2016 Targets Vs 2017 ADP

Slide1

By John Bush and Dave Cherney

Introduction 2016 Targets vs 2017 ADP

The previous 4 articles dealing with targets have been published. See the links below.

2016 RB Targets Article

2016 TE Target Article

2016 WR Target Article

2016 Team Targets Analysis

The remaining question for this preseason is a simple one to pose but not easily investigated.

“Does A Player’s 2016 Targets/Game Average explain their current 2017 ADP?”

If there is a connection then players may be able to use the 2016 data to catch players that are not so correlated. If a player was heavily targeted then should not his 2017 ADP be associated with that past season production? Answering the reverse question is also interesting.

I wanted to explore this question and determine if player could use last year’s data toward this year’s drafts.

“Does Running Back 2016 Targets/Game Average explain the current 2017 ADP Landscape?”

Running backs present the most unique situation in fantasy football. Their fragile nature is an issue and their passing catching ability is highly variable. As an investigator into the question, should RBs are rushing dominate vs others are pass catching dominate. The data as a collection would most likely show the lowest correlations in 2017 ADP and 2016 targets.

The data shown in Figure 1 and 2 presents the player, team, 2016 T/G, 2017 Average ADP and simple ranking within the NFL RB colony. I have color coded high, mid, and low T/G form 2016 using green, yellow and red coloring. (Data does not include rookies fyi)

Scan through the data sorted by 2017 ADP and make note of RBs that seem to be misplaced. For instance Theo Riddick seems lower than he should be in PPR leagues as does Jame White. Later draft targets!

Figure 1 and 2 2017 ADP Average vs 2016 Targets/Game. 

2016 Targets Slide31

Slide32

Correlations in RBs

The data graph shown in Figures 3 and 4 present a scattergram of the the RBs T/G (Y Axis_ vs 2017 ADP (X Axis). Also included is a trendline and R2 value.

A correlation is a single number  (R2 value) that describes the degree of relationship between two data sets. The number ranges from -1 to +1 in value. The -1 value implies a perfect negative correlation and a +1 implies a perfect positive correlation.  A zero value implies no relationship between the data.

Results. 

The May R2 value was 0.05 and in mid August it had moved up to 0.13. These numbers suggest no relationship in 2016 T/G and a Running Back’s ADP. As a group then we must move on to other more fruitful data analysis. The concept of a pass catching RB is still a solid idea and must be included in your drafts.

Figures 3 and 4 Correlation in RB 2017 ADP and 2016 Targets/Game Early May and Mid August 2017

2016 Targets Slide302016 Targets Slide33

Correlations in TEs

Unsurprisingly, we found no correlation in RBs between targets and ADP. Tight ends however, make their fantasy living on catching passes and many near the red zone. we would expect a better association between Targets and current ADP

The data shown in Figure 3 presents the player, team, 2016 T/G, 2017 Average ADP and simple ranking within the NFL TE colony. I have color coded high, mid, and low T/G form 2016 using green, yellow and red coloring. (Data does not include rookies fyi)

Scan through the data sorted by 2017 ADP and make note of TEs that seem to be misplaced. For instance Z Ertz at 7.6 seems undervalued at TE 10 as well as J. Witten, C. Clay and Z. Miller.  Later draft targets for TEs!

Figure 5. 2017 TEs ADP Averages vs 2016 Targets/Game

Slide35

Correlations in TEs

The data graph shown in Figures 6 and 7 present a scattergram of the the TEs T/G (Y Axis vs. 2017 ADP (X Axis). Also included is a trendline and R2 value.

A correlation is a single number  (R2 value) that describes the degree of relationship between two data sets. The number ranges from -1 to +1 in value. The -1 value implies a perfect negative correlation and a +1 implies a perfect positive correlation.  A zero value implies no relationship between the data.

Results. 

The May R2 value was 0.11 and in mid August it had moved up to 0.19. These numbers are much better than the RBs but again suggest no real relationship in 2016 T/G and a Running Back’s ADP. As a group then we must move on to other more fruitful data analysis. The concept of a pass catching TE is still a solid idea and must be included in your drafts. The idea that TEs are not readily judged by last season’s T/G is confusing to me. This surprise will drive me to deepen my research here.

Figures 6 and 7 Correlation in TEs 2017 ADP and 2016 Targets/Game Early May and Mid August 2017

Slide34

Slide36

Correlations in WRs Data Analysis

Unsurprisingly, we found no correlation in RBs but TEs also had no significant correlation between targets and ADP.  We expected a better association between Targets and current ADP than was found and I left with confusion. We finally move to the WR position and again based on their fantasy jobs some correlations are strongly expected.

The data shown in Figure 8 to 10 presents the player, team, 2016 T/G, 2017 Average ADP and simple ranking within the NFL WR colony. I have color coded high, mid, and low T/G form 2016 using green, yellow and red coloring. (Data does not include rookies fyi)

Scan through the data sorted by 2017 ADP and make note of WRs that seem to be misplaced. For instance J Mathews, K White and T Austin are deeper plays as they had 8.3, 9 and 7.1 T/G in 2016.   Later draft targets for WRs!

Figures 8 to 10. 2017 WRs ADP Averages vs 2016 Targets/Game

2016 Targets Slide38

2016 Targets Slide39

2016 Targets Slide40

Correlations in WRs

The data graph shown in Figures 11 and 12 present a scattergram of the the TEs T/G (Y Axis vs. 2017 ADP (X Axis). Also included is a trendline and R2 value.

A correlation is a single number  (R2 value) that describes the degree of relationship between two data sets. The number ranges from -1 to +1 in value. The -1 value implies a perfect negative correlation and a +1 implies a perfect positive correlation.  A zero value implies no relationship between the data.

Results.

The May R2 value was 0.38 and in mid August it had moved up to 0.60. These numbers are very strong and much better than the RBs and TEs.  The movement in correlation was surprising and the nice 0.6 number was excellent.

That finding can be reworded to say that 60% of the variation in a WR’s 2017 ADP can be explained by their 2016 Targets per Game production.

Excellent finding! 

Rookie WRs, C Davis at 49 ADP should receive 6.4 Targets per Game, Z Jones  at 52 ADP getting 6.2 Targets per Game and Kenny Golladay at 58 about 6 Targets per Game. This is a great way to classify rookies in their first year as well. I will work on the future to explain the other 40% of the variation for other hidden data.

Figures 11 and 12 Correlation in WRs 2017 ADP and 2016 Targets/Game Early May and Mid August 2017

2016 Targets Slide37

2016 Targets Slide41

Commentary from Dave Cherney on these findings

Bush Question (1)

Why are the RBs different than WRs in 2016 Target Correlation vs 2017 ADP

Dave Cherney’s Response to Q1

This data is quite fascinating, as I would have thought these numbers would have matched up much closer.

At the top, the first eight are your usual suspects with Brown, Beckham Jr. and Jones leading the way. Mike Evans took a nice jump from the previous year as the connection between he and Jameis Winston had another year to develop. Of this group, you will see practically the same order in Dynasty except for Jordy Nelson.

Our first three mid-outliers are Dez Bryant, Brandon Cooks and Keenan Allen. These can be easily explained. Bryant was a bit banged up most of the season and getting used to a new quarterback. Cooks had never been a target monster in New Orleans, but a change of venue and the loss or Julian Edelman could easily correct that. Lastly, Allan suffered a torn ACL in the first game thus a small sample size. He is back from the injury and should be featured prominently once again.

Tyreek Hill has the largest disparity among the top-20 with a paltry 5.2 targets per game. This was in part due to coming on late in the season, but he still was perceived as a boom or bust play thus the low number of targets is hardly surprising. This season, expect that to climb sharply with the departure of Jeremy Maclin and the rise to the top of the depth chart.

The biggest surprise is the number of targets, 9.4, that Allen Robinson commanded after having such a disappointing season. Finishing at WR 25 in PPR leagues, he had seven games of less than double digit points including a four-game stretch during the crucial weeks 12 thorough 15.

Bush Question (2)

I want your take on why ADP of 2017 TEs does not fit with 2016 TE Targets as measured by Correlation.

Dave Cherney’s Response to Q2

Targets for the tight end position are not surprising that they do not fit this correlation as this will dictate the type of tight end the player is. Of the three types, you have the blocking, the hybrid and finally the exclusive pass receiving.

The first ten contain some of the best pass receiving TE’s in the game and it’s not surprising to see them match up pretty much on the money with the exemption of Rob Gronkowski being on the low end at 4.8. I believe this is an aberration and will most likely be the first tight end to go off either redraft of dynasty boards.

The biggest surprise for me was seeing Kyle Rudolph leading the pack with 8.3 targets but then reminded myself he did place second among the position in PPR leagues. Another reason for the heavy targets was the short passing game Sam Bradford was forced to play behind his horrendous offensive line.

We don’t get our first disturbance in the force until we reach Hunter Henry. However, this can be easily explained as it is twofold. First, he was a rookie last year who almost always struggle their first year. In addition, he was sharing time with Antonio Gates.

Our next blip on the radar come in the form of Austin Hooper who once again, was a rookie. Expect these numbers to rise this year with the departure of Jacob Tamme. I would expect his targets to be in the green come 2018.

From a Dynasty perspective, eight of the top-10 would be among the first to go; the exceptions being Delanie Walker and Martellus Bennett due to their advanced age.

Bush Question (3)

I want your take on why ADP of 2017 RBs does not fit with 2016 RBs Targets as measured by Correlation.

Dave Cherney’s Response to Q3

In PPR leagues, the pass catching 3-down running back is invaluable.

It should come as no surprise that David Johnson (7.5) and LeVeon Bell (7.8) lead the way. They were not only the bell cow backs for their respective clubs, they were featured heavily in the passing game and made many owners extremely happy.

The first six running backs each averaged over three targets per game in 2016 are all considered invaluable. In addition, you will find that except for McCoy, these will most likely be the first running backs going off the board in most Dynasty leagues.

The next two, Elliot and Ajayi came in a good bit under, going 2.7 and 2.3 respectively. While these backs are fantastic in any format, it is nice to have that extra bit when it comes to the passing game. The good news here is each back has been told they will be featured more in the passing game this season as they both have good hands.

Some nice, late round PPR fliers that could have worked on your squad in a pinch last year would have been Darren Sproles (4.7), Giovanni Bernard (5.1), Shane Vereen (3.8), Chris Thompson (3.9) and T.J. Yeldon (4.5).

Bush Questions (4)

I want your take on why ADP of 2017 RBs does not fit with 2016 RBs Targets as measured by Correlation. Why is there a universal movement from May to August to use Target information more?

Dave Cherney’s Response to Q4

The job of a fantasy football player is to be part gypsy; the fortune teller. Reading the tea leaves each individual season remains the challenge to keep you ahead of your competition and get you started toward the championship.

That said, I had never noticed this trend before.

Perhaps fantasy drafters would revisit the real numbers in August as actual drafts approach as opposed to in May when the numbers are being recalled from the back of their minds?

Bush Question (5) Is this an influx of low-info folks?

Dave Cherney’s Response to Q5

As stated above, I’m hard pressed to think of another reason for it. It isn’t until much later in the season that the real pundits pull out the actual data from the following year and sharpen up a more realistic ADP.

Targets are a huge part of rankings, especially in PPR leagues. As it is stated, it starts and ends with opportunity.

Bush Question (6)

Are the ADP from MAY more accurate or less than August? If not using targets from last year what information filled the gaps in for early drafters that generate the May ADP? 

Dave Cherney’s Response to Q6

The closer to the season, the more serious ADP’s become thus I find the August ADP much more accurate than in May. In addition, injuries and situations are very fluid between May and August and thus will affect ADP.

Looking at the number of season ending injuries, or even some bumps and bruises, see Sterling Sheppard, can affect ADP. Change of venue, such as Jordan Matthews to Buffalo will also create large changes in opinion and I saw his ADP jump from the 11th round to the 7th.

Thanks to Dave Cherney for adding his wisdom to my metrics! 

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>