Writers of Pro Football Prospectus 2008

16 Apr 2010

ESPN: RB vs. OL? Take the Speedster

Today's feature in ESPN's draft-related "Samsung Next Level" series comes from our own Vince Verhei, who looks at which move does more to improve a team's running game: drafting a back high, or drafting a tackle high? While FO generally believes that good left tackles are harder to find than good backs, the backs actually do more for your running game -- because, of course, you are generally drafting a tackle that high because of his pass protection abilities. Thanks to the Samsung sponsorship, this article is free, not Insider.

Posted by: Vincent Verhei on 16 Apr 2010

23 comments, Last at 20 Apr 2010, 1:01am by tuluse


by Jimmy :: Fri, 04/16/2010 - 12:10pm

It is very, very difficult to determine which comes first the chicken or the egg on this stuff. Maybe a team took a tackle over a back because they already had a back. Or vice versa. Or maybe their star QB is tired of getting beaten back into the stone age every snap. Seperating the entire offensive line from a RB is nearly impossible (as most around here would agree - with one prominent dissenter), trying to distill out the effect of a single blocker has to be trickier still. And then there are cases like the Bengals last year who did take a tackle in the top ten and then left him riding the pine all year.

These NFL teams don't make it easy for you Vince.

Also Hooray for Samsung!

by Karl Cuba :: Fri, 04/16/2010 - 2:18pm

The Bengals provide another example, Benson. In Chicago he had a crap line, a crap quarterback and a crap offensive coordinator, as a result he performed like crap. He arrived in Cincy, who had a line that was good enough to leave the aforementioned chubby tackle on the bench, a good quarterback and a working offensive system, all of which combined to make him look like a franchise runner.

by Marko :: Sat, 04/17/2010 - 5:46pm

There are some additional reasons why Benson performed like crap in Chicago. He had a crap attitude, was lazy and thought that he should be given a starting job without actually earning it. He did not run with any burst or passion and rarely broke any tackles. (When I saw him this year, he looked like a completely different player. If he ran like that with the Bears, fans and teammates would have loved him.) He was a liability in the passing game both as a receiver and as a blocker.

He also had off-field issues, starting with his lengthy and acrimonious holdout which made him miss all of training camp. He did not sign until the middle of the preseason. Shortly after he finally signed, he left the sideline during a preseason game and went home. (Although he wasn't suited up and obviously wasn't playing, you can't do that.) He then pouted and became even more of an outcast when this incident was reported (which happened because one of his teammates told a reporter about it). Starting RB Thomas Jones was loved by teammates for his production, work ethic and professionalism. Benson was the anti-Jones.

Oh, and the line wasn't crap back then. It was a solid line that featured perennial Pro Bowlers at C (Olin Kreutz) and G (Ruben Brown), as well as a very good LT (John Tait). The Bears' running game was pretty good with Thomas Jones. When Benson played, it was not nearly as good.

by Karl Cuba :: Mon, 04/19/2010 - 12:38pm

I can remember some flashes of brilliance from Benson in Chicago. Didn't he put up pretty good numbers the year before Brown ran out of steam and Tait began to decline? However, I can see your point about him being an utter tool in Chicago but the charges were dropped in his 'final straw' incident on the boat.

by tuluse :: Tue, 04/20/2010 - 1:01am

His sophomore year he looked really good and every one was calling for him to replace Jones. Until the Superbowl that is, when his biggest play was losing a fumble.

by fek9wnr (not verified) :: Fri, 04/16/2010 - 1:02pm

Thanks to this sponsorship, I will now sit and watch my SamsungĀ® TV for several hours.

by GV (not verified) :: Fri, 04/16/2010 - 2:24pm

What about the teams that don't draft either a running back or a tackle? A baseline would allow for some comparison.

Bad teams are likely to win more the next year simply due to regression to the mean. If they're winning more they won't get stuck in pass-only mode as much.

by Bowl Game Anomaly :: Fri, 04/16/2010 - 3:03pm

Even if the conclusion of the article is true regarding the Redskins and the 4th pick, that still doesn't mean they should take Spiller over Okung. The article only addressed the running game. The Redskins are not planning to take Okung purely to improve their running game, being that their pass protection sucked last year and they need to replace Samuels regardless. I'm not saying I like picking up Johnson or Parker (I don't), but the article did not address whether teams get more overall value from highly drafted backs or tackles. I suspect the tackles provide better value in addition to being a scarcer commodity. So the real conclusion of the article should be, "While the Redskins will correctly choose Okung over Spiller, their running game might have been better served by choosing Spiller rather than relying on aging and washed up veteran RBs," which is hardly Earth-shattering.

by tuluse :: Fri, 04/16/2010 - 5:30pm

This article is disappointing, why not measure running game in DVOA?

by Vincent Verhei :: Fri, 04/16/2010 - 5:49pm

This was written for ESPN's audience, not Football Outsiders' audience. We didn't have space to introduce and explain DVOA to a new readership, so we got right to the point using yards per carry, which anyone can understand.

We actually studied the issue using yards per game, yards per carry, and DVOA, and the results were the same across the board: RB teams fared much better in Year 1 and Year 4, and teams were about the same in Years 2 and 3.

by tuluse :: Fri, 04/16/2010 - 6:25pm

Thanks for the reply.

by Jerry :: Fri, 04/16/2010 - 5:47pm

I can understand why it wouldn't be in the ESPN piece, but how does rushing DVOA compare for the teams involved?

ETA: I see tuluse had the same idea.

by Boesy (not verified) :: Fri, 04/16/2010 - 6:14pm

"then picked Miami's Bryant McKinnie with the fourth overall selection."

McKinnie was 7th overall that year.

by geekosphere101 (not verified) :: Fri, 04/16/2010 - 11:12pm

I'm pretty sure 'Soda was supposed to pick 4, but (at the time) inexplicably allowed time to expire, i n which the following teams rushed up with their cards. In hindsight, since McKinnie was their guy, they ended up getting him and saving some cash on his contract.

If I'm the Redskins, even if I didn't have Portis, Johnson, and Parker, I'm still taking Okung no matter what. Spiller may be a dynamic back, but the value Okung provides as a pass protector is VITAL. McNabb, who isn't exactly known for his ability to stay healthy, has no chance of surviving behind that line if they don't get a franchise left tackle.

I did find the article interesting, though. I certainly hope AD rebounds this next year.

by Jovins :: Sat, 04/17/2010 - 10:13pm

Actually, that was the 2003 draft, in which the Vikings still got Kevin Williams. They were supposed to have the 7th pick; they picked 9th. Kevin Williams was slotted into the 9th pick's salary, and is considered the 9th pick.

So wrong on both counts.

by Noah Arkadia :: Sat, 04/17/2010 - 11:49am

Yeah, it's good and a bit surprising. Nevertheless, I've never bought into the FO theory of not drafting a RB high. As a Miami fan, I've been through enough decades of RB misery before Ronnie and Ricky to be so nonchalant about it.

by Mr Shush :: Sat, 04/17/2010 - 12:32pm

My thoughts, further to what others have said:

1. The sample is way too small to have any confidence in the conclusion.

2. Left tackles generally have much longer careers than running backs. The LTs may have more aggregate career marginal value in the running game, even if they have less value per year.

3. CJ Spiller is not worth a top ten pick, so even if an elite RB prospect is better for the running game than an elite LT prospect, Okung may be better in this respect than Spiller. Spiller at 4 would be a colossal reach.

4. Why does the article look at team YPC not RB YPC for the backs in question? I find it hard to credit Darren McFadden for Justin Fargas and Michael Bush's productivity. Relatedly, a bad RB pick is unlikely to have much negative effect on his team's YPC, because he won't get many carries, whereas a good one will presumably get to tote the rock a lot, skewing the influence disproportionately towards the successes. The tackles, on the other hand, will in the vast majority of cases be on the field every down, even if they're not much good.

by DaveRichters :: Sun, 04/18/2010 - 9:19am

1. The sample is way too small to have any confidence in the conclusion.

First let me say I am not defending the article or claiming that its conclusion merits confidence, but what's wrong with the sample size? There are 9 teams in the RB group and 14 teams in the OL group, there is absolutely no reason to say that more teams and more years are needed to learn anything. The analysis in this article, like the previous ESPN QB article, is completely worthless, but it is impossible to tell if a larger sample is needed or not. Without some measure of the spread in the two groups we can't tell if there is even a difference at all.

The bigger problem with this article is even if the differences are significant, it still in no way suggests that teams wanting to improve rushing yards per game should draft a running back over a tackle. I know this is a difficult question to address with performance based statistical analysis, but this is not a worthwhile first step.

by Bowl Game Anomaly :: Sun, 04/18/2010 - 6:57pm

9 in one group and 14 in the other? Yeah, that does mean the sample size is way too small. Come back when you have 30 and 30.

Sure, you might reach statistical significance with those small samples, if the effect is huge (which is isn't). But even that doesn't tell you much except "further study is needed to confirm these results."

by DaveRichters :: Sun, 04/18/2010 - 9:26pm

Yeah, that does mean the sample size is way too small.

I guess that's a fair comment given how FO treats statistics, but it is not a fair comment in general.

Sure, you might reach statistical significance with those small samples

Sample size is taken into account in statistical hypothesis testing. Are you suggesting that about 95% of the population needs to be tested to have any confidence in the results? I'm not sure what claim you are making. Are you criticizing me for saying the sample is not too small, or are you criticizing the article for not actually using any statistics?

if the effect is huge (which is isn't)

That's the crux of this mountain. I can't tell, although I guess you can, how big the effect is. Knowing only the means of two groups it is impossible to tell how big the effect is.

But even that doesn't tell you much except "further study is needed to confirm these results."

If you mean this as a general comment, it is out of place, though I agree with it on logical terms. I'm surprised to find a kindred spirit who is not a philosopher, and I assume most people here will have another opinion about what a statistically significant difference indicates. If it is a comment specific to this study, I disagree, it doesn't even tell you that.

by capt. Anonymous (not verified) :: Sat, 04/17/2010 - 1:27pm

Anyone doesn't take Clausen is leaving a present for the rest of the league. Redskins should take Clausen and then make some other team pay them handsomely for his services.

Russell Okung is not a franchise player.

by Formersd (not verified) :: Mon, 04/19/2010 - 4:35pm

The other point to make is that the LT is key because of pass protection. Most teams run to their right, so the Guards and Right Tackle have a lot to do with the running game. Since the OL is a unit, it makes sense that adding a RB has more impact that one OL.

by Vincent Verhei :: Mon, 04/19/2010 - 4:58pm

Most teams run to their right

Not true. See the very last line of data on our offensive lines page. Last year the average team ran left on 24 percent of the time, up the middle 50 percent, and to the right 26 percent. And that was an unusually right-handed year:

2008: 25/50/24
2007: 26/50/24
2006: 27/49/24
2005: 26/51/23
2004: 26/51/24

(Rounding errors mean the totals don't add up to exactly 100 percent.)

Realistically , teams run right about as often as they run left.