Panel Recap – Avg. Depth of Target

| August 22, 2012

This past weekend, I had the pleasure of attending Fantasy Football Fest in Atlantic City. There were dozens of players in attendance (Michael Vick, LeSean McCoy, and Phil Simms among them), several fantasy football experts in the house (Matthew Berry, Dave Richard on the list), and whole lot of people trying to sell their innovative product.

The Atlantic City Convention Center proved to be a good fit for the event. There were video games (including Tecmo Bowl), live drafts, lingerie football players, and even a few informative panels.

Although the Matthew Berry/LeSean McCoy/Michael Vick panel certainly stole the show, the fine folks at ReedPOP allowed us (“us” being Jeff Ratcliffe, Bryan Fontaine, and host Jim Day) to put on a panel of our own. The topic? Next-generation statistics in Fantasy Football. As if you needed to ask.

Today, I’m going to review half of my part of the presentation. This discussion will seem very familiar, as it’s a lengthy discussion on Average Depth of Target – a topic I’ve covered before.

In the other half of the presentation, I did a detailed analysis of snap and target distribution.

Normalized aDOT Correlation

As a quick review, here’s what you need to know about Average Depth of Target, or ‘aDOT’:

What is it? Depth (in yardage) of each target divided by total targets. It focuses solely on how deep down field a player is when targeted.

Why is it better? It’s an improvement on a stat like Yards-Per-Reception for several reasons. One is a product of the sample size being near double because players only catch about 65 percent of their targets. Additionally, the very-unpredictable Yards-after-the-catch aspect of YPR is removed from the equation.

The most appealing thing about aDOT is that it’s very predictable. The graph you see shows the correlation between a player’s weighted and normalized aDOT from recent history and compares it to actual normalized aDOT data from the next season.

The obvious next question is “How and why are you normalizing aDOT?” All I’m doing there is simply removing the impact of the quarterback from the player’s aDOT. As we’ll see later, there’s a big difference between a high aDOT with Tim Tebow and one with Colt McCoy.

On our graph, the Vertical axis the player’s weighed normalized aDOT from the previous two seasons. The Horizontal axis is the next season’s normalized aDOT. We see a correlation of .95 here, which means that normalized aDOT is absurdly predictable. Essentially it means that you can predict n-aDOT with 95 percent confidence.

r-squared comparables:

Normalized aDOT

0.95

aDOT

0.55

Yards-Per-Reception

0.41

Yards-Per-Target

0.27

Next we have a few comparables. Regular-old aDOT (.55) is very predictable compared to the popular YPR (.41) stat, but not near as predictable as when it’s normalized (.95).

That was a lot of math, but the key item to remember here is that the stat is more valuable than its competitors because it’s predictable year-to-year. That’s hard to find in football statistics.

2011 Top 12 WR/TEs

Now the fun part. Now that you understand what aDOT is and why it’s valuable, we can take a look at some historical data and see how it translates to this upcoming season.

Here are the top 12 WR/TEs from last season

Rk

Player

Target

aDOT

1

Torrey Smith

99

19.6

2

Denarius Moore

73

18.5

3

Vincent Jackson

110

18.3

4

Devin Aromashodu

78

17.4

5

Sidney Rice

56

17.1

6

Malcom Floyd

68

17.0

7

Mario Manningham

94

16.3

8

DeSean Jackson

95

15.7

9

Robert Meachem

72

15.4

10

Eric Decker

92

15.4

11

Demaryius Thomas

81

15.3

12

Jacoby Jones

58

15.2

Note that over the last four years, balls thrown 13+ yards down field were caught 45 percent of the time, compared to 73 percent on balls under 13 yards. aDOT is 21.1 yards on the long balls and 3.7 on short throws.

We see Eric Decker tenth on this list, but it’s a bit misleading, as Tim Tebow – more on him later – likes to throw it deep. Decker’s aDOT on 37 targets with Kyle Orton was 8.8. With Tim Tebow, it was 19.8 on 55 targets. Similarly, Demaryius Thomas had a 9.1 aDOT on 35 targets with Kyle Orton in 2010, but was at 15.0 with Tim Tebow in 2011. That is a fine example of why I normalize aDOT and those splits certainly need to be considered when projecting future data.

Vincent Jackson, Malcom Floyd, and Robert Meachem should also jump out. Jackson and Floyd have spent the last few seasons with the Chargers. With Jackson out, Meachem figures to take on big chunk of his workload. There should be no doubts that Meachem will see a ton of deep balls, as no one provides more long balls to his wideouts than Philip Rivers.

These guys are ideal targets in non-PPR, TD-heavy, and Draftmaster leagues. They will generally catch fewer balls, but they’ll put up big yardage and touchdown numbers.

2011 Bottom 12 WR/TEs

Here are the bottom 12 WR/TEs from last season.

Rk

Player

Target

aDOT

117

Percy Harvin

118

5.9

116

Dallas Clark

63

6.1

115

Hines Ward

62

6.2

114

Brandon Pettigrew

122

6.6

113

Nate Burleson

115

6.9

112

Preston Parker

60

6.9

111

Ben Watson

69

7.1

110

Brent Celek

94

7.1

109

Aaron Hernandez

134

7.1

108

Dennis Pitta

64

7.3

107

Austin Collie

92

7.5

106

Wes Welker

193

7.5

You’ll notice that six tight ends show up here and that makes a lot of sense, as they will usually focus primarily on underneath routes. Not coincidentally, the six wide receivers shown spend a lot of time in the slot. Percy Harvin and Aaron Hernandez also work out of the backfield on occasion.

These guys are going to be better targets in PPR leagues and are better bets for consistent production. Although Austin Collie and Percy Harvin are exceptions, wide receivers with low aDOT’s generally won’t score a lot of touchdowns. Tight ends don’t fit that bill as they are targeted more often in the short field, especially inside the redzone.

aDOT > YPR

Here we’re going to use a made-up scenario to explain the advantage of aDOT over YPR.

Wide Receiver A

Target

Reception

Depth

Yardage

1st Quarter- 12:15

1

1

-1

80

2nd Quarter – 8:57

1

0

5

0

2nd Quarter – 4:14

1

0

-2

0

4th Quarter – 13:19

1

0

1

0

Total

4

1

3

80

aDOT =

0.8

YPR =

20.0

YPR Proj. =

??

Wide Receiver B

Target

Reception

Depth

Yardage

2nd Quarter – 7:54

1

1

22

28

3rd Quarter – 14:14

1

1

8

9

4th Quarter – 9:14

1

1

14

30

4th Quarter – 1:19

1

1

12

13

Total

4

4

56

80

aDOT =

14.0

YPR =

20.0

YPR Proj. =

??

Wide Receiver A has seen four targets in this game, but none were deeper than five yards down field. He caught only one of those passes – a bubble screen – and took it 80 yards for a touchdown. At the end of the day, he has a YPR mark of 20.0, but an aDOT of just under one yard.

Wide Receiver B also saw four targets, but caught all four for 80 yards and also has a 20.0 YPR. He, however, has a 14.0 aDOT.

Simply looking at YPR, these players both appear to be big play threats. Looking at aDOT, however, we see that’s not the case. WR A will catch more passes going forward and his YPR will take a big dive. WR B, on the other hand, is a mid-range/deep target and won’t keep up a high catch rate.

And – for a real life example – I’m sure you saw Andrew Luck’s first preseason pass – a short dumpoff to Donald Brown, who took it 63 yards for a touchdown. Brown’s YPR in that game was 63.0. His aDOT? 3.0. Which is more indicative of his role?

And that leads to my final slam on YPR. The NFL season is only 17 weeks long so one big play can severely inflate data for weeks on end, even while it’s regressing. aDOT gives you an accurate look at a player’s usage right off the bat, while WRA from above will show on stat pages with an inflated YPR for weeks.

Top aDOT vs. YPR

Next, we are comparing the players from 2011 who had higher aDOTs than YPRs.

Rk

Player

Targ

aDOT

YPR

Diff

1

Eddie Royal

56

12.5

9.5

2.9

2

Mario Manningham

94

16.3

13.7

2.6

3

Torrey Smith

99

19.6

17.3

2.4

4

Jonathan Baldwin

51

14.3

12.1

2.2

5

Sidney Rice

56

17.1

15.1

2.0

6

Eric Decker

92

15.4

13.9

1.4

7

Brandon Lloyd

144

15.2

13.8

1.4

8

Titus Young

89

13.7

12.3

1.4

9

Legedu Naanee

74

11.9

10.6

1.3

10

Golden Tate

54

11.8

10.9

0.9

11

Damian Williams

89

13.4

13.2

0.2

12

Ed Dickson

92

10.0

9.9

0.1

As we’ll see in the next chart, we’re analyzing 117 players here and only 13 (or 11%) had a higher aDOT than YPR.  These are guys you can generally expect to see some regression from the next year; although quite a few of these guys changed teams, so their role may change as well.

A good example is Titus Young. His 12.3 YPR was 1.4 yards lower than his 13.7 aDOT. He figures to see additional work this year and will run a fair share of deep routes. Expect a larger YPR.

Bottom aDOT vs. YPR

Our next slide is the same concept, but it’s the guys with YPRs severely higher than their aDOTs.

Rk

Player

Targ

aDOT

YPR

Diff

117

Preston Parker

60

6.9

13.9

-6.9

116

Jared Cook

74

9.1

15.5

-6.4

115

Brent Celek

94

7.1

13.1

-6.0

114

Jordy Nelson

99

12.8

18.3

-5.5

113

Percy Harvin

118

5.9

11.2

-5.3

112

James Jones

56

11.7

16.7

-5.0

111

Lance Kendricks

53

7.5

12.6

-5.0

110

Early Doucet

89

7.9

12.8

-4.9

109

Demaryius Thomas

81

15.3

20.2

-4.9

108

Rob Gronkowski

143

9.9

14.8

-4.9

107

Victor Cruz

151

12.7

17.5

-4.8

106

Wes Welker

193

7.5

12.3

-4.8

You see names like Rob Gronkowski and Brent Celek here – both guys very good after the catch. Victor Cruz and Wes Welker also make the list, but both had 99-yard receptions last season – yet another area where aDOT is more indicative of a player’s role than YPR.

Looking forward, it’s fair to expect a drop in YPR from a handful of these guys including Jordy Nelson, Demaryius Thomas, and Cruz.

Quarterback Top 12

Finally, I wanted to point out that aDOT can also be used to analyze quarterbacks. Shown here are the top-12 quarterback average depth of throw numbers from 2011.

Rk

Quarterbacks

Aimed
Throws

aDOT

1

Tim Tebow

286

13.3

2

Matt Moore

328

10.4

3

Carson Palmer

312

10.3

4

Eli Manning

698

10.1

5

Cam Newton

494

10.0

6

Joe Flacco

568

9.8

7

Ben Roethlisberger

529

9.8

8

Christian Ponder

264

9.6

9

Philip Rivers

531

9.6

10

Rex Grossman

417

9.5

11

John Skelton

262

9.4

12

Sam Bradford

324

9.4

I used Broncos wide receivers as an example earlier and this chart shows why. Tebow’s aDOT was absurdly high at 13.3, almost three full yards higher than second place Matt Moore’s. I also discussed Giants and Chargers wideouts earlier, so it shouldn’t be a shock to see Eli Manning and Philip Rivers here.

Sam Bradford is an interesting name to note. Bradford had a 6.8 aDOT in 2010 which is embarrassingly low. He was up to 9.4 in 2011. More on that in a second…

Quarterback Bottom 12

Here are the quarterbacks with the lowest aDOT marks last season.

Rk

Quarterbacks

Aimed
Throws

aDOT

34

Josh Freeman

519

7.4

33

Colt McCoy

434

7.8

32

Ryan Fitzpatrick

544

8.1

31

Alex D. Smith

463

8.1

30

Tony Romo

497

8.1

29

Blaine Gabbert

381

8.1

28

Drew Brees

730

8.2

27

Matt Hasselbeck

490

8.3

26

Kevin Kolb

226

8.4

25

Curtis Painter

224

8.4

24

Kyle Orton

229

8.5

23

Mark Sanchez

506

8.6

You may notice that the list of guys on the low list is a bit underwhelming. Aside of Brees and Romo, none of these guys really jump off the page. Brees’ low aDOT really makes his record-setting season that much more impressive.

Bradford’s data from earlier can be used to offer some insight as to what we could see from Josh Freeman in 2012. Freeman had a 9.7 aDOT in 2009, a 9.8 mark in 2010, but took a dive to 7.4 last season as a product of lacking a legit deep threat. In 2012, he’ll have Vincent Jackson running deep routes, which allow a bounceback season in the aDOT department. Bradford, on the other hand, didn’t have a strong deep threat in 2010, but took advantage of Brandon Lloyd last season. With Lloyd now in New England, anticipate a drop in aDOT for Bradford in 2012.

As predictable as aDOT is for pass-catchers, the sample size isn’t as great for quarterbacks. It’s important to take personnel into account when projecting aDOT for passers.

aDOT’s Upgrades and Downgrades

Based on what was discussed here today, listed below are a few guys who benefit and a few who take a hit. Note that this isn’t necessarily a list of players you should target/avoid; rather it’s a heads up on their potential changes in value.

Going Up

Torrey Smith / Denarius Moore – Both can maintain high yardage numbers thanks to high aDOT numbers.

Robert Meachem – Proven deep threat entering offense that supports deep threats

Josh Freeman – Expect more deep throws with Vincent Jackson in town.

Titus Young – YPR will regress as reps increase.

Going Down

Eric Decker / Demaryius Thomas – With Peyton Manning, there’s still a ton to love, but don’t expect the massive YPR numbers of 2011.

Victor Cruz – I still like this guy a lot and think there are reasons he’ll maintain WR1 production, but his YPR figures to take a hit.

Follow Mike Clay on Twitter: @MikeClayNFL

  • qpontiac

    Mike, Loving this and I’ve really been focusing on aDOT in my cheatsheets.

    Question# 1 In Excel, I’ve manually input from the PFF Draft Guide the 2009-2011 aDot for QB and WR and I’m trying to find a projection formula from the prior two seasons to predict aDOT for 2012. I’m confused, what is the difference between Normalized aDOT and just aDOT? What am I missing? I have the data in excel already, I’m just missing the calculation I suppose?

    Question #2: What to do with Matthew Stafford? Obviously if I simply take 2009-2011 and just average the 3 seasons, his aDOT and C% appear to be low because 2009 & 2010 are bring his numbers down. In an instance like Stafford, what “weights’ for lack of a better term should I place on 2011, then 2010, then 2009? I think we can all agree we project Stafford (if healthy) to play closer to his 2011 peripherals.

    Question #3: What magic mushroom if any id there to project RB for 2012?

  • http://www.profootballfocus.com Mike Clay

    1 – The difference is explained in the article. The impact of the QB is removed from aDOT to create normalized aDOT. My projection formula is complicated. It uses a weighted mix of career aDOT with the previous year to project the following year. Also, that’s just to get n-aDOT. From there, you need to convert to aDOT by considering the passer

    2 – Again, I definitely do apply weights, but it’s certainly debatable how much to lean on each year. IT really varies from player to player.

    3 – Eh, I use a weighted YPC, but there’s not a lot that’s predictable for backs/YPC.

    • qpontiac

      OK thanks a bunch Mike, I re read the article. So basically there is a lot more to just looking at 3 years of aDOT to accurately project aDOT for 2012, and the formula is proprietary info?

      Not sure if you’re familiar with Tom Tango and his Marcel Projection System (made for Baseball), but I am attempting to use the Marcels to possibly project with some accuaracy 2012 stats? Maybe I’m just wasting my time. I dunno. Great work though!!

      • qpontiac

        What n-aDOT% will I have with a basic weights calculation of the prior 3 years with a slight age regression formula in as well?

        If I’m a pain, and youre too busy to respond, totally understand by the way…

        • http://www.profootballfocus.com Mike Clay

          What do you mean? What would r-squared be? You’d need to run that regression analysis. I obviously couldn’t do that calculation in my head.

      • http://www.profootballfocus.com Mike Clay

        Right…n-aDOT predicts n-aDOT…but our goal is to find yardage, not n-aDOT. So we need to turn n-aDOT into aDOT (applying the QB) and, from there, turn aDOT into YPR.

  • kreardon

    Mike, in your article on Rotoworld you showed statistics for players based on where they lined up (i.e. Aaron Hernandez on the line, in the slot, etc). Is this data available on a large scale on PFF? I have a Gold membership, but have not seen this…am I missing it?

    • http://www.profootballfocus.com Mike Clay

      No, but we do reference it often and it’s laid out for most offensive players in the 2012 Draft Guide (2009-11 data)