Tuesday, March 15, 2011
Does Size Matter For Running Backs?
I noted in a recent column that Jerome Bettis — all 250 pounds of him — wasn't a particularly good short-yardage runner. In 13 NFL seasons, Bettis led his own team in TDs only four times. Emmitt Smith weighed 40 pounds less, but he led his team in TDs 12 times. Even as other positions become increasingly specialized, the use of a giant battering-ram RB in short-yardage situations doesn't seem to be gaining traction. Is this a mistake?
For every season from 1991-2010, I looked at the top 10 rushers that year, slicing and dicing their statistics to see how height, weight, and Body Mass Index (BMI) might play a role in short-yardage success. 1991 is a good starting point because that's the year the NFL began keeping first down data for individual players. There are a lot of numbers coming up, so for those of you whose eyes get blurry at that kind of thing — or if you just can't stand suspense — being taller and heavier does not help in short-yardage situations, and it may be detrimental.
The average top-10 rusher, looking at all 20 seasons, was 71 inches tall (5'11") and weighed 219.4 pounds, with a BMI of 30.6. Basically Garrison Hearst, or Clinton Portis in Washington — not the skinny Denver years. BMI is a very basic measure of the relationship between an individual's height and weight. A higher BMI indicates more weight per inch of height. Essentially, a higher BMI means a fatter (or at least more muscular) running back.
The height, weight, and BMI of the top 10 rushers has remained fairly constant over the past two decades, from an average of 70.5 inches, 219.4 lbs., 31.0 BMI in '91 to 70.6 inches, 215.7 lbs., 30.5 BMI in 2010 — basically the same. The year-to-year fluctuations in between are small enough that they appear to be coincidences rather than patterns.
I examined relationships between these three size factors and a number of statistical categories, including: rushing yards, rushing TDs, rushing first downs, percentage of rushes yielding first downs, and fumbles. Note that the NFL counts a touchdown as a first down, so those count in both categories. In all cases, the height and weight I used are those given on a player's NFL.com page as of March 1, 2011.
The optimal height for a running back is 5'9". This was one of the few conclusions of my research. I broke the 200 RB seasons into seven groups:
(1) less than 5'9" (14 seasons, includes Barry Sanders, Maurice Jones-Drew, and Ray Rice)
(2) 5'9" (19 seasons, includes Emmitt Smith, Priest Holmes, and Frank Gore)
(3) 5'10" (47 seasons, includes Thurman Thomas, Marshall Faulk, and LaDainian Tomlinson)
(4) 5'11" (54 seasons, includes Terrell Davis, Jerome Bettis, and Chris Johnson)
(5) 6' (16 seasons, includes Edgerrin James, Ahman Green, and Willis McGahee)
(6) 6'1" (32 seasons, includes Ricky Watters, Corey Dillon, and Adrian Peterson)
(7) more than 6'1" (18 seasons, includes Chris Warren, Eddie George, and Steven Jackson)
The players mentioned are examples, not the only top-10 RBs of that height. Group 2, the 5'9" players, outperformed the other groups in virtually every category: rushing yards, rushing TDs (by a huge margin), first downs, and first down %. The advantage is so large that you can subtract Emmitt Smith from the group entirely, and it still leads in most categories, plus it passes Group 1 (the shortest players) for best rushing average (4.84).
The further over 5'9" the players went, the lower their TD totals got. The data suggest that the advantages gained by shorter players (low center of gravity, tough for defenders to see behind the line) outweigh the advantages of a taller player (arm length, extra inch or two if they fall forward). In a goal-line situation, you simply aren't going to fall forward very often. Here's the 5'9" group with Emmitt, without Emmitt, and combined with the shortest group into a 33-season sample of all the RBs under 5'10":
This isn't about sample size, and it isn't about Emmitt skewing the stats. Shorter RBs, when they are successful, gain more yards and more first downs, score more touchdowns, and have better averages than successful RBs who are tall. Height in excess of 69 inches is a disadvantage for running backs, including in short-yardage rushing situations.
For the categories tracked, a weight between 206-225 pounds appears to be ideal. Lighter-weight RBs outperformed the heavier backs in most categories, but in the statistics linked to short-yardage performance, neither very light nor very heavy runners equaled those in the middle. I sorted players into five weight classes:
(1) 205 lbs. and below (25 seasons, includes Barry Sanders, Tiki Barber, and Chris Johnson)
(2) 206-215 lbs. (62 seasons, includes Thurman Thomas, Marshall Faulk, and LaDainian Tomlinson)
(3) 216-225 lbs. (46 seasons, includes Emmitt Smith, Edgerrin James, and Adrian Peterson)
(4) 226-235 lbs. (46 seasons, includes Rodney Hampton, Eddie George, and Shaun Alexander)
(5) over 235 lbs. (21 seasons, includes Jerome Bettis, Jamal Lewis, and Steven Jackson)
The middle category performs best in the three stats we would expect to be associated with short-yardage success: TDs, first downs, and first down percentage. The further removed a group is from the center (216-225 lbs.), the more poorly it performs, except that the lightest backs have the second-highest first down percentage. The heaviest backs are clearly the weakest, last or second-to-last in almost every category. These data suggest that speed and elusiveness outweigh mass even in goal-line and short-yardage scenarios. A lighter back might hit the hole quickly, beat defenders to the outside, leap over the pile, and slip through tackles in a way that larger players are unable to.
It bears mention that Emmitt Smith is listed at 216 lbs., his weight with the Cardinals, thus barely edging into the middle weight class. Smith played most of his career 5-10 pounds lighter than that, and was such an incredibly successful short-yardage runner that he single-handedly skews the statistics. I would suggest that the 206-215 range is actually at least as advantageous for short-yardage and goal-line situations as the 216-225 category. The stats are basically even if you move Emmitt (the only 216-er) to the lighter weight class.
The formula for BMI is kilograms divided by (meters squared): BMI = kg/m². Or, for us heathens in the U.S., multiply your weight in pounds by 703 and divide by the square of your height in inches: BMI = (703 * lbs) / in². Higher BMIs indicate more weight per height. A healthy BMI is usually between 18.5-24.9, though this is acknowledged to be inaccurate for athletes with high muscle mass.
I used five groups here:
(1) below 29 (24 seasons, includes Ricky Watters, Adrian Peterson, and Chris Johnson)
(2) 29-29.9 (54 seasons, includes Thurman Thomas, Eddie George, and Brian Westbrook)
(3) 30-30.9 (53 seasons, includes Barry Sanders, Marshall Faulk, and LaDainian Tomlinson)
(4) 31-31.9 (35 seasons, includes Emmitt Smith, Priest Holmes, and Shaun Alexander)
(5) 32 and over (34 seasons, includes Jerome Bettis, Ricky Williams, and Maurice Jones-Drew)
As you might expect, the 31s blow away the pack. Smith, Holmes, and Alexander all set single-season TD records in the last 20 years, and this group is clearly and dramatically the best in the short-yardage categories.
It seems apparent that the optimal BMI for a short-yardage back is something very close to 31.5. The other groups are all roughly equal, and all far below the 31.x-BMI group.
For all 20 seasons studied, I broke down the first through tenth in several statistical categories, producing a group of (for instance) all the leading rushers: Emmitt Smith, Emmitt Smith, Emmitt Smith, Barry Sanders, Emmitt Smith, Barry Sanders, Barry Sanders, Terrell Davis, Edgerrin James, Edgerrin James, Priest Holmes, Ricky Williams, Jamal Lewis, Curtis Martin, Shaun Alexander, LaDainian Tomlinson, LaDainian Tomlinson, Adrian Peterson, Chris Johnson, Arian Foster. I also have all the second-ranked rushers, and third-ranked, and fourth, and so on down to tenth, for every year from 1991-2010. Some of the more interesting findings:
* The leading rushers were the shortest (70.2") and lightest (215.2 lbs) of all 10 groups.
* The 10th-ranked rushers were the second-tallest (71.5"), second-heaviest (223.9), and had the second-highest BMIs (30.9).
* The leading TD scorers were the shortest (70.2") by a large margin, almost half an inch.
* The tallest group was the ninth-ranked TD scorers (71.8").
* The second-leading scorers were the heaviest (224.2) and had the highest BMI (31.4).
* The first-down leaders were the shortest (70.5"), followed by the third-leading (70.6") and second-leading (70.7").
* The 10th-ranked group for first downs was the heaviest (225.0).
The stat that most obviously correlates with size is first down percentage. The smaller running backs consistently and dramatically outperformed the larger in this category:
* The three shortest groups were the top three in this category. The leaders checked in at 70.3", followed by 70.7" and 70.4".
* Four of the top five groups in 1stD% — 1st, 2nd, 3rd, and 5th — weighed an average of less than 220 lbs., and four of the bottom five weighed over 220 lbs. The lightest group was the 2nd-ranked in first down percentage.
These data confirm the height and weight breakdowns presented already: height is a disadvantage, and weight except to the extent it correlates with BMI. Not only do smaller RBs gain more yards overall and per carry, but they gain more first downs and score more TDs. Short, powerful runners with a medium-to-high BMI tend to be the most successful RBs in goal-line and short-yardage situations. You probably don't want Dexter McCluster (5-8, 170) carrying the ball on 3rd-and-1, but you do want Jones-Drew (5-7, 208) or Rashard Mendenhall (5-10, 225) or LT (5-10, 215).
Shorter players fumble less often than tall players, and lighter-weight RBs fumble less often than heavier RBs.
* The top-10 rushers with the fewest fumbles each season were the shortest (70.3") and by far the lightest (210.6 lbs) of all 10 groups.
* The 10th-ranked group (most fumbles) was the second-tallest (71.5"), second-heaviest (225.4), and had the second-highest BMIs (31.0).
* In the height study, the shortest group (under 5'9") fumbled far less than the other groups, 2.1 per season. The tallest group (over 6'1") fumbled second-least (2.9), but the other tall groups fumbled the most.
* In the weight study, the lightest group (205 and less) fumbled the least, 2.6 per season, followed by the second-lightest group (206-215), at 2.8 per season. The other three groups were roughly equal, and almost a full fumble higher per season (3.7, 3.8, and 3.4, in ascending order of weight).
* There is no obvious correlation between BMI and fumbling. The players with BMIs between 30-32 fumbled the least (2.8), players above 32 the most (3.8).
The top 10 in rushing TDs of 1-2 yards, since 1991:
Hold the presses! This doesn't look like 5'9" is an ideal height for RBs. Nine of the 10 players are taller than that!
That's because most RBs are taller than 5'9". Only 33 of the 200 seasons included were by a running back that short — and half of those 33 seasons are just Emmitt and Barry. Players that short seldom even get a chance in the NFL, or they're forced into part-time roles and leave the game in goal-to-go situations. A team that has a choice uses Tons of Fun in that situation.
When short RBs get a chance, however, they prove themselves over and over. DeAngelo Williams is about 5-8½, and he plays with a stereotypically "big back" who gets most of the goal-line work, but he led the NFL in TDs a couple years ago. Priest Holmes (5'9") set a single-season TD record. Maurice Jones-Drew is only 5'7", but he's scored double-digit TDs in four of his first five seasons, including 18 runs of 1-2 yards. Smith, of course, is the all-time leader in rushing TDs, one of the greatest short-yardage runners of all time.
The average size for the leaders in 1-2 yard TDs correspond almost exactly to the averages of all top-10 rushers: 71 inches tall (5'11"), 219.4 pounds, 30.6 BMI. The difference is less than a pound.
The Ideal Size
On average, the most successful short-yardage RBs over the past 20 seasons were about 5'9", 215 lbs., and BMI 31.5. I would suggest that a model short-yardage runner is 69" tall and weighs 210-215 pounds, indicating a BMI between 31-31.75. This very nearly matches Emmitt Smith, LaDainian Tomlinson, Priest Holmes, Earnest Byner, Frank Gore, Barry Foster, DeAngelo Williams, and Steve Slaton.
Conversely, running backs taller than 6'1" appear to be at a disadvantage and probably should not be regularly used in important short-yardage situations if the team has another good RB. This includes Eddie George, Steven Jackson, Chris Warren, Robert Smith, Leonard Russell, Matt Forte, Harvey Williams, and Harold Green.
Domination of the game (and the goal line in particular) by big fat guys has noticeably diminished over the past several decades. I would suggest that the primary reason for this is the explosion of size among linemen. Remember when Refrigerator Perry was a punch-line, a guy so fat it was comical? Most teams now have half a dozen guys that big. It used to be that a John Riggins or Earl Campbell could get a head of steam and bowl over linemen who didn't weigh much more than he did. But now, send a 240-lb. RB crashing into a wall of 320-lb. refrigerators, and he's not going to get past them.
I would posit that the performance of the offensive line is probably the most important factor in straight-ahead rushing attempts, and that even a big, strong RB is very limited in his capacity to create opportunities without a strong push from his blockers. A faster RB who works outside, or presents the threat of an outside run, may be less handcuffed by his blocking opportunities, and presents more threats to a defense. Send in the tank, and the defense knows it doesn't have to defend against an outside run. A more nimble back forces the defenders to spread out.
Running backs can't create holes if the blockers in front don't make them. The most successful short-yardage runners in today's game threaten as much with their agility and balance as their power, and probably more than anything with their vision and intelligence. A good running back will do well on the goal line, too, regardless of size (though it's probably best if he's not too tall). An increasingly popular trend involves increasing the size of the blocking back rather than the ball-handler, for instance by using a 310-lb. defensive tackle at fullback. I'm not sure how much difference that makes, but it's a better idea than replacing your best RB with a big slow guy just when you really need a good play.
It probably was true at one time that larger backs were more effective at picking up 3rd-and-1, but that was a long time ago. Like, 30 years, son. If you still believe, in 2011, that your best bet on the goal line is the 240-pounder who snuck out of the locker room for a hot dog at halftime, you probably also believe 1,000-yard rushing seasons are really groovy. Half the starters in the league make it to 1,000 these days, big guy. Remember when a Hall of Fame defensive tackle might weigh less than a running back? Alan Page, who retired in 1978, finished his career at 225 lbs. Yeah, that doesn't happen anymore.
The best goal-line runner is your best running back, period. He doesn't have to be huge, and it might be preferable that he isn't.
That's basically it, but if you're kind of a stats nerd, you might want to keep reading about how the numbers presented above might be misleading. If that sounds boring, you can quit while you're ahead and probably won't miss anything important. Thanks for stopping by.
The numbers don't lie, but that doesn't mean they necessarily tell us what they appear to tell us. The four most important potential biases that seem likely to me are: (1) bigger RBs get more short-yardage opportunities than smaller RBs, (2) a player's weight can vary over his career, (3) bad teams are more likely to use unusually big or small RBs than good teams, and (4) possible issues with sample size.
Bigger RBs get more short-yardage opportunities than smaller RBs.
Conventional wisdom dictates that a big, heavy back (like Bettis or George) is best-suited to short-yardage situations, and that a lighter back (like Terry Allen or Terrell Davis) is not a good choice for those situations. Of course, Allen (208 lbs) and Davis (210) and Faulk (211) and Holmes (213) and Tomlinson (215) and Barry Sanders (200) all led the NFL in TDs, which Bettis (255) and George (235) and Jamal Lewis (245) and Michael Turner (244) and Larry Johnson (235) and Deuce McAllister (232) never did.
I believe the disadvantage of heavier RBs is even greater than it appears from this data, because they get many more goal-line opportunities than lighter RBs. Tiki Barber, listed at 205 lbs., was a very good short-yardage runner, but the Giants routinely subbed him out for Ron Dayne (245) and Brandon Jacobs (264). I remember thinking at the time that it was crazy. Dayne was a disaster, and Jacobs, especially early in his career, wasn't much better. But the Giants regularly removed their best running back from the game in the most crucial situations. Imagine if the Yankees pulled Mariano Rivera before the final out of a game because the next batter was left-handed. It's the same thing.
Barber scored more rushing TDs of over 10 yards (20) than rushing TDs of 1-2 yards (18), because the Giants rarely gave him the ball in short yardage. Some of you may remember that Barber had a fumbling problem, and wonder if the Giants kept him out of goal-line situations because they were afraid he'd pull an Earnest Byner. Only 32 of Barber's 53 career fumbles came on rush attempts. The others were receptions or punt returns, where fumbles are more common. Barber fumbled less often per carry (1.44%) than Jacobs (1.51%), and in 2005, he actually had the fewest fumbles (1) of any top-10 rusher.
Barber is just one example. In 2005, splitting time with Bettis, Willie Parker scored TDs of 10, 11, 37, and 80 yards. The next season, with Bettis retired, he scored from 1, 1, 1, 1, 3, 3, 3, 3, 3, 4, 8, 9, and 41. Obviously, he could handle short-yardage, but he lost opportunities because he played with a guy who outweighed him by 45 pounds. The same thing happened to Chris Johnson, who in his 2,000-yard season regularly left the game so LenDale White could stumble into a wall of defenders. Guys like Warrick Dunn and Ray Rice and Jamaal Charles don't even get a chance to prove themselves in short-yardage situations. Their TD statistics are poor not because they were bad at picking up 3rd-and-1 or 4th-and-goal, but because they were never in the game on 3rd-and-1 or 4th-and-goal. Thus, the lightest backs are probably even better at short-yardage than their stats suggest, and the big guys probably a little worse.
A player's weight can vary over his career.
As a rookie, Clinton Portis weighed about 190 lbs. He rushed for 1,500 yards, with a 5.5 average and 15 TDs. The next season, almost 1,600 yards, 5.5 average, 14 TDs. Then he got traded to Washington, and bulked up to 220 in an effort to handle increased workload. For what it's worth, all his career highs were set in those two seasons, but that's not the point — on every line of this study, Portis is 5'11", 219 lbs. That means that in '02 and '03, his weight listing is off by almost 30 lbs., and that the results are not 100% accurate.
Similarly, Emmitt Smith is listed at 216. He put on weight at the end of his career, but was below 210 in the '90s. In this study, Jerome Bettis is 255 lbs., because that's what his NFL.com page shows, but in his prime, Bettis barely weighed 240. Late in their careers, many players compensate for lost speed with extra size and strength, so again, most RBs are probably listed as weighing more than they really did for the majority of their careers. Some are even listed at different heights. Fred Taylor, 6' in 1998, was later listed at 6'1".
Bad teams are more likely to use unusually big or small RBs than good teams.
Most teams, if they have a choice, use a running back close to average size: somewhere between 5'9" and 6'1", 205-235 lbs. Players who fall outside that range are most likely to play only if the team is desperate: if there are no other good running backs on the roster, or if injuries force him into action. That's how Peyton Hillis, originally a fullback, became a fantasy football sensation for the Browns in 2010. Players who are too short, too tall, too heavy, too light, usually begin their careers as part-time players: special teamers, third-down backs, goal-line specialists, fullbacks. If the team already has an Emmitt Smith or Marshall Faulk, they might not carry the ball at all.
Bad teams sometimes have no choice but to put these players in the lineup, and often, they pleasantly surprise. But a bad team with a good running back is still a bad team. It won't score very many touchdowns, and it especially won't score many rushing touchdowns. It won't get a lot of first downs, and it will have a poor first down percentage. Thus, the size outliers — the very big and the very small — might be better than their low TD and first down numbers suggest.
Possible issues with sample size.
In any study like this, you have to balance precision with certainty. A small group that fit a certain height or weight range, maybe as few as 15 or 20 players, may not be terribly meaningful. For instance, the Under-5'9" group in the height section included only 14 seasons, and eight of them were by Barry Sanders. Does this tell us anything meaningful about players besides Sanders?
On the other hand, a larger group — say 100 seasons — eliminates that uncertainty, but with only two groups, you're lumping together as "similar" vastly different players. We could split weight at 215 lbs., yielding groups of 87 and 113: nice, big samples. But now we're putting Emmitt Smith and Clinton Portis in the same category as Jerome Bettis and Christian Okoye. That's just not a valid comparison. When we talk about big backs, we don't mean guys like Smith and Portis.
Two hundred seasons are a lot, and I think there's a decent balance in most of the data sets presented here, but it's not perfect.
Some RBs appear on the top-10 leaderboard more than once. Here are all who posted at least three appearances, with height, weight, BMI, and number of top-10 seasons since '91. They're organized by ascending weight, because alphabetical order was boring and I felt like it.
Eight Top-10 Seasons
Barry Sanders: 5'8", 200 lbs, 30.4 BMI
LaDainian Tomlinson: 5'10", 215 lbs, 30.8 BMI
Emmitt Smith: 5'9", 216 lbs, 31.9 BMI
Seven Top-10 Seasons
Curtis Martin: 5'11", 210 lbs, 29.3 BMI
Six Top-10 Seasons
Clinton Portis: 5'11", 219 lbs, 30.5 BMI
Corey Dillon: 6'1", 225 lbs, 29.7 BMI
Fred Taylor: 6'1", 228 lbs, 30.1 BMI
Five Top-10 Seasons
Marshall Faulk: 5'10", 211 lbs, 30.3 BMI
Ricky Watters: 6'1", 211 lbs, 27.8 BMI
Edgerrin James: 6'0, 219 lbs, 29.7 BMI
Rodney Hampton: 5'11", 228 lbs, 31.8 BMI
Eddie George: 6'3", 235 lbs, 29.4 BMI
Jerome Bettis: 5'11", 255 lbs, 35.6 BMI
Four Top-10 Seasons
Tiki Barber: 5'10", 205 lbs, 29.4 BMI
Thurman Thomas: 5'10", 206 lbs, 29.6 BMI
Terry Allen: 5'11", 208 lbs, 29.0 BMI
Terrell Davis: 5'11", 210 lbs, 29.3 BMI
Thomas Jones: 5'10", 212 lbs, 30.4 BMI
Adrian Peterson: 6'1", 217 lbs, 28.6 BMI
Shaun Alexander: 5'11", 228 lbs, 31.8 BMI
Ricky Williams: 5'10", 230 lbs, 33.0 BMI
Stephen Davis: 6'0, 230 lbs, 31.2 BMI
Jamal Lewis: 5'11", 235 lbs, 34.2 BMI
Three Top-10 Seasons
Chris Johnson: 5'11", 191 lbs, 26.6 BMI
Priest Holmes: 5'9", 213 lbs, 31.5 BMI
Chris Warren: 6'2", 227 lbs, 29.1 BMI
Steven Jackson: 6'2", 236 lbs, 30.3 BMI
Rudi Johnson: 5'10", 255 lbs, 32.3 BMI