Writers of Pro Football Prospectus 2008

25 Feb 2009

Speed Score 2009: Is This It?

Right around the time you see offensive linemen running 40 yards downfield for the first and last times of their NFL careers, it's reasonable to wonder whether the NFL scouting combine is really just a big waste of time -- an excuse for NFL teams to spend hours trying to figure out whether a player's bubble (read: posterior) is too big or his hands are too small, or that he doesn't run patterns well against a set of orange cones.

So then, is the combine worthless? Is it safe to throw out all the data and just look at how a player performed on Saturdays against inferior competition? No one has bothered to actually go back and check whether what happens at the combine bears any relationship to NFL performance -- besides us, that is.

The answer is that, well, it depends on the position and the player. One place where a bit of combine data can actually go a long way in predicting a player's viability in the NFL is at running back, where we've come up with a metric known as "speed score."

Speed score (explained in detail in PFP 2008 and in this 2008 article) takes into account each player's 40 time and weight to produce a number scaled around 100; the average speed score for a drafted back is 102.4, a number which rises to 111.1 for backs taken in the first round. The formula -- (Weight x 200)/(40-yard^4) -- adjusts the minuscule differences in 40 times from player-to-player by accounting for the weight each player has to lug around on his 40-yard dash. The result is a metric that has a stronger correlation to NFL performance on a one-year, three-year or five-year stretch than any other combine drill, including the standard 40-yard dash.

Last year, speed score pegged Chris Johnson (121.9) as the best back in the class, with Darren McFadden (120.0) and Jonathan Stewart (116.7) shortly behind. It predicted Matt Forte' (109.7) to be a sleeper, while believing that Ray Rice (99.8), Kevin Smith (98.6) and Steve Slaton (96.9) would struggle. Speed score is certainly not a foolproof indicator, but as you can see from that level of performance, it can be a useful one.

One really interesting case in upcoming seasons will be that of Felix Jones, who had a speed score of 103.7 despite being regarded as one of the fastest players in college football. His 4.47 40-yard dash certainly wasn't an impressive number (compared to the Chris Johnson's and Darren McFadden's of the world), but Jones showed off elite speed and blazing athleticism as a change of pace back and return guy in Dallas this year. Jones led the league in DVOA for backs with 20-99 carries, averaging a ridiculous 8.9 yards per attempt.

On the other hand, Jones only got 30 carries because he spent most of the year sidelined by a torn hamstring muscle. It's impossible to say if Jones is likely to be an injury-prone player at the professional level, but the possibility also exists that speed score does reveal something about a player's likelihood of staying healthy enough to accrue yardage as a professional; perhaps, if a player isn't fast enough or a good enough athlete to avoid taking a certain amount of hits at certain angles, he's more prone to suffering injuries, and speed score captures that. I'm not sure whether I believe that's the case or not, more just me thinking aloud.

The possibility always exists that Jones could just end up being a bust, anyway. The list of guys who averaged more than seven yards a carry in 20 or more attempts is: DeDe Dorsey, Jerry Ellison, Lorenzo Neal, Thomas Sanders, Ahmad Bradshaw, Tony Nathan, Darrin Nelson, Lamont Jordan, Brian Mitchell, Davi Meggett, Rock Cartwright, Sylvester Stamps, Michael Wiley (who did it two years in a row, was cut by the Cowboys, and never played in the NFL again), and Chris Fuamatu-Ma'afala. If we changed the somewhat arbitrary cutoff to 6.5 yards per carry, we could finally throw in some famous names: Kimble Anders, Bo Jackson, and Mercury Morris. What that all tells us? 30 carries is a pretty small sample size.

2009 Running Backs

Speed Score For 2009 Running Backs
Player School 40 Time Weight Speed Score
Andre Brown North Carolina State 4.49 224 110.2
Cedric Peerman Virginia 4.45 216 110.2
Ian Johnson Boise State 4.46 212 107.2
Javarris Williams Tennessee State 4.52 223 106.9
Beanie Wells Ohio State 4.59 235 105.9
Kory Sheets Purdue 4.47 208 104.2
Donald Brown Connecticut 4.51 210 101.5
Rashad Jennings Liberty 4.64 231 99.7
Shonn Greene Iowa 4.63 227 98.8
Mike Goodson Texas A&M 4.54 208 97.9
Chris Ogbonnaya Texas 4.61 220 97.4
Marlon Lucky Nebraska 4.59 216 97.3
Knowshon Moreno Georgia 4.60 217 96.9
James Davis Clemson 4.61 218 96.5
Glen Coffee Alabama 4.58 209 95.0
Jeremiah Johnson Oregon 4.61 209 92.5
Bernard Scott Abilene Christian 4.56 200 92.5
Anthony Kimble Stanford 4.66 216 91.6
Javon Ringer Michigan State 4.60 205 91.6
Branden Ore West Liberty State 4.67 214 90.0
Tyrell Sutton Northwestern 4.66 211 89.5
Gartrell Johnson Colorado State 4.71 219 89.0
Kahlil Bell UCLA 4.68 212 88.4
Note: Some times are unofficial

The 2009 crop of running backs isn't as highly regarded as last year's, a group that produced five first-round picks. That's borne out by their speed scores. Knowshon Moreno, regarded as the draft's top back, ran a disastrous 4.6 40-yard dash that yielded a speed score of only 96.9. Even if you go with the time of 4.55 that has also been unofficially reported for Moreno, his speed score would be only 101.3, putting him just below Chris Perry (102.7).

Going back to 1999, that would be the lowest speed score posted by a first-round pick; the only two backs selected in the first round to post a speed score under 100 are William Green (98.7) and Trung Canidate (99.3). Only one back in the 11 seasons we've got speed score data for made it to the Pro Bowl after posting a speed score below 98.0: Brian Westbrook.

In his defense, Moreno's regarded as having elite agility, which goes unmeasured in the 40. Agility is measured in other drills, though, so if Moreno's agility was really at an elite level, we'd expect to see as such in the three-cone drill and the two shuttle runs.

In the three-cone drill, Moreno's 6.84 seconds were second to Abilene Christian back Bernard Scott. Scott also topped the leaderboard in the 20-yard shuttle with a time of 4.08 seconds, while Moreno was eighth at 4.27 seconds. (In the 60-yard shuttle, which we don't track data for, Moreno finished fourth out of the six who attempted it.) Over the past ten years, the average back who's been drafted has been 5'10" and weighed 216 pounds -- almost a mirror image of Moreno's 5'11", 217-pound frame. Those same backs have averaged a 20-yard shuttle time of 4.20 seconds and a three-cone drill time of 7.07 seconds. While Moreno's three-cone drill score was better than average (and would rate as the fourth-best time for drafted backs), success in the three-cone drill actually bears a slightly inverse correlation to NFL success, while the shuttle, which Moreno was below-average in, has a much more positive relationship.

While Beanie Wells' 4.59 40-yard dash almost perfectly mirrored Moreno's, the fact that he did so with 18 extra pounds on his frame produces a speed score of 105.9 (below-average for a first-rounder, but passable for a day-one pick). He actually profiles as rather similar to another Big Ten back: Larry Johnson, who was 228 pounds and ran a 4.55 40 at the 2003 combine, yielding a speed score of 106.4. Unfortunately, Wells doesn't come with the 2006 Chiefs offensive line.

The two players who improved their stock the most are the two ACC products who sit atop the speed score leaderboard. North Carolina State's Andre Brown ran a 4.49 40 at 224 pounds; that compares comfortably with Tampa Bay's Earnest Graham, who ran a 4.50 40 at 223 pounds in 2003. The book on Brown is that of a versatile, injury-prone back without the speed necessary at the NFL level; speed score thinks he can be a useful player in the pros.

The fastest 40 time of the day for running backs belonged to the other riser, Virginia's Cedric Peerman, a back similar to Brown in both stature and proneness to injury. While Brown projects to be a mid-round pick, though, Peerman (before Sunday) rated out as practice squad fodder.

The reason why? Scouts fear that the 7 7/8-inch hands of Peerman will yield too many fumbles as a professional, despite the fact that he fumbled all of four times on 448 touches in college. Maybe the combine is a stupid idea, after all.

A final factor we don't know anything about yet is whether the literal change in playing field has affected what we should expect from players in the combine. This is the first year for the combine in Lucas Oil Stadium as opposed to the now-imploded RCA Dome, and without a real blowaway 40 time from any of the players at this year's event, there's been some chatter that the surface might not be of the same caliber.

If that's the case, well, the numbers sure don't show it. The average 40 time for this year's running backs was 4.58; the average 40 time for running backs from 1999-2008 was 4.57. At wide receiver, the average 40-yard dash this year was 4.51, actually three-hundredths of a second faster than the 1999-2008 average of 4.54. If the new field's slowing guys down, it's not showing up in their numbers.

Posted by: Bill Barnwell on 25 Feb 2009

73 comments, Last at 04 Aug 2009, 4:15pm by Rexy

Comments

1
by Lou :: Wed, 02/25/2009 - 4:20am

I really like speed score as a stat. is there any plan to see if it works for other positions? or to see if other combine stats correlate to success at other positions?

i do wonder whether its the speed score that leads to greater success or if a high speed score just indicates an athlete willing to dedicate more of his time than his peers to getting the absolute highest production out of his body. and that that dedication is what translates to the nfl (on the practice field, in the classroom, and obviously in the weight room) and not necessarily the slight difference in physical ability.

23
by KyleW :: Wed, 02/25/2009 - 3:27pm

I think your analysis could be quite accurate, combined with the momentum one below, could be the reasons for why speed scores have a correlation with NFL success. We know that all the combine prospects work out especially in order to post good scores in these drills and the speed score may simply capture those who work out the hardest in preparation and then are likely to be able to follow this through to their NFL career.

What is the measure used to judge NFL success for the running backs though - and is/should it be adjusted to reflect the talent of those blocking for them?

2
by Joe T. (not verified) :: Wed, 02/25/2009 - 8:37am

So, by the speed scores, it looks like there are no "must-have" backs this year like last year, but there do look to be some valuable backs who should fall to later rounds.

So you've established a correlation between speed score and success, what is the reason for the correlation? How does this attribute potentially manifest itself on the field in a way that leads to success?

Interesting article, looking forward to the draft.

12
by The Guy You Don't Want to Hear (not verified) :: Wed, 02/25/2009 - 12:24pm

For what it's worth, I really think the correlation is a result of speed score being closely related to the formula for momentum (mass x velocity). A bigger back at the same speed as a smaller one has more momentum and is thus more difficult to tackle, all else being equal, and simultaneously has a higher speed score. A faster back is also going to be more difficult to tackle than one of the same size who is slower due to the same increased momentum and again his speed score would also be higher.

It's been a while since I had any physics courses, but as I recall momentum is defined as resistance to a change in course or velocity. Since NFL running backs are rarely running unabated, the ability to continue forward in spite of being hit is more important than simple speed and as a result, momentum is very important. I think the speed score essentially measures the runners' general running momentum, which is an important skill. It obviously does not measure agility, vision, blocking ability, or ability to read defenses and, as such, is only a part of the picture, but is a very good measure of one very important skill.

However, to test whether it's really measuring momentum like I think it is, all I can come up with is to try to find backs whose careers have been based on those other skills (Brian Westbrook actually could be a good example) and players at the other end of the spectrum who lack those other skills (the one who comes to mind is Michael Turner, but I have admittedly not seen him enough to fairly judge) but "seem," for lack of a better term, to have good momentum and then compare their speed scores. That just seems like it would be such a subjective, weak categorization (high-momentum vs. low-momentum backs) that it wouldn't work, especially since we only have 11 years of speed scores and thus would have a hard time picking out more extreme, obvious examples.

18
by Jacob Stevens (not verified) :: Wed, 02/25/2009 - 2:08pm

Speaking of momentum, I have a sneaking suspicion that "slower" tracks, like Lucas Oil allegedly (but clearly many tracks are going to be slower than the RCA Dome) affect otherwise faster players more than slower players. To put it another way, the slower the track, the more it's an equilizer, whereas faster tracks like the RCA Dome, and those in Minnesota and St. Louis provide more separation between the fastest players and the average players.

I think slower and bigger players move their body around the field driven by less momentum, and more pure force, at least on the football field where there's few opportunities to really get it going, particularly in a single direction.

I remember Dr. Z commenting on how slow the Superdome field was, after Katrina, and how their slow corners McKenzie and Thomas were able to keep up with receivers better at home. If the theory hods true, it'd mean they'd get torched in faster fields on the road, but hold up better up a slow track at home. And for this draft, some of the faster players might indeed be getting hurt by Lucas Oil Stadium, and play much faster elsewhere, but some guys who ought to be whittled away on account of their slowness will come across as average, until gameday comes and they're molasses.

3
by BlueStarDude :: Wed, 02/25/2009 - 8:44am

RE: thinking out loud --

- I don't think Felix Jones's injuries had anything to do with contact (his second, which put him out for the season, certainly didn't; don't recall about the initial injury)

- Emmitt Smith probably had a terrible speed score

- It's funny how the two prospects with the most injury concerns going into the draft finish one-two in speed score. That doesn't correlate well with the speed score / injury thoery re: Felix Jones, does it? I think Brown will be good though.

- I'll be shocked if Knowshon Moreno isn't the best RB out of this years crop. Unless it turns out to be LeSean McCoy.

- I thought Larry Johnson was undervalued a bit the year he came out, I'd say Wells is overvalued.

- RE: Peerman's small hands: I see how that could affect receiving ability, but I don't get the fumble concerns? All you need really is to cover the nose of the ball, then secure tight to forearm, bicep, and chest. If the RB has sound technique then 7" hands shouldn't be an issue, unless I'm missing something.

- The players have rec'd so much training these days that the 40 time is even less relevant to the football field than it was ten years ago. Maybe they should have the guys run the 40 from a standing start (like a WR on the line of scrimmage), yell go, and time them that way.

20
by Vincent Verhei :: Wed, 02/25/2009 - 2:34pm

- Emmitt Smith probably had a terrible speed score

A 30-second Google search puts Emmitt's 40 time at 4.6 or 4.7, and his weight at 210 to 216 pounds. That's a speed score of 86.1 to 96.5. So yes, he was likely an anomaly.

24
by Bobman :: Wed, 02/25/2009 - 3:31pm

And that, friends, is how Emmitt Smith anomalized the NFL defenses and demathematized the front-office know-it-mostlies. His intangibles hypothesize the measurements caluclated by the speed score and disincentivize the weight metric that doesn't take into accouint the correlation of his agility and vision and durabilty.

Hence: anomalization personifized.

34
by Bill Barnwell :: Wed, 02/25/2009 - 5:52pm

I expect that Emmitt was an anomaly, but it's also worth noting that I have nothing to compare Emmitt to. I can't say for sure that the baseline for a successful back would be similar, because I don't have Speed Scores for any other late-80's guys.

44
by BlueStarDude :: Wed, 02/25/2009 - 10:11pm

Hence the "probably" -- there's no point to the 30-second Google search.

39
by armchair journe... :: Wed, 02/25/2009 - 7:44pm

Emmitt debacled the Speed Score.

_______________________________
armchair journeyman quarterback

4
by Dr. Mooch (not verified) :: Wed, 02/25/2009 - 9:30am

Seriously, you guys, why don't you use a number that's scaled closer to 100?

Using nearly the same damn formula in metric gives you a 99.99 for Peerman.

Last year Johnson would have scored a more reasonable 110, and McFadden 108.9

(Weight*400)/(40-time^4)

16
by Bill Barnwell :: Wed, 02/25/2009 - 1:17pm

Because *200 correlates much better to NFL performance. And the formula you suggested doesn't yield the results you suggested -- Peerman would have like a 220.

22
by An Ominous (not verified) :: Wed, 02/25/2009 - 3:02pm

How could *200 correlate better than *400? All you're doing is changing the constant the variables are being multiplied by, without changing the variables in the slightest. I mean, the *200 was, by the article's own admission, merely a factor designed to scale speed scores somewhat roughly around 100. If another factor accomplished it more readily, why not switch to that factor?

Also, remember that he said to use metric. I'm assuming he means (weight IN KILOGRAMS * 400)/(40-yard^4).

To the original poster- I suspect the reason they don't switch to metric is that all combine weights are given in POUNDS. And the reason they stick with 200 is because the goal of Speed Score is supposed to be quick and easy. I'm sure they could come up with a way to better center the results around 100, but then the speed score formula might wind up looking more like the infamous "passer rating formula", and I don't think anyone really wants that. For a quick and easy measure, speed score works fine.

46
by DeltaWhiskey :: Thu, 02/26/2009 - 10:06am

Many statisticians utilize a concept known as "Standardized" scores.

5
by MCS :: Wed, 02/25/2009 - 9:47am

Speed Score=(Weight x 200)/(40-yard^4)

You say, “…if a player isn't fast enough or a good enough athlete to avoid taking a certain amount of hits at certain angles, he's more prone to suffering injuries, and speed score captures that.”

How does Speed Score capture the player’s ability to take a hit and remain injury free? Maybe the case you used as an example (Felix Jones) is simply coincidence?

6
by Travis :: Wed, 02/25/2009 - 10:28am

Last year, speed score pegged Chris Johnson (121.9) as the best back in the class, with Darren McFadden (120.0) and Jonathan Stewart (116.7) shortly behind. It predicted Matt Forte' (109.7) to be a sleeper, while believing that Ray Rice (99.8), Kevin Smith (98.6) and Steve Slaton (96.9) would struggle. Speed score is certainly not a foolproof indicator, but as you can see from that level of performance, it can be a useful one.

Am I missing something, or is there not that much of a correlation between the speed scores of those 7 players and their performances last season?

7
by Yaguar :: Wed, 02/25/2009 - 10:48am

Johnson and Forte look the best, and Rice and Smith look the worst out of that group to me. It missed on Slaton, but the rest seems about right. McFadden and Stewart underperformed their speed score, but they do seem to be talented and capable of significant improvement. In McFadden's case, only injury really stopped him. He looked pretty good.

8
by Travis :: Wed, 02/25/2009 - 11:09am

I'd agree that Johnson (subjectively) looked like the best of the bunch. I'm not sure how to evaluate Smith, given that he was on one of the worst teams of all-time, but by FO's stats he was at least no worse than Forte.

A comparison in (crappy) table format:

Player ........... Score ..... DYAR ..... DVOA ..... Att-Yd ........ EYds ..... Teammate with 50+ att
C.Johnson ..... 121.9 ...... 170 ........ 9.2% ..... 251-1228 ..... 1122 ..... L.White: 7.5% DVOA, 200 carries
D.McFadden .. 120.0 ...... -14 ....... -9.1% ..... 113-500 ........ 365 ..... J.Fargas: -13.8% DVOA, 219 carries
J.Stewart ....... 116.7 ...... 111 ....... 7.8% ...... 184-838 ........ 826 ..... D.Williams: 28.5% DVOA, 273 carries
M.Forte ......... 109.7 ....... 38 ....... -7.0% ..... 316-1239 ..... 1233
R.Rice ............ 99.8 ........ 20 ........ -2.4% ..... 108-462 ........ 381 ..... L.McClain: 7.6% DVOA, 232 carries
K.Smith .......... 98.6 ........ 88 ........ -0.3% ..... 238-975 ....... 984 ..... R.Johnson: -21.3% DVOA, 76 carries
S.Slaton ......... 96.9 ........ 170 ........ 5.8% ..... 268-1285 ..... 1269 ..... A.Green: 12.4% DVOA, 74 carries

69
by phildo (not verified) :: Fri, 03/06/2009 - 12:11am

you didn't include receiving, which males forte appear much worse than he really was. he was second among rbs in rush dyar.

9
by fire SNAKE42 (not verified) :: Wed, 02/25/2009 - 11:42am

Looks quite hard to believe. If it works, would be one of the least comprehensible stats ever to me. I understand that you guys DO put a lot of validation in your stats ...

I mean, if you put a RB with a uber-Speedscore in a team with no OL, no blocking and no passing game, what do you expect in return? I would expect at least SOME correlation here (since I think the RB position is least based on individual skill and mostly on team effort). Did DMC really look so good last season?

Any other stats with regard to how "bigger" (does that correlate to "slower"?) backs perform over time? So do speedier backs show a bigger decline than slower ones? Are bigger backs more "durable" and a better "long term" investment?

11
by Sophandros (not verified) :: Wed, 02/25/2009 - 12:23pm

That's why they mentioned that, for example, Beanie Wells doesn't come with the 2006 KC offensive line...

70
by phildo (not verified) :: Fri, 03/06/2009 - 12:15am

so you don't see why a stat that measures combines size and speed could be a useful way to evaluate rbs? wtf?

it's like you've taken the standard sportscenter mouthbreather's opinion (lol wat's an o-line?) and gone way too far the other way with it (rbs don't matter at all! only the o-line is important). there's a happy middle ground there which is likely closer to the truth.

71
by phildo (not verified) :: Fri, 03/06/2009 - 12:16am

*combined

10
by vherub :: Wed, 02/25/2009 - 12:19pm

wouldn't the nfl need to rescore veteran players with 1-5 years experience to get a better sense of whether the numbers mean anything? Surely the bodies coming out of college look and play different than the bodies after a few years in the league- some positions more than others. When would speed-score peak, or arm strength or agility? And what would be the plotted progression both on the up and the down?

14
by Bowl Game Anomaly :: Wed, 02/25/2009 - 12:59pm

While that's an interesting question, it's not relevant to the issue of projecting the performance of potential draftees. Draft-year speed score is already correlated to NFL performance. If we start looking at NFL veteran speed scores, we won't learn anything new about projecting rookies. We already have good ways of measuring NFL veterans, namely DYAR and DVOA. Futher, veterans don't run 40-yard dashes, so there's no way to gather that information.

Remember also that this formula isn't based on a theory, it's based on a statistical result. The theory behind it comes second.

(Formerly "The McNabb Bowl Game Anomaly")

53
by MC2 :: Thu, 02/26/2009 - 2:36pm

Remember also that this formula isn't based on a theory, it's based on a statistical result. The theory behind it comes second.

This is part of my problem with it. Correlation is obviously not always indicative of any real, intrinsic relationship. For example, the article mentions that Moreno excelled in the 3-cone drill, which has a negative relationship with NFL success, but it offers no reason why the relationship is negative. Is there any reason whatsoever that the ability to perform well in the 3-cone drill would hamper a player's performance in the NFL? Isn't it much more likely that this is a statistical fluke? And, assuming that it is, how do we know the positive correlation between speed score and NFL success isn't also a fluke?

Speaking of Moreno, I think there's a typo in the article. Moreno is said to have run the shuttle in 4.08 seconds, while the average in the shuttle is 4.20 seconds. However, Moreno's shuttle time is later described as "below average".

By the way, the point of this post is not to say that speed score is necessarily useless. I would just caution against falling victim to the natural human tendency to recognize patterns, and then attempt to explain them with (often counterintuitive) post hoc theories, when often the better explanation is just random variance, giving the appearance of a "pattern".

59
by Bowl Game Anomaly :: Thu, 02/26/2009 - 10:17pm

I think, though I am not certain, that speed scores are not just based on eyeballed correlation. IIRC, Barnwell examined many combine stats and found that none of them correlated significantly with NFL performance, except for 40-yard time for RBs when weight was taken into account. That result, which became the speed score, was correlated with NFL success to a statistically significant degree.

What that means is that there is a real relationship, and it can be used predictively even though it does not establish any causation. (Though of course it will not be 100% accurate because the correlation is not a perfect 1.0.)

(Formerly "The McNabb Bowl Game Anomaly")

13
by Dales :: Wed, 02/25/2009 - 12:33pm

Regarding Jones, one possibility is that he just had an off day at the combine. Especially since the exponent is 4, that would have a fairly large effect on his score.

17
by Jimmy :: Wed, 02/25/2009 - 1:19pm

From memory Devin Hester ran a 4.43 or similar at the combine, but has said that he did once time in a 4.25 or so. From stuff I have read about it he wan't too bothered about putting up a really low time as he reckoned that he had enough stuff on film to make GMs think he was one of the fastest players at the combine. Having watched Hester play for three years I can't think of any players in the NFL who are definitely faster than he is (have you ever seen it look even remotely likely that he could be caught from behind?), yet there are probably loads with faster 40 yard times.

If you think about it one wasted step at the start or a moment's hesitation could easily cost you a quarter of a second, moving a 4.3 to a 4.5. I know they run these things twice, but it could still produce false results. Everyone and their mothers agree that you should really do your scouting on film - with the possible exception of Mike Mayock who seems to be infatuated with the fact that they are testing the players, and reads way too much into the intangibles surrounding the combine (what happened to the intangibles involved in the actual football?).

15
by bubqr :: Wed, 02/25/2009 - 1:12pm

I have a speed score of 84,94, but i'm pretty sure i could rush for 1000 yards easily if i were drafted by the Giants.

19
by MarkV :: Wed, 02/25/2009 - 2:23pm

I was hoping y'all would run a speed score article. Thanks, this is most informative.

It would be interesting to see other combine drills quantified... but it also seems like most coaches view "general attitude and appearance," the medical exams, the interviews, and the position drills as the most important aspects of the combine, none of which are really quantifiable.

47
by DeltaWhiskey :: Thu, 02/26/2009 - 10:10am

Really? Why not?

21
by Wait, what? (not verified) :: Wed, 02/25/2009 - 2:43pm

The problem I have with Speed Score is the same problem I have with combine stats in general: it places a lot of weight on a player's performance in one event on one specific day of his life- an event that he doesn't habitually train in, because he's a football player and not a track and field athlete. Now, Speed Score isn't at fault for this- if these guys ran the 40 in a centrally controlled environment every three months, we'd have a lot more numbers to work into the formula, but that doesn't happen. A player's combine 40 time stays with him for his career, and never gets re-evaluated so we can say, "Oh, no wonder he's so good, he's really two-tenths faster than his combine time!"

It's another piece of data to look at, at any rate, although what I'm curious to know is this: what factors are common to those players who significantly over- or under-perform their Speed Score? I'm guessing there isn't a very big sample size to work with (apparently it's good to be Brian Westbrook), so this might have to wait until 2020 or so to produce any reasonable analysis.

29
by bravehoptoad :: Wed, 02/25/2009 - 4:55pm

Good question. Besides Brian Westbrook, also Steve Slayton from last year's class. Mr. E. Smith from days of yore. Felix Jones as pointed out by Bill. Since 40 times and weights are both easy to find, you could go way, way back to increase your sample size...no need to stick to the DVOA era.

30
by Bill Barnwell :: Wed, 02/25/2009 - 5:17pm

Would love to find more 40 times if anyone knows a reliable source for Combine data from the nineties.

25
by King (not verified) :: Wed, 02/25/2009 - 3:31pm

So Lydon Murtha has a speed score of 107.38. Since he isn't projected as a top flight lineman, maybe he should switch to RB. While somewhat useful to show how fast someone is in comparison to size, this is a joke to use for primary analysis of a player.

32
by The Ninjalectual :: Wed, 02/25/2009 - 5:46pm

I work with a guy who spent a few years on the practice squad at defensive end for the Vermiel-era Chiefs. 255 lbs, 4.5 40, speed score of a whopping 124! So yes, there is much more to it than simply the score. This is much like the Lewin QB Projection system, in that it assumes the basic scouting work has been done already.

33
by Bill Barnwell :: Wed, 02/25/2009 - 5:50pm

That's true. There are all those articles I wrote that said "Speed Score is applicable on a similar scale to linemen, so you should start using it for all players."

40
by Bobman :: Wed, 02/25/2009 - 8:03pm

Yeah, but..... what useful info does it tell you about interior linemen? How often does a C run unabated anywhere? They need nimble feet and a fast burst, but beyond about 5-10 yards, is it really useful?

I thought for a moment about guys like Suggs, Freeney, Peppers, et al.... and it's more applicable to speed rushers, but even for them, without the right body positioning, hand/combat techniques, speed is all "potential" without being the biggest part of the game. Maybe. (Freeney and Mathis this season started stunting around the entire LOS, with Mathis looping all the way around right end from his LDE spot, and vice versa for Freeney. Insane that they actually got pressure that way. So I guess, yeah, being able to cover 25 yards ASAP can be a huge part of their game, not just a sudden burst and leverage....)

Is there a tiering system for positions and speed (speed scores)? Put WR and DB first, QB/P/K last, that's probably universally accepted. RB/S/pass rushers in the 2nd tier. Other linemen and LBs in the third tier. Reasonable?

42
by armchair journe... :: Wed, 02/25/2009 - 8:29pm

would imagine that different positions would require a different weight-time ratio than rb's...

10lbs on an rb is much different than 10lbs on a lineman. shoot, most linemen have at least 10lbs hanging over their belt.

38
by armchair journe... :: Wed, 02/25/2009 - 7:40pm

...pretty sure the goal is to have a more useful speed rating than the 40-time alone offers, not to be the end all be all. but, thanks for playing!

_______________________________
armchair journeyman quarterback

62
by King (not verified) :: Fri, 02/27/2009 - 1:04pm

No No.... he said this was a predictive measure of success for an NFL running back. Big difference.

49
by Rich Conley (not verified) :: Thu, 02/26/2009 - 1:06pm

"So Lydon Murtha has a speed score of 107.38. Since he isn't projected as a top flight lineman, maybe he should switch to RB. "

Maybe he should. Who knows? I certainly don't. I'm pretty sure some NFL linebackers would make great running backs.

61
by King (not verified) :: Fri, 02/27/2009 - 1:01pm

I know.... he shouldn't. There's a lot more to being a RB than your size to speed ratio including vision/quickness/balance. To say Moreno is going to be a bust based on this data and correlations is laughable.

26
by Bobman :: Wed, 02/25/2009 - 3:36pm

Any chance FO can set up an online speed score test so we could all impress one another like we did with out Wonderlic scores? It's offseason and my bragging muscles are getting flabby.

28
by Key19 :: Wed, 02/25/2009 - 4:53pm

Time yourself running 40 yards. Put your time and weight into the equation. Voila!

I posted an absolutely horrendous Speed Score of 23.7 when I ran the 40 at the NFL Experience in Tampa. Made me appreciate how fast these NFL guys are. But hey, I know I'm slow as molasses (especially for my size, I only weigh 150 pounds!). Hence my atrocious speed score.

As for Felix, not sure what to think about him. He seems to have a knack for HUGE plays even though he's not really blazing past everyone out there. In limited time, he had:

Week 1 vs Browns: 12-yard or so TD run on first NFL carry. On a draw if I remember correctly.

Week 2 vs Eagles: 99-yard Kickoff Return TD. Almost got tackled at the end though. The defender was gaining on him. Isaiah Stanback also was gaining on him, and that's kinda sad (even though the Dallas media refer to him as a "speed guy", he played QB in college and ran I think a 4.55 at the Combine or something like that, which is pretty much the equivalent of Pat White gaining on Felix, which shouldn't happen).

Week 3 vs Packers: 60-yard TD run. Madden called it a Scissor play but it looked like a counter to my *admittedly untrained* eye. Scissor = counter? I don't know. Pardon my lack of terminology knowledge. Just watching and analyzing the games without having to remember too much terminology is enough to keep me at full brain capacity. Anyways, T.O. gained on him at the end, and I know he's not as fast as he used to be. However, some people say that it's not his top speed but his acceleration that's dwindled, so if that's the case then it's not so bad (I guess).

Week 4 vs Redskins: NO CARRIES. Sorry, had to throw in that jab at Jason Garrett. At least Romo got to throw the ball 60 times (59 of them to T.O. by the way).

Week 5 vs Bengals: A few 10+ yard carries, including a 33-yard TD run on a 4th and 4 play. Totally burned everyone on that play, but hey, it was the Bengals. Chances are they were totally out of position, took bad angles, and weren't giving 100% anyways. So not the best judge of speed. Big play though.

Week 6 vs Cardinals: Two big runs and two other successful runs. 3 yards on 2nd and 5 (60% of needed yardage on 2nd down is the benchmark, right? Or is it 65%?), 5 yards on a 2nd and 1, 16 yards on a 3rd and 1 (called back due to one of the many illegal formation penalties the Cowboys had last season, but I don't think the illegal formation really made a significant difference on the run), and finally the fateful 14 yards on 2nd and 6 that ended his rookie season.

Overall, I think Felix REALLY needs a full season next year. He also needs it to be a strong one, much like his short rookie season was. Hopefully he gets back to full-speed (or better than full-speed) once the injury is finally healed and can be explosive again. We all know the Cowboys need something like that on the offense. I'm optimistic about him, but at the same time, being a Cowboys fan in these recent years has taught me to be a bit skeptical of anything that seems too good. So if he tears an ACL next season or something, I won't be surprised. Good luck to him though.

31
by Joe T. (not verified) :: Wed, 02/25/2009 - 5:37pm

You ran a 5.96 in the 40?

36
by Key19 :: Wed, 02/25/2009 - 6:31pm

5.67 hahaha Maybe I remembered the exact value of my speed score incorrectly.

My only hope is to stay as fast as I am now after gaining like 150 pounds (or more)!

It's ok though. I'm fine being slow, even though sometimes I wish I was a bit faster (like a whole second in the 40 faster!). Would make playing football casually with my friends a lot easier. I'm not horrible by any means though (on a casual level). What I lack in physical abilities, I make up for in smarts and precision (aka finding holes in zones, running good routes, essentially just playing smarter than everyone else haha). If you are good enough at anticipating and outsmarting your opponent, you can still be competitive even if they're better than you (to a point).

41
by Bobman :: Wed, 02/25/2009 - 8:07pm

I'm more of a fat 180 lb marathoner so I'll let you know how fast I time myself on an hour glass or sundial. Now I have something to shoot for (and miss).

And I wasn't being serious, I was just poking fun at our collective wonderlic one-upmanship the other day. (since we're on the topic, I got 41 out of 40 right... in eleven seconds)

48
by Sophandros :: Thu, 02/26/2009 - 10:56am

My rough calculation gives me a 78.03, but I'm a little winger.

Sports talk radio and sports message boards are the killing fields of intellectual discourse.

Oh, and Hail Eris!

51
by zlionsfan :: Thu, 02/26/2009 - 1:34pm

I ran a 5.4 40 in high school, and I weighed about 100 pounds as a freshman, so that's an outstanding 23.5.

Right now, I think I could run a 6.4. We'll say 190 pounds (I'm eating stuffed crust pizza, so that's about accurate). That's 22.6.

I almost want to run a 40 just to see how slow I would be. But not really.

57
by Bobman :: Thu, 02/26/2009 - 7:02pm

Hell, we have a loser league FF, might as well extend that to speed scores. Given a week of limbering and running I could probably manage a 5.0-5.5 time, resulting in a speed score of about 46 for me. And I was considered fast in our grad school flag football league! Might have to put some work in this weekend to see if I can manage it. Probably improve my skiing anyway.

Sure I am 44 and lumpy, but still, damn, those guys are pretty amazing.

45
by The Guy You Don't Want to Hear (not verified) :: Thu, 02/26/2009 - 1:17am

I think my speed score is 1.3, but then again I also had an ice cream break during my 40 . . .

27
by Temo :: Wed, 02/25/2009 - 3:51pm

I think the best description of Speed Score is that it provides a more accurate snapshot of a player's inherent athletic ability than raw time itself.

Similar to how everyone (non-football players, that is) refers to their ability to bench their own body weight as a benchmark, when discussing athletic ability one has to take into account other factors.

In this vein, I don't think that speed score was "wrong" on Felix Jones. The guy's measurables all around were pretty mediocre to just plain bad. I think he had something like 6 reps on the Bench (compared to 24-30 for the top 10 RBs this year).

Does doing more reps mean you'll be a better Running back? Actually, probably not-- there's little correlation there. But it does paint a picture of a guy who's probably limited physically (compared to the average football player, anyway).

And in truth, it's well documented that the Cowboys drafted him a part time running back and return specialist, so it's not like they think/thought they were going to get a future hall of famer or anything here.

35
by Big Johnson (not verified) :: Wed, 02/25/2009 - 6:20pm

I love the speed score article in the 2008 prospectus. Keep up the good work!

37
by AlanSP (not verified) :: Wed, 02/25/2009 - 7:39pm

A few questions/comments

1. In the original analysis, did you look at the 10/20 yard splits for running backs, or just the total 40 time? I know it might seem like dissecting 4.5 seconds worth of info a bit too closely, but with splits and some arithmetic, there are really 3 separate measurements: times from 0-10, 10-20, and 20-40, which might be measuring different things (i.e. top-end speed vs. getting off to a quick start). In the 3 years of data that I have, there's a much stronger correlation between the 10-20 and 20-40 times than between 0-10 and 10-20. It's possible that the one of these partial times might account for the predictive value of the full time? For example, if top-end speed is the important factor, you might expect the 20-40 time to be the best predictor. Sorry if this sounds sort of convoluted.

2. Where did you find the original data from 1999-2008?

3. It's sort of unfortunate that the combine location was moved right after the original speed score formula was developed. If there are systematic differences between the RCA dome track and the one at Lucas Oil Stadium, it sort of throws everything off. And we don't have anyone who ran in both locations, it makes it hard to determine the effect of the location.

43
by Ryan Harris (not verified) :: Wed, 02/25/2009 - 9:39pm

Last year was my first time buying the book and I absolutely loved it. I used speed score to draft my backs and took Johnson and Forte well ahead of where the "pundits" said they should go. I got a little ribbing about it from the guys in my league, until about week 2.

Anyway I think last year you guys hit a homerun with speedscore, going forward though I would say "caveat emptor". The 2008 RB's were phenomenal and I think there may be more to this.

There are other factors that need to be included that I dont believe you can exactly measure such as coaching, o-line play and ability to read the holes. That being said, I think speed score will prove to do a better job of projecting which guys not to take, rather than which guys too take.

The list of RB's who are disapproved is more accurate then the approved list. Brian Westbrook is the only elite back on the "no" list; Travis Henry, Chester Taylor and Sproles are the serviceable players on this list, and Sproles could probably run a 4.1 40 but because of his size would still not register a strong speedscore.

However on the "approved list" there are some awful players like JJ Arrington, Chris Brown, Eric Shelton and a few more. So if I was an NFL GM or a Fantasy GM (or both, I can dream) I would use the "no" list to whittle down my board, and then use film and workouts to judge the remainder.

50
by MCS :: Thu, 02/26/2009 - 1:23pm

I think you are onto something here. It shows that speed is not everything. but lack of a certain level of speed is.

When everyone in the league is fast, including linemen, a simple thing like a hesitation in the backfield or a stutter-step at the wrong time can be the difference between a big gain and tackle for loss. That is not pure speed. That is vision and decisiveness coupled with a certain level of speed.

52
by Jimmy :: Thu, 02/26/2009 - 2:10pm

In horse racing the horses perform differently on hard surfaces to soft ones. That is some horses have low running actions where they dig in their hooves and perform much better on soft courses than high action horses. High action horses are much quicker on harder ground as they can run fast without digging in their hooves. It is mainly a product of a horse running on ground that they are most suited to being far more efficient than when conditions suit others in the race.

I find it difficult to beleive that NFL players don't also have this kind of variability in their running styles, but no one seems to be paying it any attention. If I were GM of a team who played on grass most of the time I would be looking to emphasize drafting WRs, RBs, LBs and DBs who are best suited to playing on a softer, more giving surface. By the end of a football game the increased efficiency that they are able to find would (I imagine) be fairly noticable, probably enough to help your team quite a bit.

54
by Ryan Harris (not verified) :: Thu, 02/26/2009 - 5:41pm

To the FO guys,

Any chance you would be willing to do this speed score evaluation again once the backs run the 40 at their pro days?

I would do it, but I would not be able to post it and even if I could it would probably turn into Brady V Manning XXXI.

55
by Bill Barnwell :: Thu, 02/26/2009 - 6:43pm

Speed score isn't applicable to Pro Day data.

I did look at 10-20 yard stretches, subtracting those from the 40, etc. They're all around the same as the 40 time, and they're all behind Speed Score. That was actually my initial idea before I ever fell upon Speed Score.

56
by Ryan Harris (not verified) :: Thu, 02/26/2009 - 6:51pm

Sorry I missed something there, do they not run the 40 on their pro day?

I just thought that maybe there is something to the surface at Lucas Oil, and if they ran a 40 at their pro day you could re-do the speed score with the new time, assuming of course there is a significant difference.

58
by Bill Barnwell :: Thu, 02/26/2009 - 9:07pm

No, players do run their 40 on the Pro Day, but there's no correlation between Pro Day 40 time and NFL success, let alone Speed Score based upon Pro Days and NFL success. It's addressed in the PFP 2008 essay.

60
by Carson (not verified) :: Thu, 02/26/2009 - 11:36pm

Knowshon's 40-time is listed at 4.48 at nfldraftscout. They've said his and many other's times aren't listed accurately by NFL.com. I'd wait to see how he performs at his Pro-Day before casting judgement on his NFL future. He'll be an excellent pro.

At least you glossed over his agility and quickness, but his overall body-control is what seperates him from the pack. He's the class of this draft. Not only is he an excellent reciever (50+ receptions last year), he rarely fumbles, and he's an accomplished blocker who protects his QB. He shines in the Red Zone, and as a Texan fan I'd welcome him to my team as the 15th pick.

63
by panthersnbraves :: Sat, 02/28/2009 - 3:36am

At some point you wonder how someone 400 lbs walking a 10second 40 would score.

65
by Yaguar :: Sat, 02/28/2009 - 6:11pm

Very, very poorly.

64
by Slots Ville (not verified) :: Sat, 02/28/2009 - 7:43am

Player ........... Score ..... DYAR ..... DVOA ..... Att-Yd ........ EYds ..... Teammate with 50+ att
C.Johnson ..... 121.9 ...... 170 ........ 9.2% ..... 251-1228 ..... 1122 ..... L.White: 7.5% DVOA, 200 carries
D.McFadden .. 120.0 ...... -14 ....... -9.1% ..... 113-500 ........ 365 ..... J.Fargas: -13.8% DVOA, 219 carries
J.Stewart ....... 116.7 ...... 111 ....... 7.8% ...... 184-838 ........ 826 ..... D.Williams: 28.5% DVOA, 273 carries
M.Forte ......... 109.7 ....... 38 ....... -7.0% ..... 316-1239 ..... 1233
R.Rice ............ 99.8 ........ 20 ........ -2.4% ..... 108-462 ........ 381 ..... L.McClain: 7.6% DVOA, 232 carries
K.Smith .......... 98.6 ........ 88 ........ -0.3% ..... 238-975 ....... 984 ..... R.Johnson: -21.3% DVOA, 76 carries
S.Slaton ......... 96.9 ........ 170 ........ 5.8% ..... 268-1285 ..... 1269 ..... A.Green: 12.4% DVOA, 74 carries

66
by parker (not verified) :: Sun, 03/01/2009 - 2:17am

1.Speed score looks useful when also taken in the context of a players college success.
2. Everyone brings up Westbrook and Emmit Smith. These are two guys who excelled in one system. What would have happened if they were drafted by different teams?

3.I look forward to an article on the wr size correlations and pray that Gil Brandt hasn't read pfp 08.

Thanks for all your hard work.

67
by Brian (not verified) :: Sun, 03/01/2009 - 11:03am

Hi, I like to run numbers such as this myself. However, I have had a terrible time trying to acquire the combine data.Is there an way you could share your results from the NFL combine over the last decade, or to provide the website where such data can be acquired? For what it's worth, I have been using for two years a similar number- Weight/(40time-3) that produces similar results. Keep up the great work guys.

68
by Lew Alcindor (not verified) :: Sun, 03/01/2009 - 8:02pm

Knowshon...has to be the most pushed product I've seen in some time. He doesn't have break away speed, he's competitive and a darter, but he is not explosive. He doesn't have that "edge" in a skill that will make him a full time starter. My Georgia buddies even say he came out too soon--so he's probably shown breakdowns during the season that will scare some teams off.

These backs are a bunch of situational backs--Wells and McCoy might make it as starters, everyone else will be on the bench most of the game. I'm guessing about six of these guys will still be in the league in 3-4 years, one or two may be all we remember.

72
by Greg (not verified) :: Wed, 03/11/2009 - 12:22pm

Didn't the RBs complete the 40 last/towards the end of the day this year though? If so, wouldn't we maybe expect them to run a little slower as a whole?

73
by Rexy :: Tue, 08/04/2009 - 4:15pm

I'm nervous about a system where a small amount of variance (say .05 seconds) can have a large effect on the overall score, especially given the very small sample size.