Tuesday, May 6, 2014

Draft drop-off by position


Let’s say you are an NFL team and you are 100% set on drafting a DE. What do you do if the team in front of you picks the consensus number one DE on the board? Seems like it would be helpful to know what the average drop-off is for each position…

A note on the data here – I used data from Pro-Football-Reference.com on all draft picks from 1994-2009. The 2009 endpoint allows for 4 years of performance data from each draft pick since 4 years is the current level of team control and it is much easier to do this with a nice, clean endpoint.[1] 

Table: Production by Position 

DB
LB
DE
DT
C
G
T
TE
WR
RB
QB
1
26.9
29.6
26.9
27.8
17.9
23.3
30.1
24.4
26.1
34.6
30.5
2
28.0
28.3
22.0
20.2
14.0
14.6
25.1
15.4
26.4
24.5
22.0
3
21.9
23.6
21.6
14.2
11.4
15.8
20.5
10.8
18.8
21.5
16.2
4
16.8
20.8
14.3
12.8
6.4
11.6
20.8
10.8
22.9
20.9
10.8
5
19.1
21.8
15.3
10.4

6.6
14.1
10.0
16.2
18.4
8.4
6
12.1
18.9
17.3
8.9

9.6
18.8
9.6
17.4
17.0
5.9
7
19.0
14.3
14.0
11.2

7.3
16.4
8.0
14.6
10.9
6.6
8
14.1
15.4
6.4
8.9

9.0
10.5
6.8
13.1
14.1
1.4
9
18.4
8.3
14.0
5.5

7.7
7.2
7.3
13.1
8.3
5.1
10
16.4
11.4
9.6
9.8

6.5
9.6
2.4
13.1
14.8
1.7
11
16.1
10.4
11.6
9.5

5.7
10.8
5.3
11.8
8.7
2.6
12
16.4
11.5
14.5
4.1

3.2
5.3
4.5
9.9
3.9

13
15.4
10.8
4.6
6.7

7.0
5.1
6.6
7.4
6.3

14
11.8
11.4
5.9
2.5


4.6

10.9
5.2

15
12.3
10.6
6.5
5.7


4.9

9.2
6.4

16
7.9
10.9
4.1
7.9


6.9

5.8
4.4

17
6.5
5.8
2.9
6.9


3.1

5.5
4.1

18
8.5
6.5
3.7
4.6




3.7
3.4

19
9.1
7.6
3.4





5.6
3.3

20
6.6
6.1
4.9





6.1
2.8

   
This table shows the average AV produced in the first four seasons by the 1st through 20th pick at each position from the 16 drafts from 1994-2009. Any position/rank combination with 12 or fewer observations was excluded (e.g., only 9 drafts have seen a 15th QB drafted).

As you can see on here there are some interesting patterns by position. DBs and WRs, notably, show higher averages for the second player taken than for the first. Looking further down the DBs below the first couple seem relatively similar all the way through 13-15.

This table, however, doesn’t have enough information to give us the whole picture.

Table: Average Draft Position
DB
LB
DE
DT
C
G
T
TE
WR
RB
QB
1
9.4
8.9
7.2
10.4
46.1
24.1
5.3
23.8
9.4
8.1
6.7
2
13.4
17.9
11.8
24.0
76.9
45.3
15.4
40.3
13.9
16.6
20.8
3
19.3
24.6
16.6
32.8
128.9
57.1
25.3
60.5
19.6
29.1
47.1
4
25.3
33.2
24.4
52.9
147.7
74.8
37.3
77.3
29.8
39.5
69.8
5
29.8
43.9
36.1
66.7

96.4
47.8
94.5
37.4
45.6
104.1
6
33.6
51.1
46.5
82.6

115.1
58.9
108.8
43.7
59.5
128.6
7
38.3
56.5
58.3
94.5

129.8
72.3
127.0
48.9
73.4
146.4
8
43.1
63.7
69.7
113.1

146.3
87.3
140.4
56.6
81.3
166.0
9
47.5
71.7
81.4
122.3

150.7
105.2
160.1
65.6
91.8
171.5
10
53.2
78.6
93.6
129.1

170.1
118.0
174.4
73.6
97.9
182.0
11
56.4
86.8
107.9
143.4

180.6
133.5
181.3
81.4
107.5
202.2
12
59.7
91.8
119.9
155.0

195.5
144.9
196.1
89.4
118.1

13
63.4
100.8
135.7
157.1

202.2
155.8
203.4
98.2
128.8

14
69.3
105.3
146.8
169.5


173.0

106.3
136.3

15
74.6
112.3
150.9
184.7


184.4

114.6
145.9

16
79.0
117.4
164.8
194.4


182.8

124.1
159.3

17
84.5
126.4
177.3
199.5


191.2

132.3
169.8

18
89.6
133.1
190.6
210.6




143.5
181.1

19
94.7
141.1
200.8





150.1
192.6

20
98.9
147.2
209.9





157.4
193.2


 
Part of the reason that those DBs are so close together is that they’re drafted close together. The 10th DB comes off the board at the 53rd pick while the 10th tackle doesn’t get picked until 118th.

I know what you’re thinking – it would be nice to know the over/underperformance against expectation for each cell up there – but hold onto that thought, that’s tomorrow’s post.

Getting back to the hypothetical I laid out to start – what does the drop-off look like from the first to the second player taken once we account for the gap in picks between them?

If you imagine a line starting at (0, 1) with a positive slope to indicate decreasing value for the subsequent pick, you can quickly see that several positions offer a much lower premium on the first player taken. WRs and DBs, as noted above, are below the line regardless of where it actually falls[2]. Tackles, despite being in roughly the same place as DEs and RBs, appear to have a much lower premium on the first player off the board.  

Top Player

The top player for each position fits with the pattern established above, but provides what I think is a simpler way to look at the distribution of elite value.

Table: Count of top performer at each position per draft year (of 16) by order taken
DB
LB
DE
DT
C
G
T
TE
WR
RB
QB
1
3
2
8
9
6
7
8
8
3
7
7
2
5
4
1
4
2
2
1
4
4
2
1
3

1
2
1
3
3
1


1
4
4
1
1
1

2
2

1
2
2

5
1
1


3
1

1
1

1
6

1
1
1


1
1
1
2
2
7
2
2

1

1
1

1
1

8

1






1


9
2

1




1


1
10






1




11






1

1


12


2








13

1






1


14











15











16











17











18






2




19

1









20











Other
2
1






1
1

 
It’s clear at a glance which positions have a tight concentration of elite players. Those that have a wider distribution have something inherent to them that is either that much harder to scout or results in luck being relatively more important than skill compared to other positions.

As teams refine and finalize draft boards ranking players at different positions against each other perhaps some will reflect that uncertainty.



[1] I know there are options but they aren’t worth very much. As happy as a player is to be franchised, they’re likely to be 2-3x that happy to have their option picked up. It’s leverage for the teams but not realistic to calculate value here – by that principle we would just look at players’ entire careers as my draft value chart does. ESPN’s Kevin Seifert covered this in greater depth this week (see here).


[2] It would differ for each pair of picks because they have different relative values (e.g., the 9th to the 13th pick for DBs vs. the 46th to 77th pick for Cs)
Posted by on 5 comments:
Labels: , ,

Monday, May 5, 2014

Looking for evidence of better drafting


Chase Stuart put up a post last week examining whether (and why) teams had gotten any better at drafting. He looked at the proportion of value delivered by top picks relative to all picks in the draft (i.e., can teams identify the value in each draft) and at the correlation between draft position and career value from year to year (i.e., does where a player was picked predict career value).

Given that Chase examined the relative ability of teams to pull value early in the draft, I wanted to see if drafted players in general changed over time.

To see if I could tease out any improvement in drafting I looked through my draft database from 1994 to 2012 (starting with the implementation of the salary cap).


Read more »
Posted by on 3 comments:
Labels: ,

Beating the Chart: Where to get the best value in the NFL Draft


The NFL Draft has been a focus of mine for the past couple of years. I did some analysis to get at the real value of each pick in the draft and create an updated version of the draft value chart used by NFL teams.

Along with this chart there were some basic guidelines for doing well against users of the old-school chart popularized by Jimmy Johnson: trade down. The older chart valued top picks – especially the top two or three – much higher than the rest of the draft. The disparity continues throughout the draft as each round is worth less relative to the previous one than my analysis would suggest, even after subtracting out the replacement value (an undrafted free agent).  


The older chart drops off by more than 50% each round relative to the previous one. The Sports + Numbers version still has a steep drop off from the first to the second round – roughly 50% – but subsequent gaps are 40% and then 30% for later rounds.
It’s important to note that this does not suggest that teams should expect immediate starters from later rounds.


This figure shows the percentage of players drafted in each round who start in their first, second and third season. Although UFAs are shown on the graph for reference, the denominator for them is players on the roster and excludes players signed for a tryout and subsequently cut.

Read more »
Posted by on 35 comments:
Labels: , , ,
Subscribe to: Posts (Atom)