634.3612839.48414
5Marshawn LynchSEA31515905.0511729.17355
6Andre BrownNYG733855.2782612.81348
7Frank GoreSFO25812144.718689.28318
8Jamaal CharlesKAN28515095.295638.96260
9Mike TolbertCAR541833.3972412.65248
10Ahmad BradshawNYG22110154.596589.07226
11Bryce BrownPHI1155644.94339.9214
12Cedric PeermanCIN362587.1711011.89138
13Brandon BoldenNWE562744.8921810.43133
14Willis McGaheeDEN1677314.384448.81127
15David WilsonNYG713585.044179.76122
16Michael RobinsonSEA12494.0801016.58102
17Kendall HunterSFO723715.152189.46102
18Justin ForsettHOU633745.941149.5997
19Delone CarterIND321223.8131110.8489
20Dion LewisPHI13695.311613.7774
21Danny WoodheadNWE763013.964208.9669
22Arian FosterHOU35114244.0615788.2569
23Doug MartinTAM31914544.5611648.2667
24Ben TateHOU652794.292189.0666
25Evan RoysterWAS23883.832810.7863
26Jacob HesterDEN17814.762511.5359
27LaMichael JamesSFO271254.6301010.1958
28Maurice Jones-DrewJAX864144.811218.7157
29Joique BellDET824145.053168.7154
30Darrel YoungWAS14604.290711.7952
31DuJuan HarrisGNB341574.62289.3243
32Bernard PierceBAL1085324.931248.4443
33Joe McKnightNYJ301795.97079.4743
34Montell OwensJAX422094.981109.0241
35Chris IvoryNOR402175.43279.0540
36Lamar MillerMIA512504.91128.8240
37Shaun DraughnKAN592333.952168.6938
38DJ WareTAM11514.640511.4537
39Shane VereenNWE622514.053158.6537
40Marcel ReeceOAK592714.590168.6636
41Lex Hilliard2TM9333.67051236
42Ray RiceBAL25711434.459528.1835
43Ryan Grant2TM321324.13289.1334
44Cyrus GrayKAN7446.290312.7133
45DeMarco MurrayDAL1616634.124398.2532
46Ronnie BrownSDG462204.780128.730
47Mike GoodsonOAK352216.31068.8929
48Darren SprolesNOR482445.081108.6328
49Vonta LeachBAL9323.561310.7825
50Jorvorskie LaneMIA13131259.8523
51Brandon JacksonCLE8546.750210.520
52Armando AllenCHI271244.59168.6717
53Daryl RichardsonSTL984754.850228.2116
54Michael BushCHI1144113.615288.1714
55Greg JonesJAX581.60310.613
56Anthony AllenBAL16613.81148.8112
57Robert HughesIND155012012
58Travaris CadetNOR155012012
59Collin MooneyTEN5193.8029.89
60Stanley HaviliPHI6223.67119.59
61Stefan LoganDET3175.670110.678
62Bruce MillerSFO5183.6029.68
63Chris RaineyPIT261023.92258.358
64Lance BallDEN421583.761118.175
65Isaiah PeadSTL10545.4028.44
66Quinn JohnsonTEN451.25028.753
67Lawrence VickersDAL3113.67018.672
68Bernard ScottCIN8354.38028.131
69Owen SchmittOAK210.50180
70Kregg LumpkinNYG9424.670280
71Jamie HarperTEN19301.58347.89-3
72Anthony DixonSFO21783.71237.76-6
73Bilal PowellNYJ1104373.974247.97-8
74Taiwan JonesOAK6213.5016-12
75Isaac RedmanPIT1104103.732287.91-15
76Johnny White2TM8344.25016.13-15
77Brandon JacobsSFO571.4014.4-18
78Lance DunbarDAL21753.57057.14-19
79Chris OgbonnayaCLE8303.75015.63-19
80Da'Rel ScottNYG691.5014-24
81Tashard ChoiceBUF471934.11197.4-30
82Keith TostonJAX17744.35026.12-33
83Knowshon MorenoDEN1395253.784327.81-34
84LeGarrette BlountTAM411513.68277.22-34
85Jason SnellingATL18633.5036-37
86Leon WashingtonSEA23833.61136.43-37
87Fred JacksonBUF1154373.83267.71-38
88Le'Ron McClainSDG14423025.14-41
89John KuhnGNB23632.74146.22-42
90Jalen ParmeleJAX401433.58096.95-44
91Ronnie HillmanDEN843273.891197.52-44
92Mikel LeshoureDET2157983.719477.83-47
93Richard MurphyJAX23924035.96-48
94Toby GerhartMIN501693.381117.08-48
95Robert TurbinSEA803544.430167.43-50
96Montario HardestyCLE652714.171127.25-52
97LeSean McCoyPHI2008404.22457.78-55
98Brian LeonardCIN331063.21076.39-55
99Jeremy StewartOAK251014.04035.84-55
100DeAngelo WilliamsCAR1737374.265337.7-60
101Vick BallardIND2118143.862527.74-64
102Kevin SmithDET371343.62156.19-69
103Mewelde MooreIND13201.54012.69-70
104Peyton HillisKAN853093.641197.22-70
105Felix JonesDAL1114023.623247.41-71
106Baron BatchPIT25491.96145.16-72
107Reggie BushMIA2279864.346437.71-76
108Curtis BrinkleySDG391152.95086.03-79
109Phillip TannerDAL25612.44044.84-80
110Donald BrownIND1084173.861237.24-87
111Rashard MendenhallPIT511823.57096.22-93
112Pierre ThomasNOR1054734.51177.12-97
113Kahlil Bell2TM29762.62044.69-97
114James StarksGNB712553.591136.62-101
115Daniel ThomasMIA913253.574156.92-102
116Jackie BattleSDG953113.273196.91-109
117William PowellARI592163.660106.2-109
118Mark IngramNOR1566023.865297.29-118
119Matt ForteCHI24810944.415457.54-127
120Jacquizz RodgersATL943623.851166.62-134
121Cedric BensonGNB712483.491105.89-153
122Jonathan DwyerPIT1566233.992297.04-157
123Chris JohnsonTEN27612434.56467.44-168
124Jonathan StewartCAR933363.611146.09-182
125Ryan WilliamsARI581642.83084.9-183
126Michael TurnerATL2228003.610407.21-187
127BenJarvus Green-EllisCIN27810943.946557.33-198
128Chris WellsARI882342.665115.67-209
129Steven JacksonSTL25710424.054497.23-211
130Shonn GreeneNYJ27610633.858527.26-218
131Ryan MathewsSDG1847073.841356.8-229
132LaRod Stephens-HowlingARI1113573.224155.96-231
133Rashad JenningsJAX1012832.82165.57-250
134Alex GreenGNB1354643.440175.33-367
135Trent RichardsonCLE2679503.5611386.52-409
136Darren McFaddenOAK2167073.272355.89-466

I think I like this better than a system based purely around yards per carry, but there are some obvious drawbacks. I don’t like that Andre Brown is ahead of Arian Foster, and in general, Foster seems way too low (you can read more about my thoughts on him here). And obviously it’s hard to separate out a player’s ability from his production, and I still like Trent Richardson even if he didn’t produce in an unfavorable situation last year.

Four years ago, I ran a series on the most dominant running backs of all time. At some point, I plan to revisit that topic. I dismissed the idea of using yards per carry to grade running backs then based on the following logic:

Is 270/1100 better than 330/1100? One argument that I’ve certainly used before is “they got the same number of yards, but the first guy’s team had an extra 60 plays with which to gain more yards!” The natural response to that is “why would a coach give a RB 330 carries if he was only getting 3.6 per carry?” From there, we have a two different answers. Either: a) the RB wasn’t that good but either the coach was dumb or the backups were really bad, or b) the RB was good and his YPC is misleading.

Once again, with rushing, I think median carry is a more telling number than average carry. Yards per carry is not a very good measure of central tendency. On the other hand, we can infer that if a RB is getting a high number of carries, he’s doing something right. Carries themselves are highly correlated with greatness.

Terrell Davis, Edgerrin James, Curtis Martin, LaDainian Tomlinson, Eric Dickerson, Clinton Portis, Eddie George, Walter Payton, Barry Sanders, Jim Brown, Emmitt Smith, Ricky Williams and Earl Campbell. Those are the RBs with over 19 carries per game for their careers. A RB that gets carry after carry is doing something right. Maybe he’s consistently getting gains, maybe he’s running hard despite a bad OL, or maybe he’s able to kill the clock without fumbling. All of those things are good. A list of the top RBs by yards per carry? Jim Brown, Barry Sanders, Paul Lowe, Robert Smith, Joe Perry, Wendell Tyler, Greg Pruitt, James Brooks, Tiki Barber, Hugh McElhenny, O.J. Simpson, Fred Taylor and Charlie Garner all have career averages over 4.6 YPC. Ignoring the overlap, I’d prefer the first list.

Remember, teams can choose to pass instead of run. So if a RB is getting 350 carries, it can’t be just because his RB teammates are bad. It’s got to be because the team’s QB is bad, too. And if a team’s QB is bad, its other RBs are bad and one guy keeps getting carry after carry, then he’s pretty valuable to his team. And if year after year he gets these carries, he’s definitely doing something right. A RB with 1600 yards on 400 carries may be just as valuable or dominant as one with 1600 yards on 325 carries. After all, the obvious question for the latter RB is — why didn’t he get more carries? Perhaps the former RB was getting three and four yards on every carry, a very valuable trait.

I’ll open up the rest of the post to you. What formula would you devise to grade running backs? Leave your thoughts in the comments.

  1. Based on the following formula: ((10.5-8.0)*348). []
  • The obvious thing that seems to be missing at this point is fumbles, which should be a big part of the total. I would guess they are worth approximately -28 yards each (if an interception is worth 45 yards and the defense recovers fumbles on rushing plays by non-QBs 62% of the time, that comes out to -27.9). The worst offenders in 2012, Willis McGahee, Bryce Brown, and Fred Jackson, would all lose substantial value.

    For what it’s worth, adjusting for fumbles does also substantially reduce the value of some of those career ypc leaders, like Wendell Tyler, Greg Pruitt, James Brooks, Tiki Barber, and Hugh McElhenny.

    • Chase Stuart

      Yes, that is definitely correct. I was short on time otherwise I would have included it, but I agree that they need to be accounted for in any rushing formula.

  • Eric

    If I knew the formula I wouldn’t have to read you!

    My completely random, typing on an iPhone with a 4 year old laying one me and Disney Jr. on the TV thoughts are:
    Several formulas are needed. Too
    many systems, too many running backs asked to do different things. Look at Alfred Morris. Above he’s three, while Brian Burke has him ranked much lower. Part of it was he was not the high leverage guy on that team. Not (I hope) for lack of skill but for lack of play calling predicated on him performing in high leverage situations. Doesn’t mean he’s a bad option or would’ve failed but as a non-pass catching back and on a team with RG3 it just wasn’t his thing.

    So formula ideas:
    -Throw out yards above 40 or 50 per carry. At that point it seems safe-ish to say that it’s a broken sprint to the end zone. How does the back run when near the other defenders is what I care about because that’s where he’ll run 98% of the time.
    -instead of high leverage, how about low leverage? It may yet change as passes continue to yield the best outcomes, but how a guy runs on 1-10 still matters to me a lot. Run well on that down and you may reduce the number of high leverage plays because your team is converting shorter 2nds and gain the lead. Hell, a guy running well on 1-10 could very well not score as the team scoring throwing long on 2nd and 3. So 20 yards of value go out the window because you punish the shit out of the defense, setting your WR up for a field day?

    Now the 4 year old is asleep.

    The point is: I would like to see several advance stats for RBs so fans could match player to style. Maybe a guy is not a big play back (Morris) but the team doesn’t need him to be. Or maybe the team needs the RB to change the game in one play (Spiller). In this day of myriad styles of offense there should be advance stats quantifying it all, not one formula trying to fit all in one.

    • Chase Stuart

      Agreed with all of this. I think for current players, you’re right that we can and should get much more granular.

      My goal here is to start the process of building a formula to measure across players from all eras, which unfortunately restricts us a bit.

  • Eric

    To add to that: A scoring free stat. Again, it’s not to be used to say a is better than b but that if a’s value comes from scoring and b doesn’t *need* to score on his team than b being better at everything but scoring make him a better fit on his team than a.

  • Scott Tanner

    I’m admitting out front to being biased as a Bears fan, but this formula seems like it was specifically designed to under-value Matt Forte. There are broader issues, but talking about Forte teases out some of the problems. First off, it doesn’t take into account the receiving value of a RB, which is a really important part of a lot of offenses today. It’s also unfairly punitive to guys who play on teams with designated goal-line backs, since that’s obviously going to play a role in their rushing TD totals. Nobody who watched the Bears last year would say that Michael Bush was better than Matt Forte (and while the eye-test is imperfect, you do get to see 2 RBs running behind essentially the same line week in and week out) and yet Bush is ranked nearly 50 spots higher almost entirely due to being a TD vulture.

    Maybe there’s an argument to be made that being able to get into the endzone in short yardage situations is a specific skill. I’m open to that argument, actually. Forte in particular has proven pretty inept at it (2010-2011 was a comedy of errors on the goal line for the Bears) but it still seems weird to value this alleged skill so high. Even granting it’s a skill with certain value, I’d much prefer the guy who is really good from the 20 down the 5 than the guy who has a “nose for the endzone” from 5 yards in.

    So I’d try to incorporate the receiving totals somehow, and also reduce the value TDs give. And as someone else mentioned, fumbles are quite important as well. The Bears fan in me who thinks Forte is still somewhat underrated would love for there to be a way to account for his atrocious offensive line, but it’s obviously pretty difficult to incorporate that into an empirical model without making a lot of subjective judgements.

    • Richie

      Maybe there’s an argument to be made that being able to get into the endzone in short yardage situations is a specific skill. I’m open to that argument, actually. Forte in particular has proven pretty inept at it (2010-2011 was a comedy of errors on the goal line for the Bears) but it still seems weird to value this alleged skill so high.

      I’m going to agree with you there.

      After all these years of reading football analysis, I am still unconvinced that individual players deserve such a high bonus for scoring touchdowns, when scoring touchdowns is often pretty reliant on play calling and situations.

      Did Emmitt Smith really have some reduced skill for scoring touchdowns in 1993 when he scored 9 rushing TDs (on 3.2% of his attempts) compared to 1992, 1994 and 1995 (when he averaged close to 20 TD’s on about 5.5% of his carries)? As we learned on Saturday, he won the MVP that year. (Which is also probably unusual for a RB to win an MVP with only 9 rushing TD’s.)

      • Chase Stuart

        I think it’s helpful to think of it as a rearview/explanatory thing and not a predictive thing. We still give players huge penalties for INTs even though that may not be appropriate, right?

        • Richie

          Yeah, but to me it’s clear that when a QB throws an interception he has (usually) made a mistake that is very costly. But when CJ Spiller gets tackled at the 1-yard after a 15-yard run, and then Fred Jackson scores a 1-yard touchdown, I’m just not convinced that he has contributed 20 additional “yards” of value.

          I am unconvinced that there is any kind of different skill in scoring touchdowns. Maybe I need to re-read the part of “Hidden Game of Football” that talks about the touchdown bonus.

          Side note: is it just me, or is this site loading very slowly today?

          • Chase Stuart

            The site was loading very slow for me earlier, too. I have no idea why and my web expertise is limited.

            I think your example is a bit unfair. That’s like saying why penalize a QB for a tipped interception? The average rushing touchdown is not a one yard plunge on 1st and goal. I’m not sure there is a significant difference in skill level when it comes to scoring touchdowns, but I think some touchdown bonus is needed. What do you think of giving a different bonus for a different length?

            • Richie

              Yeah, I know not all interceptions are created equal. (That’s why I said ‘usually’.)

              Most of my thoughts on touchdown bonuses are just based on my feel and intuition. I could be totally wrong. But I’m really not sure if a RB deserves any bonus for touchdowns. I think scoring short touchdowns is mainly due to the good fortune of having a team get the ball close, and then the coach calling a running play. I think long touchdowns are also just kind of the luck of having no more defenders left to tackle you. A good RB has the ability to break tackles and make people miss, and doing that enough will result in touchdowns.

              I just don’t buy in to the theory of “player x has a nose for the end zone”. I think all players want to score touchdowns.

              I did a quick search of RB seasons to look for guys with similar seasons, except big TD differences. I found a good one. In 2005 Shaun Alexander and Tiki Barber had similar yardage totals and similar yards per carry. But Alexander scored 27 TD’s to just 9 for Barber. Was Alexander really 3 times better at scoring than Barber? My memory of that season for Alexander was that every time they would get near the goal line, Walter Jones would open a huge hole and Alexander would score without being touched. I realize that this is an extremely small sample size.

              • Richie

                18 of Alexander’s touchdowns were 5 yards or shorter. 10 of them were 1 yard. Does he really deserve a huge bonus for this? (Admittedly, part of the reason he scored those TD’s was because he helped get the team in position.)

                3 of Barber’s touchdowns were 5 yards or shorter. (I have no idea how many 1-yard opportunities he had, but he didn’t score any from 1 yard out.)

                • Chase Stuart

                  Would you be more comfortable with a formula that gives a small amount of credit for short TDs and a bigger amount for larger ones?

                  • Arif Hasan

                    What about a formula that set 6 points to 20 yards, then scaled the yard value gained on touchdowns by the amount of EPA gained? The EPA gained on that 1-yard TD was not very high, and not worth 20 “yards” of value.

                    I will say this could overvalue those 60+ yard TD runs, but given that the running back is sacrificing a ton of “first down” bonuses in “yards” gained, that could be OK.

              • Chase Stuart

                How would you create a running back ranking system?

    • Chase Stuart

      Thanks for the comment, Scott.

      No doubt receiving yards will be included in the final formula. I punted on that issue today.

      I don’t think this is unfairly punitive to guys who plan on teams with designated goal line backs. For starters, they miss out on a lot of the “run for 0 yards” carries that a goal line back would have, and that benefits them in this formula. In addition, there’s no reason to pretend that rushing for a touchdown isn’t a good thing for which a player should be rewarded.

      As for Forte/Bush, this says Bush is an average back, and I don’t think he’s the real problem. It’s Forte. His actual rank is a bit of a red herring — this says he was below average, but since he had so many carries, he comes in far below average. As to why he is below average? His TD and first down rates were not impressive, and that’s not good if your YPC average is poor. Of course, I have many issues with YPC, so I don’t necessarily agree that this is a correct way of doing things.

      Thanks for helping flesh these things out.

  • Wade Iuele

    I agree that receiving should be accounted for.

    A few thoughts on situational stats that could also prove useful:

    * 5+ yards gained on 1st or 2nd down (not resulting in a 1st down).
    * 1st downs achieved in the second half while leading.
    * Touchdowns scored or long yardage plays when trailing.

    • Chase Stuart

      Good stuff, Wade. However, my main goal is to come up with a formula that we can use across eras, so unfortunately, I’m restricting myself to the more basic stats. But I agree that we should use things like that for modern players (and I will count receiving production for all players).

  • sn0mm1s

    I posted in your original article and my opinions haven’t changed. I still loathe short TDs being worth much more than the yardage gained. They are an easy stat to game and the most successful play call from the 1 or 2 is a QB run. Burke’s EPA is a pretty good measure but I wish he would break it out into rushing and receiving. I think FO’s DYAR/DVOA is poor measure of an individual and leads to some ridiculous results. I do like FO’s broken tackle project though. I would be willing to bet that Burke’s EPA/play in combination with BT/play, and yards after contact (with contact defined by FO’s broken tackle) would be a very good methodology.

    • Eric

      Completely agree about a rushing/receiving breakout being needed for Brian Burke’s EPA. It’s meaningless to compare a back that catches a lot in situational ball (and rewards, highly, those who do this well) and those that only run, rarely catch.

      • Chase Stuart

        I e-mailed Brian to see if he would accommodate.

    • Chase Stuart

      Thanks. One thing I could do is provide different bonuses for the lengths of the TD, which I do have data on going back to 1940. The issue is we would only have the length and not the down. What sort of bonuses would you think are appropriate for each length?

      Re: Burke, I will e-mail him and see if he can start splitting it out.

      • sn0mm1s

        In theory, Burke’s EPA bakes that bonus into the play (along with turnovers) – which is why I really like what he does. I don’t believe he takes into account what is expected for a particular play (which is something that I think is required for judging an individual). For instance, if it is 3rd and 1o on your own 20 and you hand off to the RB I don’t think any reasonable coach would expect the RB to convert the 1st down. I *think* Burke penalizes a RB pretty heavily, in regards to expected points, for not converting the 1st down even though the likelihood of it happening is slim. It would probably be more accurate for Burke to tally up the expected average (or median) rush across the entire NFL given a particular DnD and then determine the expected points difference from the expected result compared to the actual result.

        • sn0mm1s

          EDIT: determine the points difference from the expected result compared to the actual result.

  • I was thinking the same thing as many others have said, it puts too much value on an area that often times isn’t controllable and ignores receiving production as well as undervalues rushing in general, this seems to really benefit short yardage rushers, I mean what is Jacob Hester doing on the list, but because he had two TD’s and five first downs he’s ahead of much more productive backs.

    In this modern NFL that has countless specialist positions (from the slot WR and CB to purely receiving TE’s and the nickel LB) the short yardage RB is one of them. While there are still the obvious three down running backs most teams have two backs and one of them is a short yardage back whose going to come in on 3rd and short as well as the red zone. If you combine that situation with a formula that puts extra value on 3rd downs and TD’s you are going to see the undervaluation of backs who aren’t those types of backs, especially ones who stay in on 3rd downs as receivers or blockers.

    This was a fun way to look at RB’s but it’s meant to find what you were looking for than anything else really. Also I’m concerned about the devaluation of YPC, hardly a perfect stat but this formula actually puts a ton of value on it by really penalizing those with more carries and just average yards. Those who actually have high YPC are going to do well on this list, especially if they had very few carries, mostly in the RZ and on 3rd downs. We’ve seen that backs who maintain a high YPC in situational football don’t usually succeed once they have to carry the ball 250+ times but this seems to reward those backs for limited play time.

    Overall, fun formula, seems weak unless you are trying to prove the goal line back is more important than anyone else.

    • Chase Stuart

      Hester’s spot doesn’t concern me much with the formula — simply using an “over replacement” standard or a minimum carries threshold will change that.

      What I think is important to remember is in traditional statistics, short-yardage backs are heavily punished for their 3rd and 1 carry for 1 yard or if they run twice on 1st and goal for the one and get a touchdown the second time, that looks bad on the stat sheet without bonuses for touchdowns/first downs.

      I also think a third down back who is good in short yardage is valuable, so I don’t have a problem with him ranking high here. Being successful on such a high leverage play should be rewarded.

      In the end I’m not sure what I will do with first downs as far as historical comps, because I only have team level rushing first data historically.

      • I agree about the value of short yardage backs and how they do tend to get penalized by the traditional YPC. Like most positions there are different role players. Slot WR’s are generally going to have a lower YPR than an outside WR, a short yardage back will have a lower YPC than the teams main back. That doesn’t diminish what they do they just have different roles. That is why I’d rather stick with a range of metrics over just one metric since no matter which way you slice it, one type of player gets snubbed, no matter the position.

        This is a nice way to give more value to short yardage backs but I think your formula overvalues them instead of bring balance. You are right that getting a 1st down should be rewarded but it comes down to the question of value and role and it’s getting increasingly hard to compare a backs value when they aren’t asked to do the same thing.

        Overall in reply to your lower comment, I know you don’t intend this to be a super metric for RB’s and like you did with the WR’s you are just exploring different ways to value backs. Overall these are meant to cause disagreement, if no one disagreed with it though that does likely mean it’s a perfect metric, otherwise we adjust and perfect through discussion.

        Love your work as always.

        • Chase Stuart

          Thanks Topher. I’m in the middle of a 100 hour work week here, so I haven’t had the time to artfully explain my thoughts. That should come at some point, though.

  • As for creating a formula to judge RB’s, I don’t really see that as necessary, I do find it interesting that so many advanced statistics guys in the NFL are focused on creating the one perfect metric for every position but in reality it’s best to just look at a broad range of statistics and form your image based on lots of data instead of constantly tweaking formulas and weights. When I judge a RB I look at many things, most of which you left off the formula. I look at yards from scrimmage, yards per carry, 1st downs, total TD’s, drops, fumble rate, attempts per game, and broken tackles. Depending on what I am looking for in that situation (whether it be whose the best goal line back or who do I want to touch the ball more than 20+ times a game) I’m able to easily solve those questions.

    Obviously you aren’t presenting your formula as the be all and end all of RB value so it’s not fair to be too harsh, but it did have a goal in mind and it accomplished that goal, we just may not have agreed with it. I think we all agree 1st downs and TD’s are important but I think we disagree on how much value you put on them, there seem to be more exceptions to your formula than actual rules.

    • Chase Stuart

      I understand where you’re coming from, Topher Doll. Mind you this is just step one in the process, and soliciting reader feedback is an important part of that. I hope you like where we end up.

  • Richie

    Totally unrelated.

    I was just looking for info about the release of Footballoutsiders almanac for 2013, and saw this little gem:
    ” including the addition of Chase Stuart of Football Perspective ” Nice!


  • Dave

    I think you solve the Andre Brown problem by doing the following:

    Figure out the max number of carries for the RB in the given season. Then regress all other RB’s stats as if they were phat many carries.

    So regress their yards per carry, TD rate, and first down rate. I picked the top 32 rbs in terms of carries. The avg ypc was 4.36, td% 2.8% and 21%FD. I then recalculate their adjusted yards per carry based on assuming every RB had 351 carries. Obviously the closer the RB is to 351 carries the less his stats are going to be regressed to the mean.

    But then I calculate their value based on their true carries.

    Brown still ends up ahead of Foster but it’s much closer.

    Some of the extreme changes:

    Foster moves up 10 spots. Doug Martin moves up 9. Murray and Rice move up 16 spots.

    Lewis drops 17 spots. Hester drops 13. Young drops 11/ DJ Ware drops 8.

    This method seems to help deal with the small smaple size RB’s.

  • Red

    Regarding the TD bonus…I don’t think scoring TD’s is a special “skill”. However, I do think a bonus is necessary to compensate players with a lot of goal-to-go attempts, as their YPC is artificially deflated by the truncated field. Scoring from the 1 yard line shouldn’t boost a RB’s value too much, but it definitely shouldn’t hurt his value, either. Could you go through the play-by-play database to see how much YPC is depressed inside the 10 yard line? That would give us a clue as to how much compensation a player should receive for those carries.

    BTW, I think the above paragraph also applies to QB, WR, and TE.

  • Johan Moran

    “Remember, teams can choose to pass instead of run”

    And they can choose to run instead of pass.

    If you’re going to give extra credit to wide receivers that happened to play for run first oriented teams, don’t you have to do the same for running backs who played for the inverse?

    It seems to me that if you are going to infer that a running back who is getting lots of carries must be doing something right, then we must also assume that a wide receiver who is getting lots of balls thrown in their direction must also be doing something right. Yet, in your analysis of wide receivers, you tend go soft on many who’s numbers don’t stack up as just being a product of a run oriented offense. Maybe there is a reason they weren’t relied on as much by their coach that goes beyond the teams offensive philosophy