NFL Draftnikology: Euclidean Metrics and Other Approaches to Scoring Ranking Predictions

Abstract

With viewership of NFL (National Football League) football in the US rising above 20 million, interest in the NFL Draft has also been at all-time highs in recent years. Much of this interest is directed toward the “NFL draftniks” who provide draft predictions—so-called “mock drafts”—leading up to the NFL Draft. Despite increasing interest in “NFL draftnikology,” the scoring methodology used to evaluate mock NFL drafts lags far behind. This study offers a few alternative approaches, including a Euclidean metrics approach to evaluating mock NFL drafts. The usefulness of our methodologies extends to evaluation of economic and financial analysts.

Keywords:

• Euclidean distance
• Ranking accuracy
• Regression analysis

Mathematics Subject Classification:

• 5IN20
• 62J05

Acknowledgments

The authors thank an anonymous reviewer of this journal for many invaluable comments and suggestions for improving a previous version. The views expressed herein do not necessarily represent those of TSYS, Inc. The usual caveat also applies.

Notes

¹Median attendance across all 32 NFL teams stood at almost 542,000 over eight home games, averaging almost 68,000 per home contest. Even the 25th percentile NFL team produced an average of almost 63,000 fans per home game (www.nfl.com).

Leahy (2010) reports that “an unprecedented” 175 million television viewers had watched at least a portion of an NFL game through the first nine weeks of the 2010 season. The previous record was 170 million in 2009 (Leahy, 2010). Average viewership per game jumped from 16.6 million during 2009 to 18.3 million during 2010, representing an increase of 9.3% (Leahy, 2010). By comparison, average viewership of the major networks’ primetime shows during 2010 was only 8.6 million (Leahy, 2010).

The best-selling sports jersey in 2008 was Dallas Cowboys quarterback Tony Romo's #9 game jersey. Estimates place sales of Romo's jersey at about 500,000 units (Van Riper, 2008).

There is a large and growing academic literature that addresses the development of various ranking systems, particularly those involving sports teams/individuals (for recent contributions, see Beard and Caudill, 2009; Beaulier and Elder, 2011; Caudill, 2009; Caudill and Mixon, 2009). These typically make use of various seminal studies in statistical science (e.g., Berndt et al. 1974; Bradley and Terry, 1952). As the exposition here points out, The Huddle Report's scoring of NFL Draftniks falls outside of this genre. In essence, the scoring systems developed by The Huddle Report and in the current study constitute evaluations of rankings (and/or rankers) against a standard, which in this case is the actual outcome of the NFL Draft (ranking).

In NFL Draft terminology, “Mr. Irrelevant” is the tag commonly applied to the final player selected during a given year's NFL Draft. The 2011 NFL Draft consisted of seven rounds and 254 total selections, with “Mr. Irrelevant” representing the 254th selection. In terms of the hypothetical mock draft discussed here, picks 1–16 are represented by the names from picks 254–239 of the actual draft.

For readers’ convenience/clarity, the exposition of our Euclidean metrics scoring system for mock NFL drafts follows the presentation of the mathematical concept of vector space found in Chiang (1984: 72–74).

In more general terms, the Euclidean distance function explored here, d(A, M), is equal to .

Jesse Bartolis of NFL Mocks finished tied for eighth according to The Huddle Report. His 2011 mock NFL draft was not available for analysis in this study.

Figure 2 Euclidean distance and NFL draftnikology: 2011 results. (color figure available online)

We are grateful to an anonymous referee for suggesting this line of analysis, and for helping shape the discussion that follows.

This mirrors the result, , found earlier when b ⩽ 1.

A Spearman rank-correlations approach would, however, fail to handle an extreme case, such as the hypothetical scenario described earlier in this study, as adequately as its Euclidean metrics-based alternative.

We are grateful to an anonymous referee for suggesting that we formalize the exposition of this alternative metric.

We are grateful to an anonymous referee for suggesting that we formalize this exposition of this alternative metric.