ADP vs Player Performance Part 2

Slide2

 By John Bush

ADP vs Player Performance Part 2

Welcome to part 2 of my study from 2013 to 2017 ADP vs Player Performance Part 2 on the journey from before to after the season. What does the ADP really guarantee a drafter? Is it reliable? What positions are more predictable by the ADP?

See part 1 _ Link

ADP vs Player Performance Part 1 http://www.fakepigskin.com/?p=36752

———————————————————————————————————————

Remember that this chapter is from my kindle textbook of 1700 pages of original investigations and rankings with risk analysis. More than a list or a few sprinkle sure things, busts, or sleeper players

Slide1

Winning Dynasty and Redraft Fantasy Football Drafts_May 30th 2018 Version: First of Three 2018 Preseason Drafting Textbook Versions.

By John Bush and David Cherney Kindle Edition 

scheduled update in early July if you wish to wait!


Figure 6-7 A and B. Average of EOS Performance by Position and Positional ADP.

Quarterbacks

The positional ADPs based EOS PF averages are shown in Figure 6-7 Part A and B. Starting with the QB position the trend is a similarly high EOS PF into the 7th QB drafted with 8th to 10th generating a few PF points less. The 11th QB to 17th QBs historically can yield only a 2/7 failure rate.  That is a “late “QB model and this data would support that draft strategy.

Tight Ends

The TE position is much different in pattern from QBs. The top TE only gives terrific value followed by a lowered EOS PF average in the same level into the 15th TE all at the 65 EOS PF level. The 2018 strategy for a late TE drafting should be modified to the state to take the 1st TE or wait very late!

Running Backs

The RB position shows the top 3 RB are in the same zone (73 EOS PF). The following RBs 4th to the 14th RB were below the top 3 RB, but most are at the same level (57 EOS PF median).

The next 18 RBs were 8 of 18 RBs that were above 50 EOS PF. Clearly a drop (4/9ths) but getting about half at 50% ish level chance of an ok RB into the 32nd RB!

The RB Draft Plan would be to grab within the top 4 or wait into the 12th RB to acquire an RB or two and 1 or more into the 32nd RBs.

Wide Receivers

Finally, looking into the WR position, we see the first WR is at EOS PF of 97 and the second WR is at 88 PF! They stand above the rest.

The next 4/5 WRs (3rd to 7th WRs) are almost all at 76 PF. After the 7th WR to 12th WR, we are at 71 PF. After the 12th WR to the 24th, the median is at 58 PF. The next 12 WR 25th to 36th WRs are at a median of 50 PF. The WRs are very slowly dropping but a smaller rate than the other positions at these late draft picks.

The draft plan included after the early WRs you are going to have nice chances for above average WRs into the 36th slot. It seems that 3rd to 7th WR are the same player and 8th to 12th are also the same player. If you miss the first 2 wait to close to the 7th WR on average and them closer to the 12th WRs etc.

ADP vs Player Performance Part 2 slide136.jpg

ADP vs Player Performance Part 2 Slide14


Figure 6-8. Graphical Plot of Positional EOS PF Average vs Positional ADP

The figure 6-8 below plots the data from 6-6 and presents linear forecast trend lines for positional comparisons.

The main takeaway is in the trendlines of RB vs WR. On average across the entire range, the WRs are producing at 10% or so higher levels vs RB into the 69th players

Thus in the draft, WR will yield 10% higher fantasy points into the 69th player level. If 2 players are tied in your rankings, the WR will be the tiebreaker on average.

ADP vs Player Performance Part 2 Slide15


Figure 6-9 A and B.  Overall Average of EOS PF vs ADP Round with Value Analysis.

We collapsed the positional data and ADP data into groups of ADP rounds to get the entire landscape. The EOS PF starts at 73 and moves downward by rounds to 43.4.

We see a sharp EOS PF drop after round 6 (73 to 58 EOS PF Rounds 1 to 6) and that drop slows from the 7th round to 15th round (51 to 43 EOS PF).

The plot shown in part B tracks the EOS PFs vs ADP. I converted the EOS PFs into a Value Scale of  100 to 0. The entire dataset gives a steeper drop than looking at the data from the 7th round downward. Note, those rounds have the TE and QB thrown out.

The take-home in the Value supports the 3 rounds as being high (above 90% in Value) followed by player value near 78 % in the next 3 rounds.  The 7th round player value is at 70 and that drops to 11th within a median value of 66.7 %. Finally, in rounds, 12th to 15th plus we on average have players of  59% value.  Thus a sequence of 90 to 78 to 66.7 to 59% player value over the 15 rounds.

This data support the critical nature of the first 3 rounds. On average if you get 1 player in each of those round, that would make your team worth 283.9 value points.

If you can draft players above that value level then the next 3 round you only need average. (less pressure more success?)

If you have a failure in your top 3 then you have to be able to pull several players hitting way above average in rounds 4 to 6. (high pressure less success?)

We can live or die by 1 or 2 players missing the average in the top 3 or 6 players drafted.

ADP vs Player Performance Part 2 Slide16


ADP vs Player Performance Part 2 Slide17


Figure 6-10 A and B. Yearly Breakdown of EOS PF vs ADP Rounds

Does the data vary by year? If we see interesting variation, then 2018 might be biased by the 2017 results. In 6-10 Part A, we present the ADP Rounds vs the Year for all positions combined! We applied colorization to highlight the highs and lows in the EOS PF averages within the rounds.

Looking at the figure of the ADP Draft rounds and the colorization, we can conclude that 2017 was a very different year vs the preceding years. The first four 2017 draft rounds produced nice returns! The median was 71 EOS PF vs 2016 (63), 2015 (64), 2014 (70) and 2013 with a 4 year 64 EOS PF 4 round median score. Another way to say it is 2017’s 4 ADP Rounds were 10% above norms!

Note the high returns in 2016 from rounds 6 to 14th round! That year had a median of 55 as to say 2017 which returned a median of 48.  2016 also had the 5-year highs in rounds 6, 8, 10, 11,12,13 and 14. Note also that 2015 was the best year for ZERO RB in general at 65% int he first 2 rounds.

2017 was such a great year for the first 4 round player holding their value. Can 2018 continue? The regression to the mean would suggest not so!

ADP vs Player Performance Part 2 Slide18


Part B shows the graph plot of the table in Part A as well as overall linear forecast trendlines. Analysis of the trendlines highlights that 2016 was a strange year and players have more value leading to a lesser slope in that year’s trendline. The remaining 4 years were much closer together in similar trendlines with noting the early 2017 high values.

If you save your data plot your drafting vs EOS PF and see where you fall in the trends!

ADP vs Player Performance Part 2 Slide19


Figure 6-11. Meta Data of All Five Years by Position and ADP based Draft Rounds.

The table presented in 6-11 lists the positions by columns and has been colorized by columns to highlight the highs and lows of the EOS PF averages by draft rounds! Note that in the first round of 12 players the QB position returned 87 while the TE returned 70? That is due to low sample size for any TE to be drafted in the top 12. WR came in at 81 vs RBs at 69. We have already shown the extreme nature of the early 1st round RBs vs back half of round one RBs.

The QB positions hold value into the 6th round and TE into the third round of the drafts. WRs maintain the highest value staying above 50% to the 7th round vs RBs only to the 4th round. WRs are worth more in value later in the draft.

ADP vs Player Performance Part 2 Slide20


Figure 6-12 A and B. Scaled to Average Positional EOS PF values with Graphs and Trendlines

Part A of Figure 6-12 has the calculated values vs positional Highest PF Generated. For example, in the EOS PF QB column, the highest EOS PF was 90.5. Thus I subtracted that number from all the QB EOS PFs leaving a range for highest 0 to -27.2 in QB round 14. The metric from a data transformation gives a landscape view of each 5 years averaged ADP round of QB EOS PF values vs the top. You can quickly focus on these meta-patterns in value change vs ADP rounds!

What is the change in the EOS PF averages within the position as compared to the positions overall 15 round average? The entire table was colorized to show the hot spots for the positions.

The QB position Figure 6-12 A, shown in the first column supports the later QB drafting model until round 6. The drops are very small and even after round 6 the drops are not dramatic in a typical 12 team PPR draft!

The RBs has a drop by the 2nd round by 8% down and 12% by round 3 and after 5 rounds the drop-in value sits at 30% or less. That, however, plateaus outward into the 15th round. There is a bottom floor for RBs and they reach it as an average in the 6th round.  That fact does encourage looking for deeper RBs in those later rounds as you on average have reached the bottoms.

WRs drop by 9% into round 2 and to 14% by round 3. After the 3rd round, the WRs are return -22% into the 6th round. The 7th round starts at about 30% down and that plateaus at round 10 at the 40% less level. Looking for deep WR plays is supported in the 10th round and out!

In the TE we see that the by 3rd round you have lost 18% value and that quickly doubled by the 4th round into a 38% drop. The concept of later TE drafting is that the after the first few it’s all the same on average and no use in rushing a TE pick. The Barbell approach of early or late TE drafting might be supported by this data!

ADP vs Player Performance Part 2 Slide21


Figure 6-12 Part B gives a trendline view of the positional scaled values. These data confirm the discussion above. Part C displays the landscape view of each position separately for further analysis. Note the least slope is found in the QB position vs the WRs have the greatest slope drop.  WR and RB are paired vs TEs vs QBs. In some ways the QB and TE need separate analysis.

  ADP vs Player Performance Part 2 Slide23


Figure 6-13 A, B, C and D. Players Drafted with 15 Rounds vs Not Drafted.

Are there worthy players that are left out in a typical 12 teams PPR draft?

Is there a position that has more chances of finding a deep undrafted sleeper?

These are two viable questions to address.

The 4 Parts of Figure 6-13 contain these answers and are an interesting bit of knowledge for 2018 drafts.

The data in 6-13 Part A shows the grand 15 round positional EOS PF averages vs all players left on the table (deep sleepers?). Analysis of this tabular data reveals that the QB position has players that are not drafted but are higher in value than RB, TE, and WRs. 36% higher value for non-drafted QB vs 20% value in non-drafted RB, TE, and WRs.

Take-home is that you have a better shot at finding a significant non-drafted QB. There are not a lot of bargains in the cheap seats! We fantasy football players have a strong wisdom of the crowd going and we as a group draft every year well.

ADP vs. Player Performance Part 2 Slide24


Part B is a plot of the Part A data and gives a visual view of this data. We need however to normalize the data to see a more accurate view of these questions and answers.

ADP vs Player Performance Part 2 Slide25


In part C we normalized the data and set the drafted EOS PF average to the 100% level. We then calculated the % of that PF number for the non-drafted players by positions. These results are shown in Part C and graphed in Part D.

ADP vs. Player Performance Part 2 Slide26


ADP vs. Player Performance Part 2 Slide27


The non-drafted QBs were worth 48% vs the RBs at 36%, TE at 35% and WR at 39%. There is a difference between the QB and the rest of the positions by 10% or more but the other three positions in the non-drafted players were very close.

Thus, if you randomly added a non-drafted player they would be worth in QB’s nearly 50% vs 38% of the others on average. That is a loss of 62% of value for the 3 and 50% for the QB. 6-13 Part D has the plot for a visual aid!

 

 Part 3 continues this research and it should be out this later week as well. Thanks for reading

***** Previous Articles Please Enjoy

More Information Increasing Accuracy?  http://www.fakepigskin.com/?p=35319

Redraft ADP Pattern Analysis 5/9/18   http://www.fakepigskin.com/?p=36403

Non-PPR Player Rankings Part 1  http://www.fakepigskin.com/?p=36513

Non-PPR Rankings Part 2  http://www.fakepigskin.com/?p=36541

PPR Rankings Part 1   http://www.fakepigskin.com/?p=36549

PPR Rankings Part 2   http://www.fakepigskin.com/?p=36554

Fantasy Football Planning  http://www.fakepigskin.com/?p=36562

Why Late QB Drafting?  http://www.fakepigskin.com/?p=36601

Best at Best Ball Part 1   http://www.fakepigskin.com/?p=36666

Best at Best Ball Part 2   http://www.fakepigskin.com/?p=36677

Best at Best Ball Part 3   http://www.fakepigskin.com/?p=36708

Best at Best Ball Part 4   http://www.fakepigskin.com/?p=36726

Best at Best Ball Part 5   http://www.fakepigskin.com/?p=36734

FF Players Biased Decisions Part 1 http://www.fakepigskin.com/?p=36738

FF Players Biased Decisions Part 2  http://www.fakepigskin.com/?p=36743

 

 


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>