# How deep is a draft?

The upcoming 2014 draft is generally considered to be one of the deepest in recent history. Steelers front office executive Kevin Colbert, for instance, was quoted as saying that this is the deepest draft he has seen since he broke into the NFL. In Colbert’s case, that is a long time, as he has been the de facto general manager of the Steelers since 2000 and entered the league as a scout in 1984.

But how much a difference is there among draft classes? This article addresses the issue from the perspective of both an entire draft class and individual playing positions within a draft class. The metric used as the measuring stick in this article is the number of games started by members of a draft class.

The analysis was restricted to the draft classes from 2004 through 2013. While information would be more complete if older draft class data is included, the older data is also less current and generally of less interest. As a result I decided to restrict the years covered by the analysis to the past 10 years.

**By Entire Draft Class**

In evaluating the degree of difference among entire draft classes, I considered the cumulative number of starts after one, three and five years for each draft class. All 10 draft classes were included in the one-year analysis; the 2004-2011 draft classes were included in the three-year analysis; and the 2004-2009 draft classes were included in the five-year analysis.

The actual years used in the measurements were dependent on the year of the draft class. For example, the one-year analysis for the 2004 draft class included only 2004; the three-year analysis included 2004-2006 and the five-year analysis included 2004-2008. The 2005 draft class analyses included 2005, 2005-2007 and 2005-2009 and so on.

The general conclusion regarding the variation among full draft classes is that there can indeed be a significant amount of difference between the best and worst draft classes. A few more specific observations are as follows:

• The differences among draft classes are the widest when measured at the end of the first year

• The degree of difference among draft classes, as measured by games started, is 15-16% after the third and fifth years

• At this writing, 2006 looks to have been the best draft year out of the past 10

o This is a tentative conclusion as comparisons to the last couple of drafts will play out as those draft classes evolve

• The relative success of draft classes (as measured by number of starts) seldom changes much after the third year

The following table shows the best and worst among the draft classes for each of the time periods measured:

**By Playing Position**

Not unexpectedly, the variations are more pronounced when reviewed by individual playing position. The evaluation is also more complex as the measurement exercise must consider variables such as (1) the number of players drafted at the position, (2) the success of the players drafted and (3) a combination of both.

When just looking at the number of draftees, Wide Receivers have the narrowest range between the highest and lowest number of players in a draft class during the period from 2004 to 2013. For the same period, Running Backs and Tight Ends have the widest range.

The following table shows a summary by position.

It might be expected that the positions with the narrowest range among draft classes might also have the narrowest band when looking at the best and worst of the cumulative games started. This turns out not to be true.

In reviewing the variation among drafts by cumulative games started, I accumulated the number of starts by playing position at the end of five years. (This restricts the analysis to the 2004-2009 draft classes.) The analysis shows pretty wide variations at every playing position.

The following table summarizes the best and worst years among the draft classes by playing position and shows the number of games started at the end of the five-year measurement period. To help focus on the variation among draft classes, the final four columns show (1) the variance between the highest and lowest number of starts, (2) the portion of that variance attributable to the number of players drafted (volume), (3) the portion of the variance due to the quality of the selections and (4) the quality variance divided by the average.

This ratio shown in the final column is intended to give a quick way of assessing which playing positions have the greatest year-to-year variance. The higher the percentage in the final column, the higher the variance for that playing position. A high variance is indicative of a wider range between the best and worst years. For example, there is little difference year-to-year among offensive linemen. On the other hand, draft classes for linebackers can be very good or very bad.

A few observations based on this table:

• I found it very surprising that line backers had the highest ratio

o This is due to the 2006 draft class registering about 1 ½ times as many starts as any other year while staying close to the average number of draftees

• I was not surprised that Running backs showed so much volatility

o The injury factor may be at play here

o The 2008 draft yielded about double the number of starts than any other draft class in the measurement period but 24 RBs were drafted versus the average of about 19

• There is lower variation year-to-year among offensive and defensive linemen than any positions

Follow Tony on Twitter @draftmetrics