Writers of Pro Football Prospectus 2008

01 Mar 2010

ESPN: Speed Score 2010

Fresh off the presses, our Speed Score article reveals the sleeper back of the 2010 class (it's Ben Tate), and provides a list of middling comparables for C.J. Spiller.

Posted by: Bill Barnwell on 01 Mar 2010

29 comments, Last at 03 Mar 2010, 9:32am by jebmak

Comments

1
by Anonymousse (not verified) :: Mon, 03/01/2010 - 3:33pm

Weights, 40 times, and the speed score formula are all already freely available.

2
by dk240t :: Mon, 03/01/2010 - 4:21pm

Name/40time (sec)/weight (lb)/SS (lb/sec^4)
Tate, Ben 4.43 220 114
Mathews, Ryan 4.45 218 111
Best, Jahvid 4.35 199 111
Hardesty, Montario 4.49 225 111
Spiller, C.J. 4.37 196 107
Starks, James 4.5 218 106
Miller, Lonyae 4.53 221 105
McKnight, Joe 4.47 198 99.2
McNeal, Shawnbrey 4.56 194 89.7
McCluster, Dexter 4.58 172 78.2

And my list even comes with correct significant digits.

3
by The Ninjalectual :: Mon, 03/01/2010 - 4:35pm

=P

4
by KarlFA :: Mon, 03/01/2010 - 5:06pm

No love for Toby Gerhart? He scored out at 110.

Karl, Miami

7
by dk240t :: Mon, 03/01/2010 - 8:53pm

I copied that list from NFL.com for the 40 times, then went to the individual player pages to get their weights. If someone isn't on the list, then they weren't on the combine RB 40 time list on nfl.com.

6
by bravehoptoad :: Mon, 03/01/2010 - 8:43pm

Since the average for a first-round back is 113, only Ben Tate comes out above average this year. Matthews, Best, Hardesty, and Gerhart are close to average for a first rounder.

Speed score doesn't think this is a great class. Look at Chris Johnson's 122 or Brandon Jacobs' 123.5.

29
by jebmak :: Wed, 03/03/2010 - 9:32am

No one close to Bruce Campbell's 120.

5
by ebongreen (not verified) :: Mon, 03/01/2010 - 5:30pm

So, in round numbers, and based on a generally desirable speed score of around 105, the rule of thumb follows:

At 200 lbs, you want to run about a 4.4 or better (speed score = 107). For every 20 lbs more you weigh, you get an extra 0.1 second slack; for every 20 lbs under, you better run 0.1 second faster.

So if your 200 lb RB runs a 4.3, he's pretty studly; if he runs a 4.5, he's pretty weak sauce. A 220 lb-er running 4.4 is fab, running a 4.6 is bad. And so on and so forth.

LeGarette Blount running 4.6 at 240 is decent, for example. It's all nice and nearly linear. Don't you like pretty numbers?

captcha: blandest missile

12
by Dan :: Tue, 03/02/2010 - 3:57am

You are correct. 20 pounds is worth about 0.1 seconds for speed score. I'd wondered if there was a linear formula that gave almost the same results as speed score, and I just ran some numbers and this is it. For this year's RBs, there's a correlation of 0.9987 between speed score and a measure that assumes a linear relationship with 20 pounds worth 0.1 second (I just rescaled 40-times by multiplying them by 200, and subtracted that from the weight). If you let the scaling factor vary instead of fixing it at 20, it looks like the optimal scaling to match speed score is 19.3 pounds to .1 second (r = 0.99905), so 20 is close enough.

There is even a simple way to get a scale centered at 100: just take the 40-time after the decimal point (ignoring the 4 seconds), double it, and subtract that from the weight. For instance, Joe McKnight had a pretty average speed score of 99.2 (198 lbs., 4.47 40-time); double that 47 to get 94 and then subtract that 94 from 198 to get an alternate speed score of 104. Ben Tate (4.43 at 220lb.) is the leader on this measure too: 43 doubled is 86, subtracted from 220 gives you 134. Another nice feature: everyone's score is an integer.

How good is a 134? This alternate speed score has a distribution similar to IQ: 100 is average for the combine (speed score around 98), 115 is good (speed score around 105), 130 is great (speed score around 112, a typical first rounder), and 145 is elite (speed score around 120, equal to #10 on the all-time list). Brandon Jacobs' 155 is the highest of all time. So Ben Tate's 134 is excellent, slightly better than a typical first rounder. Jahvid Best is in second with a 129 (a 4.35 at 199lb., so subtract 70 from 199), closely followed by Mathews (128 = 218-90) and Hardesty (127 = 225-98).

8
by BigWoody (not verified) :: Mon, 03/01/2010 - 11:26pm

What the hell is Spiller's 40 time? I've seen it reported at 4:27, 4:37, 4:34, 4:33 !!!??? And why??

9
by The Ninjalectual :: Mon, 03/01/2010 - 11:56pm

why not?

10
by BigWoody (not verified) :: Tue, 03/02/2010 - 1:17am

Sorry, can you explain that?

14
by tally :: Tue, 03/02/2010 - 10:54am

It's called variability. If he ran the 40 four different times, I wouldn't expect every result to be identical. Different surfaces and how the athlete's condition on a given day would produce some spread in the results.

11
by Dan :: Tue, 03/02/2010 - 2:14am

Based on this source for 40 times, this year's speed scores are:

speed score name 40-time weight
114.2 Ben Tate 4.43 220
111.2 Ryan Mathews 4.45 218
111.2 Jahvid Best 4.35 199
110.7 Montario Hardesty 4.49 225
107.8 Toby Gerhart 4.55 231
107.5 C.J. Spiller 4.37 196
106.3 James Starks 4.50 218
105.0 Lonyae Miller 4.53 221
99.2 Joe McKnight 4.47 198
97.5 Charles Scott 4.70 238
97.2 Javarris James 4.57 212
97.1 Anthony Dixon 4.68 233
96.3 LeGarrette Blount 4.73 241
95.5 Jonathan Dwyer 4.68 229
93.8 Chris Brown 4.60 210
91.7 Joique Bell 4.68 220
89.7 Shawnbrey McNeal 4.56 194
88.5 Stafon Johnson 4.69 214
88.5 Andre Dixon 4.64 205
87.3 Keith Toston 4.70 213
85.7 Pat Paschall 4.70 209
82.7 Darius Marshall 4.63 190
78.2 Dexter McCluster 4.58 172

I'm not sure if these are the actual, official 40 times. The ten players who made the NFL.com top performers list have the same times in both places, which counts in favor of the reliability of these numbers, but the ten players on the top performers list are not the ten fastest according to these numbers, which counts against their reliability (e.g. why didn't Gerhart make the top performers list if he ran a 4.55?).

13
by Dr. Mooch :: Tue, 03/02/2010 - 9:31am

I still think we should use speed score with metric weights and a 400 coefficient, but no one else cares a whit about the aesthetics.

15
by tally :: Tue, 03/02/2010 - 10:55am

It's all normalized to a mean of 100, so I don't see how it matters.

17
by Dr. Mooch :: Tue, 03/02/2010 - 11:25am

FO's glossary only notes that the average tends to be around 100. My understanding is that the scores are merely calculated, and I don't recall reference to any procedure by which the scores are normalized.

20
by drobviousso :: Tue, 03/02/2010 - 1:05pm

It was part of the formula creation, not a post calculation normalization.

25
by billsfan :: Tue, 03/02/2010 - 6:45pm

Numbers already confuse the casual NFL fan, so it's nice to normalize it to 100.

What would really be aesthetically pleasing is if speed score were measured in Joules, since it's basically just a bastardized kinetic energy calculation.

Or just give RBs ratings in horsepower.

(I also like the Eagles)

27
by Dr. Mooch :: Tue, 03/02/2010 - 11:43pm

Yeah, kg*400 would bring the top scores in the 100 range, instead of the lbs*200 having the average around 100. But it would be fun to give measurements in actual power.

28
by Dr. Mooch :: Wed, 03/03/2010 - 1:16am

And for some kind of stupid geekiness, here are the numbers from an earlier list reprocessed as Joules in the bastardized kinetic energy sort of way:

Tate, Ben 4.43 220 3409
Mathews, Ryan 4.45 218 3348
Best, Jahvid 4.35 199 3198
Hardesty, Montario 4.49 225 3394
Spiller, C.J. 4.37 196 3121
Starks, James 4.5 218 3274
Miller, Lonyae 4.53 221 3275
McKnight, Joe 4.47 198 3014
McNeal, Shawnbrey 4.56 194 2837
McCluster, Dexter 4.58 172 2494

You could normalize them to Tate's score, giving a series of (100, 98.2, 93.8, 99.6, 91.6, . . . 73.1) but that uses a coefficient with too many decimal places. If you figure Tate's not perfect and round it off nicely you can give him a more modest score of 95.5 and McCluster a 69.8, and get a very simple formula for a new speed score of 8.515 * Weight / 40-time^2 that tends to top out around 100. (Chris Johnson a 93, McFadden a 97, although it still gives Brandon Jacobs a 109 - Or, hey, add in distance and time once more each to get yourself a very bastardized power formula that irons out as 36.71 * Weight / 40-time^3 that gives these guys scores between 65 and 92, gives Johnson and McFadden 95's and drops Jacobs in just a smidge over 103.)

16
by drobviousso :: Tue, 03/02/2010 - 11:17am

I've been telling my brother, who's a college football fan first and foremost, that CJ Spiller might be 'the most explosive back in college football', but he's not the best pro prospect.

Anyone know anything about this James Starks guy? Looks like he was injured last year, but with this speed score, he's got to be worth a mid/late round pick at least.

18
by Ben Stuplisberger :: Tue, 03/02/2010 - 12:21pm

According to this source, http://www.mercurynews.com/sports/ci_14495424 Gerhart ran a 4.53 officially at the combine, which puts his speed score at 110. It seems he has what it takes to be a halfback, let's see what happens.

19
by bravehoptoad :: Tue, 03/02/2010 - 12:49pm

A white halfback. Imagine that. Next thing you know, they'll be making white cornerbacks, too. What a world.

21
by CJ (not verified) :: Tue, 03/02/2010 - 2:33pm

Too bad Taylor Mays' 4.24 unofficial didn't stand, would have been a ridiculous 142.3 speed score. Got knocked down to a 4.43, still a good score of 119.4

22
by drobviousso :: Tue, 03/02/2010 - 3:31pm

How on earth are these things measured. Is it really done by hand? Are they just asking Easterbrook what looks right after the fact?

23
by tuluse :: Tue, 03/02/2010 - 3:40pm

the start is based on a pressure plate, but the end is hand timed.

24
by Karl Mal0wned! (not verified) :: Tue, 03/02/2010 - 4:59pm

They're looking into using laser for the finish, similar to the way they were gonna use lasers to determine where punts went out of bounds. At least that what Feagles told me.

26
by Nathan :: Tue, 03/02/2010 - 9:53pm

Anyone care to explain this?

http://www.youtube.com/watch?v=oDhYKZHKh4Q

It pretty clearly shows Mays (4.43) finishing ahead of Holliday (4.34).