Writers of Pro Football Prospectus 2008

06 Sep 2012

Look, Kids: Snap Data!

Hey folks. I mentioned a few days ago that the NFL is now making snap data available to the media a few hours after each game. Like the all-22 film, this is one of the Holy Grails of NFL stat analysis that is finally public in 2012. We're still playing around with some fun new ways to display this data on a weekly basis, but I'm so excited about having it that I wanted to share with the class.

Every Tuesday afternoon, we'll be able to put up a table that looks something like this. Thanks to Abe vander Bent for helping write a parser that grabs the data for us from the NFL's media website. This table is in order by snap total, but I may decide to display things separated into offense and defense (with a third group for guys who play special teams only). We're also playing around with ideas for sortable tables, an oft-requested feature for FO. (Right now only the tables in FO Premium are sortable.)

WK TEAM PLAYER POS ACT TOT
SNAP
OFF
SNAP
OFF SNAP
PCT
DEF
SNAP
DEF SNAP
PCT
ST
SNAP
ST SNAP
PCT
1 DAL 71-N.Livings LG TRUE 72 68 100% 0 0% 4 18%
1 DAL 73-M.Bernadeau RG TRUE 72 68 100% 0 0% 4 18%
1 DAL 68-D.Free RT TRUE 72 68 100% 0 0% 4 18%
1 DAL 63-R.Cook C/G TRUE 69 65 96% 0 0% 4 18%
1 DAL 77-T.Smith LT TRUE 68 68 100% 0 0% 0 0%
1 DAL 9-T.Romo QB TRUE 68 68 100% 0 0% 0 0%
1 DAL 43-G.Sensabaugh FS TRUE 67 0 0% 56 100% 11 50%
1 DAL 42-B.Church SS TRUE 64 0 0% 56 100% 8 36%
1 DAL 88-D.Bryant WR TRUE 60 58 85% 0 0% 2 9%
1 DAL 82-J.Witten TE TRUE 57 53 78% 0 0% 4 18%
1 DAL 19-M.Austin WR TRUE 57 57 84% 0 0% 0 0%
1 DAL 50-S.Lee WLB TRUE 56 0 0% 56 100% 0 0%
1 DAL 93-A.Spencer SLB TRUE 55 0 0% 55 98% 0 0%
1 DAL 39-B.Carr LCB TRUE 54 0 0% 54 96% 0 0%
1 DAL 24-M.Claiborne RCB TRUE 54 0 0% 54 96% 0 0%
1 DAL 29-D.Murray RB TRUE 52 52 76% 0 0% 0 0%
1 DAL 94-D.Ware JLB TRUE 51 0 0% 51 91% 0 0%
1 DAL 97-J.Hatcher DT TRUE 43 0 0% 43 77% 0 0%
1 DAL 47-L.Vickers FB TRUE 40 29 43% 0 0% 11 50%
1 DAL 36-M.Silva S TRUE 39 0 0% 22 39% 17 77%
1 DAL 54-B.Carter MLB TRUE 37 0 0% 30 54% 7 32%
1 DAL 89-J.Phillips TE TRUE 37 30 44% 0 0% 7 32%
1 DAL 32-O.Scandrick CB TRUE 34 0 0% 27 48% 7 32%
1 DAL 85-K.Ogletree WR TRUE 33 33 49% 0 0% 0 0%
1 DAL 95-S.Lissemore DT TRUE 32 0 0% 27 48% 5 23%
1 DAL 92-J.Price-Brent NT TRUE 28 0 0% 23 41% 5 23%
1 DAL 99-K.Coleman DT TRUE 24 0 0% 24 43% 0 0%
1 DAL 52-D.Connor LB TRUE 22 0 0% 4 7% 18 82%
1 DAL 84-J.Hanna TE TRUE 20 9 13% 0 0% 11 50%
1 DAL 55-A.Albright LB TRUE 18 0 0% 0 0% 18 82%
1 DAL 96-M.Spears DT TRUE 18 0 0% 18 32% 0 0%
1 DAL 28-F.Jones RB TRUE 16 12 18% 0 0% 4 18%
1 DAL 57-V.Butler LB TRUE 15 0 0% 6 11% 9 41%
1 DAL 31-M.Butler CB TRUE 14 0 0% 7 12% 7 32%
1 DAL 34-P.Tanner RB TRUE 14 1 1% 0 0% 13 59%
1 DAL 70-T.Crawford DT TRUE 13 0 0% 3 5% 10 45%
1 DAL 15-A.Holmes WR TRUE 12 0 0% 0 0% 12 55%
1 DAL 17-D.Harris WR TRUE 11 0 0% 0 0% 11 50%
1 DAL 5-D.Bailey K TRUE 9 0 0% 0 0% 9 41%
1 DAL 6-C.Jones P TRUE 6 0 0% 0 0% 6 27%
1 DAL 78-J.Parnell T TRUE 6 2 3% 0 0% 4 18%
1 DAL 91-L.Ladouceur LS TRUE 6 0 0% 0 0% 6 27%
1 DAL 11-C.Beasley WR TRUE 4 4 6% 0 0% 0 0%
1 DAL 76-D.Dockery G TRUE 4 0 0% 0 0% 4 18%
1 DAL 67-P.Costa C TRUE 3 3 4% 0 0% 0 0%
1 DAL 18-K.Orton QB TRUE 0 0 0% 0 0% 0 0%
1 DAL 21-M.Jenkins CB FALSE 0 0 0% 0 0% 0 0%
1 DAL 37-M.Johnson S FALSE 0 0 0% 0 0% 0 0%
1 DAL 40-D.McCray S FALSE 0 0 0% 0 0% 0 0%
1 DAL 51-K.Wilber LB FALSE 0 0 0% 0 0% 0 0%
1 DAL 62-D.Arkin G FALSE 0 0 0% 0 0% 0 0%
1 DAL 87-C.Cochart TE FALSE 0 0 0% 0 0% 0 0%
1 DAL 90-J.Ratliff NT FALSE 0 0 0% 0 0% 0 0%
WK TEAM PLAYER POS ACT TOT
SNAP
OFF
SNAP
OFF SNAP
PCT
DEF
SNAP
DEF SNAP
PCT
ST
SNAP
ST SNAP
PCT
1 NYG 23-C.Webster LCB TRUE 70 0 0% 68 100% 2 9%
1 NYG 21-K.Phillips SS TRUE 68 0 0% 68 100% 0 0%
1 NYG 26-A.Rolle FS TRUE 68 0 0% 68 100% 0 0%
1 NYG 85-M.Bennett TE TRUE 62 54 96% 0 0% 8 36%
1 NYG 90-J.Pierre-Paul RE TRUE 62 0 0% 59 87% 3 14%
1 NYG 75-S.Locklear LT TRUE 59 56 100% 0 0% 3 14%
1 NYG 77-K.Boothe LG TRUE 59 56 100% 0 0% 3 14%
1 NYG 64-D.Baas C TRUE 59 56 100% 0 0% 3 14%
1 NYG 76-C.Snee RG TRUE 59 56 100% 0 0% 3 14%
1 NYG 66-D.Diehl RT TRUE 59 56 100% 0 0% 3 14%
1 NYG 97-L.Joseph LDT TRUE 59 0 0% 55 81% 4 18%
1 NYG 93-C.Blackburn MLB TRUE 59 0 0% 51 75% 8 36%
1 NYG 10-E.Manning QB TRUE 56 56 100% 0 0% 0 0%
1 NYG 88-H.Nicks WR TRUE 54 54 96% 0 0% 0 0%
1 NYG 44-A.Bradshaw RB TRUE 53 53 95% 0 0% 0 0%
1 NYG 37-M.Coe RCB TRUE 52 0 0% 47 69% 5 23%
1 NYG 80-V.Cruz WR TRUE 51 51 91% 0 0% 0 0%
1 NYG 91-J.Tuck LE TRUE 51 0 0% 47 69% 4 18%
1 NYG 95-R.Bernard RDT TRUE 50 0 0% 50 74% 0 0%
1 NYG 57-J.Williams LB TRUE 46 0 0% 33 49% 13 59%
1 NYG 59-M.Boley LB TRUE 43 0 0% 38 56% 5 23%
1 NYG 30-J.Tryon CB TRUE 41 0 0% 22 32% 19 86%
1 NYG 72-O.Umenyiora DE TRUE 41 0 0% 41 60% 0 0%
1 NYG 94-M.Kiwanuka SLB TRUE 39 0 0% 34 50% 5 23%
1 NYG 28-J.Hosley CB TRUE 34 0 0% 28 41% 6 27%
1 NYG 87-D.Hixon WR TRUE 29 27 48% 0 0% 2 9%
1 NYG 55-K.Rivers WLB TRUE 26 0 0% 24 35% 2 9%
1 NYG 45-H.Hynoski FB TRUE 21 16 29% 0 0% 5 23%
1 NYG 78-M.Kuhn DT TRUE 21 0 0% 14 21% 7 32%
1 NYG 52-S.Paysinger LB TRUE 19 0 0% 0 0% 19 86%
1 NYG 27-S.Brown S TRUE 18 0 0% 0 0% 18 82%
1 NYG 58-M.Herzlich LB TRUE 16 0 0% 1 1% 15 68%
1 NYG 35-A.Brown RB TRUE 15 0 0% 0 0% 15 68%
1 NYG 13-R.Barden WR TRUE 12 1 2% 0 0% 11 50%
1 NYG 86-B.Pascoe TE TRUE 12 12 21% 0 0% 0 0%
1 NYG 31-W.Hill S TRUE 10 0 0% 0 0% 10 45%
1 NYG 22-D.Wilson RB TRUE 8 3 5% 0 0% 5 23%
1 NYG 63-J.Cordle C TRUE 8 0 0% 0 0% 8 36%
1 NYG 5-S.Weatherford P TRUE 7 0 0% 0 0% 7 32%
1 NYG 9-L.Tynes K TRUE 7 0 0% 0 0% 7 32%
1 NYG 51-Z.DeOssie LS TRUE 7 0 0% 0 0% 7 32%
1 NYG 98-A.Tracy DE TRUE 7 0 0% 0 0% 7 32%
1 NYG 65-W.Beatty TE TRUE 5 5 9% 0 0% 0 0%
1 NYG 81-A.Robinson TE TRUE 2 2 4% 0 0% 0 0%
1 NYG 82-R.Randle WR TRUE 2 2 4% 0 0% 0 0%
1 NYG 8-D.Carr QB TRUE 0 0 0% 0 0% 0 0%
1 NYG 12-J.Jernigan WR FALSE 0 0 0% 0 0% 0 0%
1 NYG 20-P.Amukamara CB FALSE 0 0 0% 0 0% 0 0%
1 NYG 33-D.Scott RB FALSE 0 0 0% 0 0% 0 0%
1 NYG 62-M.Petrus G FALSE 0 0 0% 0 0% 0 0%
1 NYG 71-A.Ojomo DE FALSE 0 0 0% 0 0% 0 0%
1 NYG 73-J.Brewer T FALSE 0 0 0% 0 0% 0 0%
1 NYG 96-M.Austin DT FALSE 0 0 0% 0 0% 0 0%

Posted by: Aaron Schatz on 06 Sep 2012

22 comments, Last at 08 Sep 2012, 7:42pm by Aaron Schatz

Comments

1
by RC (not verified) :: Thu, 09/06/2012 - 5:12pm

What is Mike Reiss going to do with all his time now?

3
by PatsFan :: Thu, 09/06/2012 - 7:30pm

You stole my line! :)

2
by Joseph :: Thu, 09/06/2012 - 6:53pm

I'll put in my 2 cents: If someone played 100% of offensive or defensive snaps, put them in green, for ex. If they played in 80+ of snaps, put them in blue. If they played between 20-50%, in yellow; if less than 20%, in red (whatever color combo works easiest). That way we can identify top-notch players, starters, prominent subs (like WR3, CB3, TE2, etc), and niche/bench players.
I would also change the true/false for active/inactive to either T/F, or IMO, better to be Y/N or A/I. That column isn't very important to the table. Also, since it's a SNAP COUNT table, eliminate "SNAP" from every column. It just makes the table "bulkier."

10
by Eddo :: Fri, 09/07/2012 - 9:56am

I'd rather see one color chosen, but adjust the shading based on the percentage. In grayscale, for example, a 100% player would have a black (#000000) background, a 0% player white (#ffffff), and a 50% player medium gray (#777777).

Joseph's color idea is OK if you're strictly bucketing (i.e. no shading), but could be difficult for those of us who are colorblind.

12
by Joseph :: Fri, 09/07/2012 - 10:26am

Hey, Eddo, I'm colorblind to a certain extent too. I must be the opposite of you, because I have trouble with black vs. navy blue vs. dark gray. If I'm not paying close attention, I mix up spades & clubs at cards, too (but not diamonds & hearts). Anyway, as you discerned, my idea is just to make the numbers easier to sort visually.
I second the idea of another commenter to group them by position on offense and defense.
Another idea might be to have a table for everyone who takes an offensive snap, a defensive step, and a third for ST snaps. (Even though there might be the occasional defensive/ST snap for a Randy Moss type on a Hail Mary, or a NT getting a snap at FB on the goal line.) In this way, the table wouldn't be so wide.

16
by Eddo :: Fri, 09/07/2012 - 12:56pm

Huh, that's interesting. I have no trouble with shades (that is, I can tell black from dark gray, or navy blue from regular blue), but I have a lot of trouble discerning different colors themselves, unless they're more pure.

To elaborate: I can tell apple red from kelly green really well, but not a dark red from a dark green. I've asked co-workers to avoid using the standard red-green dichotomy on heat maps for this reason; I can pick out the extremes, but not the detail.

19
by Joseph :: Fri, 09/07/2012 - 4:48pm

I get you. When the eye doc asked my parents years ago if I had trouble with colors, I didn't get it--I mean, I never mixed up different colors. But I can't look at my sock drawer and tell if I picked up black or navy until I compare it to another pair. (That's why most of my dark socks have a stripe or other identifying characteristic :)) Same with dress slacks.

4
by fb29 :: Thu, 09/06/2012 - 8:02pm

I think position groupings would make sense on offense. All the O-Line together, QBs, WRs, RBs and TEs. Then sorted by # of snaps or maybe %age.

17
by jebmak :: Fri, 09/07/2012 - 2:35pm

I like this idea. Also the shaded colors for 100%, 75%+, etc.

5
by Karl Cuba :: Fri, 09/07/2012 - 5:38am

It could be me being thick but I haven't the foggiest why the chart has TRUE and FALSE in one column.

6
by Raiderjoe :: Fri, 09/07/2012 - 6:29am

True- active

False- inactice

7
by Revenge of the NURBS (not verified) :: Fri, 09/07/2012 - 8:34am

Yeah, it took me a minute to figure that one out too. It definitely draws the eye more than it probably should. In the case of anyone who played even a single snap, the TRUE is implied. It's really only relevant information for players that got 0 snaps, and in this particular case, for both teams, the only active player to get 0 snaps was the backup QB.

8
by Bowl Game Anomaly :: Fri, 09/07/2012 - 8:36am

I think this data could go a long way toward telling us which WRs, TEs, and RBs are good at blocking (or being decoys I guess). In the past we only had target data but arguably if WR A has more targets but WR B has more snaps, WR B is more valuable to his team.

9
by Revenge of the NURBS (not verified) :: Fri, 09/07/2012 - 9:25am

Arguably, perhaps, if injury is not the cause of the target/snap disparity.

Just for discussion purposes, I'll throw this out there -- if WR A gets more targets than WR B, but WR B is getting more snaps, perhaps this indicates that something is off in terms how these players are being utilized. (Again, taking injury out of the equation.) Obviously, having never seen this level of detail in the data before, it's hard to know if this situation even comes up. But it would seem to me that either WR A is better at getting open and should be playing more than WR B, or the QB locks in on WR A and forces passes to him (perhaps in an attempt to satisfy the Randy Ratio).

11
by rageon :: Fri, 09/07/2012 - 10:05am

Perhaps yards after catch by other receivers when given players are on/off the field. Is Player A more likely to have a lot of yards after catch with Player B on the field or Player C on the field? Perhaps the same could be said about certain types of running players as well. There would be a lot of noise in the date, and receivers who demand double coverage all the time may skew the numbers, but it may be worth investigating.

20
by Jerry :: Fri, 09/07/2012 - 7:31pm

We don't have who's participating on individual plays yet, so we can't measure those interreactions. Having this snap data is better than not having it, but it's a long way from being seriously useful.

13
by Wondererer (not verified) :: Fri, 09/07/2012 - 10:45am

Why is the backup QB for each team showing as "True" with 0 credited snaps?

14
by Revenge of the NURBS (not verified) :: Fri, 09/07/2012 - 11:54am

On the active roster, but did not play.

15
by Biebs :: Fri, 09/07/2012 - 11:56am

True means they were on the 46 person active roster, even if they didn't play any snaps.

18
by The Powers That Be :: Fri, 09/07/2012 - 3:57pm

Question: the Cowboys ran 57 offensive plays, but the table lists 68 offensive snaps. Does it include all the pre-snap penalties as actual snaps?

21
by ch (not verified) :: Fri, 09/07/2012 - 8:08pm

Most presnap penalties are counted. Illegal Substitution and some Delay of Game penalties where neither team lines up (usually punt plays) will not be listed.

22
by Aaron Schatz :: Sat, 09/08/2012 - 7:42pm

And he knows, because that's Chris Hoeltge, the man at the NFL office responsible for distributing this data to the media!