ArticlePDF Available

An Alternative to the NFL Draft Pick Value Chart Based upon Player Performance

Authors:
Michael Schuckers at St. Lawrence University
Michael Schuckers
Download full-text PDF
Read full-text
Download citation

Abstract and Figures

In this paper, we consider the National Football League Pick Value Chart and propose an alternative. The current Pick Value Chart was created approximately 20 years ago and has been used since to determine the value of draft selections for trading of draft selections. For this paper, we analyze the first 255 draft selections for the years 1991 to 2001. As part of our analysis, we consider four non-position dependent metrics to measure and model player performance at each of the first 255 draft selections. We perform a nonparametric regression of each performance metric onto player's selections. A comparison is then made between each fitted line and the Pick Value Chart. Having considered these comparisons, we propose an alternative Pick Value Chart.
Figures - uploaded by Michael Schuckers
Author content
All figure content in this area was uploaded by Michael Schuckers
Content may be subject to copyright.
Content uploaded by Michael Schuckers
Author content
All content in this area was uploaded by Michael Schuckers on Aug 14, 2014
Content may be subject to copyright.
Journal of Quantitative Analysis in
Sports
Manuscript 1329
An Alternative to the NFL Draft Pick Value
Chart Based upon Player Performance
Michael Schuckers, St. Lawrence University
©2011 American Statistical Association. All rights reserved.
An Alternative to the NFL Draft Pick Value
Chart Based upon Player Performance
Michael Schuckers
Abstract
In this paper, we consider the National Football League Pick Value Chart and propose an
alternative. The current Pick Value Chart was created approximately 20 years ago and has been
used since to determine the value of draft selections for trading of draft selections. For this paper,
we analyze the first 255 draft selections for the years 1991 to 2001. As part of our analysis, we
consider four non-position dependent metrics to measure and model player performance at each of
the first 255 draft selections. We perform a nonparametric regression of each performance metric
onto player's selections. A comparison is then made between each fitted line and the Pick Value
Chart. Having considered these comparisons, we propose an alternative Pick Value Chart.
KEYWORDS: National Football League, draft, nonparametric regression, performance, football,
LOESS
1 Introduction
The National Football League (NFL) is a professional league of American Football
based in the United States. Each year the NFL allocates eligible players among
its teams via a draft. In each round, each of the 32 teams selects a single player.
The order of team selections is based upon the previous year’s performance. Teams
can trade their draft selections for players or for future draft selections or for a
combination of these two. Additionally, teams can be given draft picks at the end
of rounds as compensation for losing current players to free agency. In 2010, the
NFL draft had seven rounds and a total of 255 players were selected.
Sometime around 1990, the Pick Value Chart (PVC), or The Chart
(Massey and Thaler, 2010) , was introduced as a way of assigning value to each
draft selection. Figure 1 plots the value of the PVC against a players selection. (A
complete version of the PVC is found in Table 5 at the end of this article.) This chart
allowed teams to have a single currency for determining if a trade of draft picks pro-
vided equivalent value. McGuire (2010) states the that original PVC was developed
by Jimmy Johnson who coached the Dallas Cowboys from 1989 to 1999. Johnson
supposedly used historical trade data to devise the PVC (Smith, 2007) . Recently,
Massey and Thaler (2010) evaluated the PVC on the basis of the value of subse-
quent contracts awarded to players to determine that the PVC overvalues early draft
selections particularly those in the beginning of the first round of selections. Berry
(2001) looked at comparing success rates for first round picks across a variety of
sports where success was defined by making an All-Star team. Others who have
looked at creating updated versions of the PVC include Stuart (2008) and Maier
(2010) though these versions primarily involved minor modifications to the current
PVC. In this paper we show that the PVC is not reflective of player performance and
we provide an alternative chart to the PVC based on past player performance. The
remainder of this paper is organized in the following way. Section 2 discussion the
notation and variables that we will use here and gives univariate summaries of our
performance metrics. Section 3 compared the PVC to other performance metrics.
We create an alternative version of the PVC and discuss it in Section 4. A summary
of the work here as well as a discussion are summarized in Section 5. An appendix
that contains a copy of the PVC as well as our alternative is found in Section 6.
2 Data
The data that we will use to evaluate the PVC comes from
Pro-Football-Reference.com Pro Football Reference (2010) . We are using
every player selected among the first 255 players in the NFL drafts that occurred
1
Schuckers: Alternative NFL Pick Value Chart
0 50 100 150 200 250
0 500 1000 1500 2000 2500 3000
Selection
Value
Figure 1: Pick Value Chart
from 1991 to 2001. Selection of these particular years was based upon a tradeoff
among the need to have recent selections to reflect current trends in players, the
need for sufficient samples sizes at each selection and the potential bias from in-
cluding large numbers of players who are still active. Since we are going to predict
draft value, we used the length of the 2010 NFL draft, 255 selections, as our tar-
get length. Along with the name of the player selected we have several variables
for each player including their selection number — the order in a given draft that
they were selected, the team that selected the player, and the position of the player.
Further we have a set of performance measures for each player through the end of
the 2009 NFL season. The performance measures that we have for every player are
the number of games in which they appeared (G), the number of games that they
started (GS), the Career Approximate Value (CAV) and the number of Pro Bowl
(PB) appearances they had. CAV is a player metric devised by Drinen (2008) and is
calculated by Pro-Football-Reference.com and is a method for comparing the
value of players over their careers. Number of Pro Bowl appearances is the num-
ber of times that a particular player has been selected to the end of season all-star
game. Additionally, it is possible to obtain performance measures that are position
specific. In this paper we will not discuss these position specific metrics for two
reasons. First, for a measure to be appropriate for use in assessing trade value it
must assess value across positions. Second, the typical value derived from these
measures tends to follow a similar pattern to those given by the metrics described
above.
2
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
Table 1: Player Counts By Position
Position DB DL K LB OL QB RB TE WR
Counts 507 446 42 365 438 120 281 162 332
DB represents a defensive back, DL a defensive lineman, K a kicker, LBa linebacker,OL an offen-
sive lineman, QB a quarterback, RB a running back, TE a tight end and WR a wide receiver.
Table 2: Player Performance Summary Statistics
Metric 25th 50th 75th Mean Standard
percentile percentile percentile Deviation
Games Played (G) 24.0 66.5 119.0 75.3 59.0
Games Started (GS) 0.0 9.0 62.0 36.5 50.2
CAV 2.0 10.5 33.0 20.7 23.6
Pro Bowls (PB) 0.0 0.0 0.0 0.3 1.2
Among the n=2693 players in our sample, Table 1 has a breakdown of
the number of players selected by position. Kickers (K) includes both punters and
placekickers. Summaries including means and standard deviations for player per-
formance metrics are given in Table 2. Noteworthy in that table is that over 25%,
33.3%, of all drafted players never start a game and at least 75% of all players never
play in a Pro Bowl. 9.3%, or 213, of the players in this sample were still active dur-
ing the 2009 NFL season. As a consequence we will use robust measures of player
performance that are less dependent on these players in the tails of our distributions.
3 Player Performance Evaluation
In this section, we will consider how our performance metrics on NFL players com-
pare to the PVC. We begin by considering Games Played (G). Figure 2 plots G
against the selection for each player in our sample. We fit a non-parametric re-
gression line to these data and this is also plotted in Figure 2. This regression
line was fit using the loess function in the statistical software R (R Development
Core Team, 2007) with a span of 0.5. A locally weighted scatterplot smoothing or
3
Schuckers: Alternative NFL Pick Value Chart
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
xxx
x
x
xx
x
x
x
x
x
x
x
xx
x
xxx
x
x
x
x
x
x
x
x
x
x
xx
x
x
xx
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
xx
x
xxx
x
x
x
x
x
x
xx
xx
x
x
x
x
x
xxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
xx
x
xxx
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
xx
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
xx
x
x
x
x
xx
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
xx
x
x
x
x
xxx
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
xxx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
0 50 100 150 200 250
0 50 100 150 200 250 300
Selection
Games Played (G)
Figure 2: Games Played (G) by Selection with LOESS regression line
LOESS is fit at each value of the predictor by taking a locally polynomial — in this
case quadratic — weighted fit of response values where the weights depend upon
the cubic distance from the value. As we would expect, that fit line is a monoton-
ically decreasing one. We note that the decline in G on this graph as the selection
number increases is more linear than was the case for the PVC (Figure 1). The
non-parametric fit here seems to be composed of two nearly linear line segments.
The first line segment goes from selection 1 to selection 110. The second is not as
steep as the first and goes from selection 110 to selection 255.
We next consider the response GS by selection. The relationship between
GS and selection can be seen in Figure 3. That relationship is noticeably non-linear.
As was the case for games played, the relationship is monotonically decreasing
though the decrease is more exponential in shape. Though this relationship is more
similar to the PVC, the rate of decrease is much slower. The fitted line here reflects
the average decrease in games started as the selection number increases. The num-
ber of players who did not start a game is higher, 896, than to the number of players
that never appeared in a game, 398. This accounts for the intensity of values along
the x-axis in Figure 3.
CAV is the next performance metric that we consider. As mentioned above,
this measure was created by Drinen (2008) as a metric of player value to allow the
comparison of players across positions. We will use it for similar purposes here.
4
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
xxx
x
x
x
xxxx
x
x
x
x
x
x
x
xx
x
x
x
xxx
x
x
xx
x
x
x
x
x
x
xxxxxxx
x
x
x
x
x
x
x
x
xxxx
x
x
xx
x
xxx
x
x
x
x
xx
x
xx
x
x
x
x
x
x
xxxx
x
xx
x
xxxxxxxxx
x
xxx
x
xxxx
x
xx
x
x
xxxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
x
x
x
xx
x
xxx
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
xx
x
x
xx
x
xxxxxxxx
x
xxxx
x
xx
x
x
x
x
x
x
x
xxxx
xx
x
x
xx
x
xxx
x
x
x
xxx
x
x
x
x
x
xxx
x
x
x
xx
x
x
x
xxx
x
x
xxxx
x
x
x
x
x
x
x
xxxxxxxxxxx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
xxxx
x
x
x
xx
x
x
x
x
x
x
xxx
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xxx
x
x
x
x
x
x
x
x
xxx
x
xxxx
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xxx
x
xxx
x
xx
x
x
x
x
x
x
x
x
x
x
xxxxxxxxx
x
x
x
x
xx
x
x
x
xxxxx
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
xxxxx
x
x
x
xxxx
x
x
x
x
xxx
x
x
x
xx
x
x
x
xxxx
x
x
xx
x
xxxxxxxxxx
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
xxxx
x
xxxxxx
x
x
xxx
x
x
x
x
x
xxxxxx
x
xx
x
x
x
x
x
x
xxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
xx
x
xx
x
x
x
x
x
x
xx
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
xxxxxxx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
x
xx
x
xx
x
x
x
xxx
x
x
x
x
x
x
xx
x
x
x
xx
x
xx
x
xxx
x
xx
x
x
x
xxxxx
x
xxx
x
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxx
x
x
x
x
x
x
xx
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
xx
x
xx
x
x
x
x
xxxxxx
x
x
x
xx
x
xxxxx
x
xx
x
xxxx
x
xxxxxxx
x
xxx
x
xxx
x
x
x
x
xxx
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
xxxxx
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xx
x
x
x
x
x
xxxx
x
x
xx
x
x
xxxxxx
x
x
x
xxx
x
x
x
xx
x
x
x
x
x
x
x
xx
x
xxxxx
x
x
xx
x
x
x
x
xxxx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
xx
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
xx
x
xxxxxx
x
x
x
xx
x
x
x
x
xx
x
x
x
x
xxx
x
x
xx
x
xxx
x
xxxxx
x
xx
x
x
x
x
x
xx
x
xxxx
x
xxxx
x
x
xxxx
x
x
x
x
x
x
xx
x
x
xxxx
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xxxx
x
xxx
x
xx
x
x
xx
x
x
x
x
x
x
x
xxx
x
x
x
xxxxxx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxx
x
xxxxx
x
x
xx
xxx
x
x
x
xx
x
xx
x
xxxxxxxxxxx
x
x
x
x
xx
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
x
xx
xx
x
x
x
x
x
x
xx
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
x
x
xx
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
xxxxx
x
x
x
x
x
0 50 100 150 200 250
0 50 100 150 200 250
Selection
Games Started (GS)
Figure 3: Games Started (GS) by Selection with LOESS regression line
Figure 4 shows the relationship between CAV and selection number along with a
LOESS regression line. The overall form of this line is very similar to that of the
LOESS regression line for GS. Both have a exponential rate of decline and both are
monotonic. A comparison of the two curves normalized so that their sums are equal
shows that the difference between the two is never more than 10% and only exceeds
8% after the 230th selection. The correlation between the LOESS regressions for
GS and CAV is 0.995.
The final measure of player performance that we will consider is the number
of Pro Bowl Appearances. Figure 5 gives a scatterplot of PB versus Selection for
all of the players in our sample. The relationship displayed there is dominated by
the number of players, 2385 out of 2693 in our sample, who never appeared in a
Pro Bowl during their careers. The LOESS line that is given in Figure 5 is fairly
linear and includes some non-monotonic behavior between selections 117 and 160.
We attribute this anomaly to three players who had values of PB of more than 5
in that range of selections. We suspect that the typical relationship between PB’s
and Selection is monotonically decreasing. A further analysis of early NFL drafts
suggests a monotonic relationship.
5
Schuckers: Alternative NFL Pick Value Chart
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
xx
x
x
x
xxxx
x
x
x
x
x
x
x
xx
x
x
x
xxx
x
x
xx
xx
x
x
x
x
x
xxxxxx
x
x
x
x
x
x
x
x
xxxx
x
x
xx
x
xxx
x
xx
x
xx
x
x
x
x
x
x
x
x
x
xx
xx
x
xx
x
xxxxxxxxx
x
xxx
x
xxxx
x
x
x
x
x
xxxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xxx
x
x
x
x
xx
x
x
xxx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xxxxxxx
x
x
xxx
x
xx
x
x
x
xx
x
x
xxxx
xx
x
x
xx
x
xxx
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
xxx
x
xxx
x
x
x
x
xx
x
x
x
x
x
x
x
xxxxx
x
xxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xxx
x
xxx
x
xx
x
x
x
x
x
x
x
x
xx
xxxxxxxxx
x
x
x
x
x
x
x
x
x
xxxxx
x
x
x
x
xx
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
xxx
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
xxxx
x
x
x
x
x
xx
x
x
xxxxxx
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
xx
x
xxx
xxx
x
x
x
x
x
x
x
x
x
x
x
x
xxxx
x
xx
x
x
x
x
x
x
xxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
xxx
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
xx
x
x
x
xxx
xx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xxx
x
xx
x
x
x
xxxxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
x
xx
x
xx
x
xx
x
xxxx
x
xxxxx
x
x
x
x
x
x
x
xxx
x
x
x
x
xxx
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
xx
x
x
x
x
xx
x
xx
x
x
x
x
x
x
xxxxxx
x
x
x
xxxxx
x
xx
x
x
x
x
x
x
x
xx
x
xxxxx
x
x
xx
x
x
x
x
xxxx
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
xxxxxx
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xxx
x
x
x
x
x
xxx
x
x
xxxx
x
x
x
x
x
x
xxx
x
x
x
xx
x
x
xxx
x
x
x
x
xxx
x
x
x
x
x
x
xx
x
x
x
xx
x
x
xxxx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxxx
x
xxx
x
x
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
xxxxx
x
xx
x
x
x
x
xx
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
x
x
xx
x
x
xx
xx
x
x
x
x
x
x
x
xx
xx
x
x
x
x
xx
x
x
x
xx
x
xx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
xxx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xx
xxx
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
0 50 100 150 200 250
0 50 100 150
Selection
Career Approximate Value (CAV)
Figure 4: Career Approximate Value (CAV) by Selection with LOESS regression
line
4 An Alternative to the PVC
In the previous section we looked at four different measures of NFL player’s career
performance. All four of those metrics are useful metrics for assessing the value of
a player over their career. What is clear from the above analyses is that based upon
the LOESS regressions none of these metrics matches the PVC in the way that they
value players. All four of the metric that we considered value players in different
ways. Games played (G) values players who contribute to a team in some manner.
Games started (GS) values players who are the best at their position on their team.
Career Approximate Value (CAV) is an measure that assigns to a player a portion
of his team’s value for each year and then uses a weighted sum over his career. Pro
Bowl appearances (PB) rewards a player for being among the best players at their
position for a given year. Figure 6 has a comparison of the LOESS regressions for
each of these metrics versus the PVC taking the natural logarithm of each value
for each selection. The values there are normalized so that the total value for all
metrics is the same as the total value of the PVC. Table 3 has the normalized fitted
values for each of these lines at several selections. Relative to the other metrics, the
current PVC overvalues early draft selections particularly the first 50 selections and
undervalues selections from 150 to 255. From that graph we can see that the PVC
6
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
x
x
xx
x
x
xxx
x
xxxxxxx
x
xxxxxx
x
x
xxxxxx
x
xxxx
x
xxxxxx
x
xxxxxxxxxxxxx
x
xxx
x
xxxxxx
x
xxxxxxx
x
xxxx
x
xxxxxx
x
xxxxxxx
x
xxxxx
x
xxxxxxxx
x
xxxxxxxx
x
x
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
x
xxxxxxx
x
xxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xx
xx
x
x
xx
xx
xxx
x
xxx
x
xxxxx
x
x
xx
x
x
xxxx
xx
x
xxxxxxxxxxxxxxxxx
x
xxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxx
x
xxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
x
x
xxxx
x
x
x
xx
x
xxxxxxx
x
xxxx
x
xxxx
x
x
x
xxxxxx
x
x
x
x
x
xxxxxxxxxxxxxxxxxxxx
x
xxxx
x
xxx
x
xxxxxxx
x
xxxxxx
x
xxxxxx
x
xxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxx
x
xxxxx
x
xxxxxx
x
xxxxxxxxxxxxxxx
x
xxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxx
x
x
x
x
x
xxxxxx
x
xxxxxx
x
xxxx
x
xx
x
xxx
x
x
x
x
x
x
x
x
x
x
x
x
xxxxx
xx
xxxxxxxxxx
x
x
x
xx
x
xxxxx
x
xxx
x
xxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxx
x
xxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxx
x
xxx
x
xxx
x
xxxxxxxxxxxxxxxx
x
xxxxx
x
x
x
xx
xxxxx
x
x
x
x
x
xxx
x
xx
x
x
xxx
x
xxxxxxxxxxxxxxxxxxx
xxx
xxxxx
x
xxx
x
xxxxx
x
xxx
x
xxx
x
xxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxx
x
xxxxxxxxxxxxx
x
xxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxx
x
xxxxxxxxxxxxxxxxxxx
x
x
x
x
xx
x
xx
x
xx
x
x
xxxx
x
xxx
x
x
x
x
x
x
xxxx
x
x
x
x
xxxxxx
x
x
xxxxxxxxxxxxxxxx
x
xxx
x
xxxxx
x
xxxxxxxxxxxxxx
x
xx
x
x
xxxxxxx
x
xxx
x
xxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxx
x
x
xxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
x
x
x
x
x
x
x
xxx
x
x
xxxxx
x
xxxxxxxx
x
xxxxxxx
x
xxxxx
x
xx
xxxxxxx
x
xxxxxxx
x
xxxx
x
x
xxxxxx
x
xxxxxxxxxxxx
x
xxxx
x
xxxxxx
x
xxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xx
x
xxx
xx
x
x
x
x
xxxxxxx
x
xxxx
x
xxxxxx
x
xxxx
x
x
xxxx
x
x
x
x
x
xxxxxxxxx
x
xxxxx
x
x
xxxxxx
x
xxx
x
xxxxxxxx
x
xxxxx
x
x
xxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxx
x
xxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
x
x
x
x
x
x
x
xx
xxxx
x
x
xxxxx
x
xxxxxx
x
x
xxxxxx
x
xx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
x
x
xx
x
x
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxx
x
xxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxx
x
x
x
xxx
x
x
xx
x
xx
x
x
xx
x
xxxxxxxxxx
x
xxxxxxx
xxx
xxx
x
xxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxx
x
xxx
x
xxxxxxxxxxxxxx
x
x
xxxxxxxxxxxxxxxx
x
xxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxx
xx
x
x
xx
xx
x
x
x
x
x
xx
x
x
x
x
x
x
x
x
xxx
x
xx
x
x
x
x
x
x
x
xx
x
xxxx
x
x
x
x
x
xxx
x
xxxxx
x
xx
x
x
x
x
xxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
0 50 100 150 200 250
0 2 4 6 8 10 12
Selection
Pro Bowls (PB)
Figure 5: Pro Bowl Appearances (PB) by Selection with LOESS regression line
roughly mirrors the LOESS value of PB over the first approximately 120 selections,
though the PVC does exceed the fit PB by 0.8 on the log scale for the first selection.
In this graph the non-monotonicity of the PB fitted line is apparent. Additionally,
based upon Figure 6 all of the other metric diverge notably from the PVC after the
100th selection which suggests that the PVC undervalues players taken after this
selection relative to these other measures. As mentioned above, GS and CAV have
very similar LOESS lines while G has a line that values later selections higher than
any of the other methods.
Keeping all of this in mind, we propose to use LOESS predicted GS as the
metric for our alternative PVC (APVC). We do so for several reasons. First, we
concur with Massey and Thaler (2010) that the current PVC overrates early draft
selections. Second, our analysis finds that the PVC underrates late draft selections
relative to all of our other measures of player performance. Third, we find that
GS is a better overall metric than G or PB for evaluating the worth of a player
even allowing for a monotonic version of a fitted PB line. Fourth, we choose GS
over CAV since it is a simpler metric and, therefore, is easier to calculate and to
predict. Our proposed APVC is found in Table 4. To make the APVC somewhat
comparable in value to the PVC, we inflated the value of the first selection to 1000
and inflated all of the other selections the same amount. (The inflation factor here
is 6.71.) We do note that the relative values of the 1st and 255th selections are a
7
Schuckers: Alternative NFL Pick Value Chart
Table 3: Normalized Fitted Value of Performance Metrics for Specific Selections
Selection 1 50 100 150 200 250
VPC 3000.0 400.0 100.0 31.4 12.4 0.7
G 420.3 314.3 234.7 196.6 174.4 157.3
GS 739.6 397.9 214.1 122.3 99.5 77.9
CAV 704.8 348.7 208.1 150.7 132.6 114.6
PB 1363.5 350.8 117.8 101.0 79.0 26.4
factor of approximately 10 for the APVC while they are 7500 for the VPC. There is
an additional benefit to using an APVC based upon games started which is that the
AVPC provides a way to evaluate how individual draft selections perform relative
to what is expected in a straightforward manner.
0 50 100 150 200 250
0 2 4 6 8
Selection
log(Value)
PVC
G
GS
CAV
PB
Figure 6: Comparison of PVC and alternative player value metrics
8
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
Table 4: Proposed Alternative NFL Pick Value Chart
Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value
1 1000 33 670 65 449 97 301 129 209 161 149 193 135 225 123
2 988 34 661 66 443 98 297 130 206 162 148 194 135 226 122
3 976 35 653 67 438 99 293 131 204 163 147 195 135 227 121
4 964 36 644 68 433 100 289 132 202 164 146 196 135 228 121
5 953 37 636 69 427 101 286 133 200 165 145 197 135 229 120
6 941 38 628 70 422 102 282 134 197 166 145 198 135 230 119
7 930 39 620 71 417 103 279 135 195 167 144 199 135 231 118
8 918 40 612 72 412 104 275 136 193 168 143 200 134 232 117
9 907 41 604 73 407 105 272 137 191 169 143 201 134 233 117
10 896 42 597 74 403 106 269 138 189 170 142 202 134 234 116
11 885 43 589 75 398 107 265 139 187 171 141 203 134 235 115
12 874 44 581 76 393 108 262 140 185 172 141 204 134 236 114
13 863 45 574 77 388 109 259 141 183 173 140 205 133 237 113
14 853 46 566 78 383 110 256 142 181 174 140 206 133 238 113
15 842 47 559 79 378 111 253 143 179 175 139 207 132 239 112
16 832 48 552 80 374 112 250 144 177 176 138 208 132 240 111
17 821 49 545 81 369 113 247 145 175 177 138 209 131 241 111
18 811 50 538 82 364 114 244 146 173 178 138 210 131 242 110
19 801 51 531 83 360 115 241 147 171 179 138 211 131 243 109
20 791 52 525 84 355 116 239 148 169 180 137 212 130 244 109
21 781 53 518 85 351 117 236 149 167 181 137 213 130 245 108
22 771 54 512 86 346 118 234 150 165 182 137 214 130 246 107
23 761 55 506 87 342 119 232 151 164 183 137 215 129 247 107
24 752 56 500 88 337 120 229 152 162 184 137 216 129 248 106
25 742 57 494 89 333 121 227 153 160 185 137 217 128 249 106
26 733 58 489 90 329 122 225 154 158 186 137 218 128 250 105
27 724 59 483 91 325 123 223 155 157 187 137 219 127 251 105
28 714 60 477 92 321 124 220 156 155 188 137 220 127 252 104
29 705 61 471 93 316 125 218 157 154 189 137 221 126 253 104
30 696 62 466 94 312 126 216 158 152 190 136 222 125 254 103
31 687 63 460 95 308 127 213 159 151 191 136 223 125 255 103
32 679 64 454 96 305 128 211 160 150 192 136 224 124
9
Schuckers: Alternative NFL Pick Value Chart
5 Discussion
In this paper, we have evaluated the current NFL Pick Value Chart (PVC). We
have analyzed other performance metrics for evaluating the value of a draft selec-
tion. These metric included the number of games in which a player appears, the
number of games that a player starts, the number of times they are chosen for the
Pro Bowl and their Career Approximate Value (CAV). Our analysis agrees with
previous work done by Massey and Thaler (2010) and finds that the current PVC
overvalues early draft selections. Thus, we have proposed an alternative PVC or
APVC that is based upon an non-parametric regression of games started onto draft
selection. This APVC can be found in Table 4. This alternative provides a new
method for determining the value of a draft selection. One future direction for this
work is to develop methodology for considering which teams drafted best during
the time period covered by our sample.
There are certainly other factors that go into how a team might value a par-
ticular section besides the factors that we have considered here. Foremost among
these is certainly the needs that a team might have at a particular position. If such
a need exists and a team feels that a particular player can fulfill such a need, then
they might attach additional value to a selection. Additionally, a player might pro-
vide revenue to a team through additional ticket or merchandise sales. We have
not considered player salaries as part of this analysis but they certainly do impact
player selections. Early selections are paid substantially more than later selections.
For the 2010 draft, the first selection, Sam Bradford, was paid an annual rate of
approximately $13,000,000 and the 255th player selected, Josh Hull, was paid an
annual rate of approximately $456,000, a ratio of approximately 28.5.
10
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
6 Appendix
Table 5: Current NFL Pick Value Chart
Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value Sel. Value
1 3000 33 580 65 265 97 112 129 43 161 28 193 15.2 225 2.9
2 2600 34 560 66 260 98 108 130 42 162 27.6 194 14.8 226 2.8
3 2200 35 550 67 255 99 104 131 41 163 27.2 195 14.4 227 2.7
4 1800 36 540 68 250 100 100 132 40 164 26.8 196 14 228 2.6
5 1700 37 530 69 245 101 96 133 39.5 165 26.4 197 13.6 229 2.5
6 1600 38 520 70 240 102 92 134 39 166 26 198 13.2 230 2.4
7 1500 39 510 71 235 103 88 135 38.5 167 25.6 199 12.8 231 2.3
8 1400 40 500 72 230 104 86 136 38 168 25.2 200 12.4 232 2.2
9 1350 41 490 73 225 105 84 137 37.5 169 24.8 201 12 233 2.1
10 1300 42 480 74 220 106 82 138 37 170 24.4 202 11.6 234 2
11 1250 43 470 75 245 107 80 139 36.5 171 24 203 11.2 235 1.9
12 1200 44 460 76 210 108 78 140 36 172 23.6 204 10.8 236 1.8
13 1150 45 450 77 205 109 76 141 35.5 173 23.2 205 10.4 237 1.7
14 1100 46 440 78 200 110 74 142 35 174 22.8 206 10 238 1.6
15 1050 47 430 79 195 111 72 143 34.5 175 22.4 207 9.6 239 1.5
16 1000 48 420 80 190 112 70 144 34 176 22 208 9.2 240 1.4
17 950 49 410 81 185 113 68 145 33.5 177 21.6 209 8.8 241 1.3
18 900 50 400 82 180 114 66 146 33 178 21.2 210 8.4 242 1.2
19 875 51 390 83 175 115 64 147 32.6 179 20.8 211 8 243 1.1
20 850 52 380 84 170 116 62 148 32.2 180 20.4 212 7.6 244 1
21 800 53 370 85 165 117 60 149 31.8 181 20 213 7.2 245 0.95
22 780 54 360 86 160 118 58 150 31.4 182 19.6 214 6.8 246 0.9
23 760 55 350 87 155 119 56 151 31 183 19.2 215 6.4 247 0.85
24 740 56 340 88 150 120 54 152 31.6 184 18.8 216 6 248 0.8
25 720 57 330 89 145 121 52 153 31.2 185 18.4 217 5.6 249 0.75
26 700 58 320 90 140 122 50 154 30.8 186 18 218 5.2 250 0.7
27 680 59 310 91 136 123 49 155 30.4 187 17.6 219 4.8 251 0.65
28 660 60 300 92 132 124 48 156 30 188 17.2 220 4.4 252 0.6
29 640 61 292 93 128 125 47 157 29.6 189 16.8 221 4 253 0.55
30 620 62 284 94 124 126 46 158 29.2 190 16.4 222 3.6 254 0.5
31 600 63 276 95 120 127 45 159 28.8 191 16 223 3.3 255 0.45
32 590 64 270 96 116 128 44 160 28.4 192 15.6 224 3
References
Berry, S. M. (2001): “Do you feel a draft in here?” Chance, 14.
Drinen, D. (2008): “Approximate value in the NFL,”
http://www.pro-football-reference.com/blog/?p=465.
Maier, C. (2010): “Updating the NFL draft trade value
chart,” http://www.nfldraft101.com/draft/articles/
407/Updating the NFL Draft Trade Value Chartjsp .
11
Schuckers: Alternative NFL Pick Value Chart
Massey, C. and R. Thaler (2010): “The loser’s curse: Overcon
fi-
dence vs. market efficiency in the National Football League draft,”
http://ssrn.com/abstract=697121.
McGuire, M. (2010): “Matt Mcguire’s NFL Draftology 321: Revising the trade
value chart,” http://walterfootball.com/nfldraftology321.php.
Pro Football Reference (2010): “NFL draft,”
http://www.pro-football-reference.com/draft/.
R Development Core Team (2007): R: A Language and Environment for Statisti-
cal Computing, R Foundation for Statistical Computing, Vienna, Austria, URL
http://www.R-project.org, ISBN 3-900051-07-0.
Smith, M. D. (2007): “Value of high draft picks
can be misleading,” http://www.nysun.com/sports/
value-of-high-draft-picks-can-be-misleading/52792/.
Stuart, C. (2008): “The draft value chart: Right or wrong?”
http://www.pro-football-reference.com/blog/?p=527.
12
Submission to Journal of Quantitative Analysis in Sports
http://www.bepress.com/jqas
... In the National Football League (NFL), the Chart (Fig 1B), as it is commonly known, shows the typical exponential decline from high (the initial selections, with the first selection having a re-scaled value of 3000 points) to low draft selections. The Chart was estimated in 1991 by staff of the Dallas Cowboys using trades from 1987 to 1990 [12,19]. Massey and Thaler [12] compare the ratio of market values of alternative draft selections (from the Chart) against surplus value (defined as the performance value less the salary paid). ...
... Whether performance of the recruited players meets expectation in reality is purely probabilistic and, as such, a closer consideration of the distribution (and in particular the variance) of possible outcomes can be enlightening. As illustrated by Schuckers [11] and Hurley et al [19], distributions of performance are highly skewed with a high density of zeroes, depending on the metric chosen, for lower draft selections. This immediately leads to questions of whether the measure of central tendency used to compare trades should be the mean or instead the median. ...
... (C) The draft value from selection 1 (S1), two of selection 9 and one selection 18 (2 × S9 + S18), and three of selection 18 and one selection 9 (3 × S18 + S9). Schuckers [19] show figures with similar characteristics when considering games played, games started, career approximate value and Pro-Bowl appearances as a function of draft selection. ...
Risk equivalence as an alternative to balancing mean value when trading draft selections and players in major sporting leagues
Article
Full-text available
In sports leagues that use an annual draft to assign eligible players to clubs, having a value associated with a draft selection can allow clubs to anticipate future growth of players and, if a trading period exists, assist negotiations when exchanging draft selections and players. Typically, mean draft values often decline in either an exponential or geometric manner with increasing draft selection number. Aggregate mean values have been used to compare trade packages. However, clubs may also want to ensure that a trade does not increase the probability of obtaining poor players in the draft. This paper therefore considers equivalence of risk as an alternative trading strategy for club list managers. Here, risk is defined as the probability of the aggregate value of the received draft selections being below a minimum acceptable level. For risk equivalence, a premium over and above mean market value may need to be provided when trading to secure higher draft selections.
View
... Further, as early selections do tend to gain more tenure and game time irrespective of their talents [30,31], a case could be made against the use of player payments as the determinant of the DVI. To circumvent this, Mitchell et al. [32] and Stewart et al. [23] use games played and champion data playing ranking points (Champion Data (CD), in conjunction with the school of mathematics at Swinburne University, introduced the CD player ranking points as a proprietary method of objectively evaluating player performance by assigning values to a range of on field statistics that are important for team success) as alternative measures of performance to value picks within the AFL, consistent with studies in other leagues [16,[33][34][35]. Their findings characterise further inefficiencies within the AFL caused by the F/S rule and the selection of Aboriginal and Torres Strait Islander players, which is unique to the league itself. ...
... the sport. Hence, many authors have chosen games started [35], ranking points [23] and contribution to the margin of victory [31]. Studies that use games played have described it as a determinant rather than an outcome or measure of draftee performance [37]. ...
Valuing Australian football league draft picks
Article
Full-text available
To ensure uncertainty in match outcomes, professional sporting leagues have used various competitive balance policies, including player salary caps, revenue sharing among teams and player drafts. The Australian Football League (AFL) introduced a player draft in 1986, and to refine its operation, a draft value index (DVI) was introduced in 2015. The DVI allocates a numeric value to each individual player draft pick, with these values determined by the AFL using historic player compensation or wage and salary data. The AFL DVI plays an essential role in the operation of its player draft; however, other research has questioned the validity of such indexes. This paper aims to produce an alternative to the AFL DVI. The former index uses career compensation as the determinant of value, whereas we use other measures of player performance. First, various models were developed to predict on-field performance, such as games played (both in a recruit’s career and season) after a draftee was selected for the first time by a team. This was then retrofitted to the pick used to select these draftees to create the new DVIs. Even though the predicted DVI followed an inverse monotonic function like the existing index, the decline in value for the DVI produced here was less steep, unlike the AFL’s. This allowed us to conclude that players’ salaries did not always strongly correlate to performance. The change in performance between players selected at different points in the draft did not vary as much as their wages. Though this scheme is applied to the AFL, the underlying concept could be directly exported to other player drafts.
View
... In the domain of professional leagues, varied research has been initiated for obtaining meaningful insight of its dynamics and functionalities in the recent past (Tingling, 2017;Schuckers, 2011). A lot of focus has been on the Indian Premier League, which is a cricketing tournament organized by the Board of Cricket Control in India, in a twenty-twenty format. ...
... In this section, we illustrate the proposed backward subset selection technique (see Section 2.1) through a simulation study. For this purpose, we simulated the performance data of the players by considering 8 variables with the mean vector (20,30,40,50,45,55,35,25) and the covariance matrix At first, we obtained the full model (8) by considering all the variables and noted the corresponding value of the objective function. We obtained the subsequent sub-optimal solutions thereafter, by dropping the variable which had the least associated weight, as described in Section 2.1 and the corresponding the scree plot is shown in Figure 2. We clearly observe that the objective function attains the highest value when all the variables are considered, and there is a drop once we start to eliminate variables. ...
Player selection strategy: A quantitative perspective
Conference Paper
Full-text available
  • Mar 2020
In most of the professional sporting leagues across the world, players are selected by team management and owners through various types of auctions. A ranked list of players is very useful for bidders to decide upon the preferences of possible buys. In this paper, we propose an efficient scoring system and ranking scheme of players based on performance and experience related attributes from the historical data. The proposed methodology maximally discriminates the players and incorporates both averages and the consistencies of the performance variables. The method is illustrated with real life data analysis and simulation study.
View
... Their findings, while upholding their hypothesis, showed a lower rate of decline in pick value when compared to the PVC. Conlin and Emerson [20] expanded on this to use games started and active contract lengths as determinants of pick value in the NFL reconfirming the previous findings, similar to Schuckers who plotted this relationship with the help of local polynomial functions both in the NFL [21] and NHL [22]. Studies within the AFL have also used player ratings to determine pick value through linear regression models [23,24] as the league's own draft value index (DVI) mimicked certain properties in the NFL PVC. ...
Optionality in Australian Football League draftee contracts
Article
Full-text available
Though player drafts have commonly been utilised to equitably disperse amateur talent and avoid bidding wars, often they have also been accused of creating a monopsony labour market which restricts player movement. Within the Australian Football League (AFL) some have called for the increase of the initial draftee contract from two to three seasons, which further pushes the envelope on monopsony power. Instead of increasing the contract length, this paper suggests a call option to be purchased by the teams allowing them to add a further season to the draftee contract at a predetermined compensation package should they choose to do so at the end of the initial contract. The call prices per pick were calculated using the Black-Scholes model and were valued between 1% and 1.5% of the pick value. However, it failed to follow a monotonic function similar to pick value, owing to managerial overconfidence and sunk investment plays. Overall, the findings allow teams to procure the option of increasing initial draftee contracts and not impede further on a player’s ability to move.
View
... If a team is intending to choose a player that they know would not be selected prematurely and they have an early pick, it would be in their best interest to relieve the pick in lieu for selections later on, to avoid hefty salary negotiations. 2 Current studies have suggested pick values by regressing games played (Conlin & Emerson, 2005;Hurley, Pavlov, & Andrews, 2012;Schuckers, 2011a), time on field (Schuckers, 2011b) and player ratings (Mitchell, Stavros, & Stewart, 2011;Stewart, Stavros, Phillips, Mitchell, & Barake, 2016) to name a few. 3 Taylor, Mecalfe, Dugan, and Ezell (2018) proposed a model by which picks in the current year could be traded against those in the future within the NFL. the medical profession, survival analysis is used to evaluate similar binary outcomes over time, such as in clinical trials of new pharmaceuticals. The outcome there would be dead or alive, while a cohort of predictors dictate the time to this event. ...
An Experimental Method in Juxtaposing Draft Picks and Active Players
Article
  • Apr 2021
  • Econ Paper Econ Soc Aust
Trading picks is a common transaction that is observed in sporting leagues that conduct player drafts. However, with its evolution, current iterations allow for the trading of picks and players, which in turn raises questions about valuing such trades. This study proposes a pick valuation system using survival analysis techniques, which was further extended to value players who were already enlisted in the league, creating a common denominator. The model proved to be viable, allowing to evaluate trades involving both picks and players.
View
... In an attempt to verify the theory both in the immediate contract years 7 and a draftee's lifetime, the career cross section and seasonal panel were broken into two for the total career and the initial contract period. Previously, various metrics including games played (Schuckers, 2011a), time on field (Schuckers, 2011b) and the years taken to achieve a set number of games (Glasson and Bedford, 2009) have been used to quantify the contribution of a draftee after his initial selection. However, in this study, a player's contribution to the team's margin of victory (Chandrakumaran, in review) 8 was used to analyse post draft 4 Teams within the AFL maintain two lists called the seniors (which is the regular roster) and the rookies. ...
The AFL Pick Trading Market as a Coasian Utopia
Article
  • Aug 2020
  • J SPORT ECON
Sporting leagues have various competitive balance measures including player drafts, where teams are awarded picks based on prior performance. Teams, however, have the option to either exercise this pick or trade it as they see fit. An analysis of this trading market in the NFL stated that players obtained through traded picks contributed more to their recruiters in comparison to their counterparts. This paper retested this within the AFL and disproved this fact, establishing the Coasian notion of the efficient allocation of resources, based on some differences between the list management systems between the two leagues.
View
... In addition, if a team does not covet a certain player when they are scheduled to make a selection in the draft, they might consider trading their draft selection away to another team. Though literature is limited in this area of study, one proposed method incorporates nonparametric regression in order to help managers identify the trading value of draft picks (Schuckers, 2011). Another concept associated with player replacement involves decisions related to immigrating the risk associated with injuries. ...
A Team-Compatibility Decision Support System for the National Football League
Article
Full-text available
  • Jul 2020
Many factors are considered when making a hiring decision in the National Football League (NFL). One difficult decision that executives must make is who they will select in the offseason. Mathematical models can be developed to aid humans in their decision-making processes because these models are able to find hidden relationships within numeric data. This research proposes the H euristic E valuation of A rtificially R eplaced T eammates (HEART) methodology, which is a mathematical model that utilizes machine learning and statistical-based methodologies to aid managers with their hiring decisions. The goal of HEART is to determine expected and theoretical contribution values for a potential candidate, which represents a player’s ability to increase or decrease a team’s forecasted winning percentage. In order to validate the usefulness of the methodology, the results of a 2007 case study were presented to subject matter experts. After analyzing the survey results statistically, five of the eight decision-making categories were found to be “very useful” in terms of the information that the methodology provided.
View
Identifying the Ideal Salary Distribution Under the MLS Salary Cap Using Machine Learning
Article
Full-text available
  • Jan 2023
As one of the fastest-growing leagues in the world, Major League Soccer has garnered a lot of attention recently. Its most notable feature, however, is the stringent salary cap all teams must adhere to. This paper looks at how Machine Learning can inform a better salary allocation model. This meant primarily determining the best salary variability and distribution by position. To determine the best allocation model, several models were created and the two most accurate were used to determine the best model. The four models made were two decision trees and two linear regression models, an XGBoost model and a Random Forest model and a LASSO model and a Ridge Regression model. The results found that increased salary variability is beneficial for teams as well as homegrowns being beneficial. In addition, it was found that a redistribution of salaries by position would be beneficial. Less money for defenders, more for a starting goalkeeper, and less for the rest of the goalkeepers were all found to be distinctly positive adjustments.
View
On the Value of a Premium College Football Player: Evaluating the Literature
Article
  • Mar 2024
  • J SPORT ECON
Over the past decade, the issue of player compensation in college sports has been the subject of several successful legal challenges. Athletes contend that the compensation they receive falls significantly short of the value they generate, attributing this gap to unlawful National Collegiate Athletic Association restrictions. Numerous tools exist in the sports economic literature that estimate the value of college athletes, with an emphasized focus toward premium college football players. In addition to providing updated estimate of player marginal revenue product (MRP), we review past and contemporary methodologies for estimating college player MRPs. We contend that, while presenting some evidence that restrictions on player compensation resulted in the extraction of the majority of the value generated by top college athletes, existing methods leave considerable uncertainty over the magnitude of exploitation.
View
Relative age effects in American professional football
Article
  • Jun 2019
We test for the existence of relative age effects in professional American football. In a sample of 18,898 football players born on or after 1940, there is an excess of January and February births – consistent with a relative age effect associated with calendar year – as well as a slight increase in September births – consistent with the fact that some football players we analyze attended high school in states with fall school cutoff dates. We consider the possibility that relative age effects may affect skilled football positions more than positions relying heavily on player weight, and we find suggestive evidence of this. Lastly, and contrary to what has recently been shown in professional hockey, we find no evidence that misguided preferences for relatively older players lead to selection-based inefficiencies in football player drafting. Our results have implications for evaluating potential football players and speak broadly to the role of physiological factors beyond player control on athletic success.
View
View
Approximate value in the NFL
  • Jan 2001
  • S M Berry
Berry, S. M. (2001): "Do you feel a draft in here?" Chance, 14. Drinen, D. (2008): "Approximate value in the NFL," http://www.pro-football-reference.com/blog/?p=465.
Updating the NFL draft trade value chart
  • Jan 2010
  • C Maier
Maier, C. (2010): "Updating the NFL draft trade value chart," http://www.nfldraft101.com/draft/articles/
Matt Mcguire's NFL Draftology 321: Revising the trade value chart
  • Jan 2010
  • M Mcguire
McGuire, M. (2010): "Matt Mcguire's NFL Draftology 321: Revising the trade value chart," http://walterfootball.com/nfldraftology321.php. Pro Football Reference (2010): "NFL draft," http://www.pro-football-reference.com/draft/.
The draft value chart: Right or wrong
  • Jan 2008
  • C Stuart
Stuart, C. (2008): "The draft value chart: Right or wrong?" http://www.pro-football-reference.com/blog/?p=527.