Research Update 1– The Difference Between a “Good” and “Great” Player

Introduction

The first goal of my project has been to create a statistical threshold that a player must reach before he will be considered for the Hall of Fame. In this post I will detail the methods and process I used to develop the threshold, which variables I used, and how I plan to use it during the rest of my research project.

I focused on the Wide Receiver position because their contributions to a team are simple and easy to quantify. The following is a brief description of the position for readers less familiar with the sport. In American Football, a team puts 11 players on the field at a time. A Wide Receiver plays when the team has possession of the ball and is trying to score. A Football game is divided into events called “plays,” which generally start with all players completely motionless and last between 5-15 seconds. All plays begin with the Quarterback putting the ball in play, either by handing the ball off to a running back, throwing the ball to a receiver, or, less frequently, running the ball himself. A receiver’s primary role is to get open downfield and catch a forward pass from the quarterback, and thus can be measured simply in catches, yards, and touchdowns (scores). Because of the inherent statistical simplicity of the position, I chose to focus on this position.

 

Data 

As a starting point, I used two data sets: the receiving statistics of all Hall of Fame Receivers and the annual league receiving statistics for all players combined. I hypothesized that I could find a consistent statistical difference between Hall of Fame Receivers and their peers. The statistic I am using is Yards per Game, which is explained at the end of this post.

Table 1 shows the creation of the statistically “average” receiver in a given year. The Yards per Game statistic in column 2 is the average per team per game. Since a team can have exactly 5 eligible receivers on a given play, a logical average player would be 1/5th of the team’s total receiving yardage in an average game, as shown in column 3.

Table 2 shows the Yards/Game statistic of each Hall of Fame Receiver in column 3. Column 2 shows a representative year for each receiver, the midpoint year of their career. This is a simplification that I will replace with complete career arcs in the next stage of the project. Column 4 shows the difference between the player’s statistic and the average league statistic for that player’s representative year.

 

Table 1                                                                        Table 2

Year (1931=0) Yds/Game Yds/Game/5 Player Year (1931=0) Yds/Game Difference
81 231.3 46.26 Jerry Rice 64 75.56 31.40
80 229.7 45.94 Cris Carter 64 59.40 15.24
79 221.6 44.32 Michael Irvin 63 74.87 32.15
78 218.5 43.7 Andre Reed 62 56.40 16.28
77 211.3 42.26 Art Monk 57 56.79 16.65
76 214.3 42.86 James Lofton 55 60.10 19.00
75 204.8 40.96 Steve Largent 52 65.45 24.53
74 203.5 40.7 John Stallworth 50 52.87 11.99
73 210.5 42.1 Lynn Swann 47 47.09 15.33
72 200.4 40.08 Charlie Joiner 47 50.82 19.06
71 212.2 42.44 Fred Biletnikoff 41 47.23 16.81
70 205.8 41.16 Paul Warfield 40 54.55 23.41
69 206.9 41.38 Charley Taylor 40 55.21 24.07
68 212.3 42.46 Don Maynard 35 63.62 27.96
67 205 41 Raymond Berry 30 60.23 24.13
66 201.8 40.36 Tom Fears 21 62.03 30.03
65 207.4 41.48 Elroy Hirsch 21 55.35 23.35
64 220.8 44.16 Dante Lavelli 20 52.75 20.21
63 213.6 42.72 Pete Pihos 20 52.51 19.97
62 200.6 40.12
61 187.6 37.52
60 199.1 39.82
59 194.8 38.96
58 210.9 42.18
57 200.7 40.14
56 203.9 40.78
55 205.5 41.1
54 204.5 40.9
53 205.9 41.18
52 204.6 40.92
51 199.4 39.88
50 204.4 40.88
49 196 39.2
48 180.4 36.08
47 158.8 31.76
46 141.9 28.38
45 152 30.4
44 162.8 32.56
43 153.2 30.64
42 140.9 28.18
41 152.1 30.42
40 155.7 31.14
39 161.4 32.28
38 177.5 35.5
37 168.5 33.7
36 179.5 35.9
35 178.3 35.66
34 183.8 36.76
33 173.6 34.72
32 185.7 37.14
31 193.8 38.76
30 180.5 36.1
29 170.6 34.12
28 169.8 33.96
27 180.4 36.08
26 153.4 30.68
25 147.6 29.52
24 159.8 31.96
23 191.6 38.32
22 173.8 34.76
21 160 32
20 162.7 32.54
19 165.7 33.14
18 161 32.2
17 173.9 34.78
16 180.6 36.12
15 143.1 28.62
14 143.8 28.76
13 129.3 25.86
12 141.4 28.28
11 123.7 24.74
10 121.8 24.36
9 125.3 25.06
8 128.8 25.76
7 105.8 21.16
6 93 18.6
5 83 16.6
4 79.7 15.94
3 59.3 11.86
2 77.9 15.58
1 55.2 11.04

 

Creation of the Threshold

The “Difference” in column 4 became my idea of a threshold. I think of it as exactly how much more production a Hall of Famer like Jerry Rice gave his team each week than his peers did. In order to apply this to any season, I found the mean of the “Difference” data set– essentially how much better the average Hall of Fame player was than his peers. I created the first of my lines by adding this number, 21.66, to the average receiving yard value from each season. But this gave me a representative benchmark of an “average” Hall of Fame production level for each season.

In order to get a threshold, or a line that divided Hall of Famers from their peers, I subtracted one standard deviation from the “Difference” set (5.83), giving me a line that, by definition, approximately 85% of Hall of Fame players land above. In fact, only 2 Hall of Famers, Lynn Swann and John Stallworth, fall below the threshold. They played almost their entire careers together for a Pittsburgh Steelers team that won 4 Super Bowls, so they get a pass (no pun intended) for falling just short in this category.

The Graph on the Left shows my process of developing my threshold. The blue dots show the League averages that I started with (from Table 1, Column 3). The Orange dots each represent a Hall of Fame player. The gray line represents the statistically “average” Hall of Fame player, and the Yellow line represents my threshold.

 

The equation for the threshold is as follows:

Y=(1/5)L(x) + 15.83

where

x is a given year after 1931 (1931=0)

L(x) = the average yards/game for all teams from that season (from Table 1, above)

 

Note: To get a clear view of the graph, click on it, and then on the next screen click on it again.

 

Testing the Threshold

I hypothesized that this receiving threshold set my Hall of Fame Receivers apart from all others. In order to test this, I found receiving data for a lot of very good, but not Hall of Fame receivers. In the middle graph, I represent them in the same way as I represent the Hall of Famers– as a point with their career average of receiving yards coinciding with the year representing the “middle” of their career. As you can see, the threshold does not work by itself. There are easily 100 current and former players whose Yards/Game statistic is comparable to, or better than, Hall of Famers from the same era. But, the vast majority of these players are not, and will never be,  Hall of Famers. So what else sets Hall of Famers, the greatest players, apart from the very good players in the middle chart?

The first answer I noticed was a player’s career receiving yards. Hall of Famers have far more career yards than the other players with high Yards/Game numbers. This is indicative of the other statistic that sets great receivers apart from good ones: longevity. A Hall of Fame player must produce at a high level for a long time, but how long?

Table 3 shows the Hall of Fame receivers once again organized chronologically, along with the length of their careers in column 3. Only two players who played after the founding of the AFL (1960, year 29) played for fewer than 12 seasons: 4-time Super Bowl Champion, and MVP of Super Bowl X, Lynn Swann, and 3-time Super Bowl Champion Michael Irvin. I hypothesized that, barring the impressive postseason accomplishments that these rare players can boast, a player who played the vast majority of his career after 1960 must have played for at least 12 seasons in order to be considered for the Hall of Fame. A player who played before then probably must have played for at least 10 seasons in order to be retroactively considered for the Hall of Fame, but I cannot make a real conclusion about this based on only 5 players.

In the graph on the right above, as well as in Table 4, below, you can see that there are only 8 players who meet both of these criteria (are above the statistical threshold and played for long enough). Several are not even eligible for the Hall of Fame, others have committed serious crimes since leaving the game, and a few appear to have been simply forgotten by the voters for one reason for another. Their stories, as well as some history and development of the game, will appear in my next post.

 

Table 3                                                                                     Table 4

Player Year (1931=0) Career Length Player Year (1931=0) Yds/Game Career Length
Jerry Rice 64 19 Randy Moss 67 70.1 14
Cris Carter 64 15 Marvin Harrison 65 76.7 12
Michael Irvin 63 11 Terrell Owens 65 72.8 14
Andre Reed 62 15 Isaac Bruce 63 68.2 15
Art Monk 57 15 Jimmy Smith 61 69 13
James Lofton 55 15 Tim Brown 57 58.6 16
Steve Largent 52 13 Henry Ellard 52 60.4 15
John Stallworth 50 13 Billy Howton 21 59.6 11
Lynn Swann 47 8
Charlie Joiner 47 17
Fred Biletnikoff 41 13
Paul Warfield 40 13
Charley Taylor 40 13
Don Maynard 35 15
Raymond Berry 30 12
Tom Fears 21 8
Elroy Hirsch 21 11
Dante Lavelli 20 10
Pete Pihos 20 8

 

Assumptions 

I decided that I could ignore a couple of variables,  such as offensive system and surrounding talent. My thought process is outlined below.  By “offensive system,” I mean the types of plays a team ran, and how many they ran per game. The Buffalo Bills of the early 1990s, for example, led the league in offensive plays per game.

It would be logical that a receiver in this system would get way more yards per game than his peers– not so. The offensive production is still distributed between 4-5 players, and in the case of the Bills, they had a Hall of Fame Quarterback (Jim Kelly), a Hall of Fame Running Back (Thurman Thomas), and two Hall of Fame Receivers (James Lofton and Andre Reed). As is the case for many great offenses, the overall production  increased, but the team produced multiple players with Hall of Fame statistics in the Yards/Game category, not just one statistical monster.

The same trend repeats itself on teams like the 1970’s Steelers (Hall of Fame receivers Lynn Swann and John Stallworth) and the Vikings of the late 1990’s (Hall of Fame receivers Cris Carter and Randy Moss). The only exception is Jerry Rice, who played most of his career with the 49ers, where he had two Hall of Fame quarterbacks (Joe Montana and Steve Young). He is the NFL’s all-time leader in every major receiving statistic, and is widely considered the Greatest Player of All Time at any position. But, again, he is the exception, not the rule. Even the greatest offense of all time, this year’s (2013)  Denver Broncos, continued the trend by producing 3 1,000-yard receivers instead of one 3,000-yard receivers.

On the flip side, even the NFL’s worst offenses do not lag that far behind the rest of the pack. A great player on a bad team will still produce like a great player, even if his surrounding talent is bad. Every team has a “go-to” guy, whether it be a receiver or a running back. Ottis Anderson, for example, is one of the most overlooked players in NFL History, played the majority of his career with the St. Louis Cardinals in the 1980’s. This was perennially one of the worst teams in football, yet he is still one of only 28 players in NFL History to rush for more than 10,000 yards in his career.

Another example comes from this year’s Cleveland Browns, who ranked near the bottom of the league in every offensive category, but still produced the League’s leading receiver, Josh Gordon, who gained more yards than big names like Calvin Johnson, or anyone on the record-setting Broncos team. This trend is almost consistent through history as well– bad surrounding casts will not statistically prevent Hall of Fame players from producing like Hall of Famers.

 

Why Yards per Game?

I chose yards per game because it best reflects an receiver’s overall contribution to his team each week. It also ignores historical anomalies that mess with statistics based on “per season” or “per career.” The schedule was not always 16 games. It varied from 10-14 until it was set at 16 in 1978. There were also 2 strike-shortened seasons, in 1982 and 1987. Using Yards per Game eliminates the statistical headache caused by accounting for scheduling differences from different years, and made it easier to compare players from different eras.

 

Conclusion and Next Steps

In this post I outlined how I accomplished the first goal of my research. I determined that there is a Yards per Game threshold that sets Hall of Famers apart, and exactly what that is. I also discovered that the length of a player’s career is also very important. In my next post, I plan to explore player’s career arcs to see what else is different about Hall of Famers. Many things can affect longevity, such as injuries and problems with the law. I also plan to discuss exactly how much qualitative accomplishments like postseason success and League MVP awards can predict whether a player will be a Hall of Famer.