This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.

AFL versus NFL: draft class peaks and declines

Posted by Jason Lisk on September 17, 2009

Before moving on to the draft classes of 1967-1969, I thought I would do another quick thought experiment, and while this post is included in the AFL versus NFL series, it certainly has broader application. Even if you are not interested in the specifics of the AFL versus NFL breakdown, you may find it informative as you assess how much the 2009 season will be determined by the 2007 draft class versus the 2002 class.

I've already looked at the draft classes of the early (1960-1963) and middle (1964-1966) of the decade. But how the heck am I going to combine all of that information to come up with a coherent overall view of how the two leagues compared? It's not going to do a lot of good to simply know how much talent the AFL got in 1960 versus 1964 if I cannot combine it in some meaningful way.

In order to do that, I'm going to look at the behavior of draft classes post-merger. Approximate value is a great tool for this kind of project, because we have really large sample sizes, and it combines years starting and playing in the league, pro bowls and other honors, and whether and when a player was most likely to be playing on a productive unit into one number.

I've pulled the AV data for every draft class from 1970-1994, and found the cumulative AV and total number of contributing players from each draft class for each season from year 1 (rookie year) to year 15 for each class. The following chart shows the average total AV for each year, and the average number of "starting" players, where "starting" does not just include the primary starters, but rather all players who contributed enough to earn at least 1 point of AV in a season.

Yr	AV	no. of players
1	397	138
2	584	147
3	636	137
4	633	125
5	583	111
6	519	96
7	447	81
8	371	67
9	295	53
10	224	40
11	165	29
12	113	20
13	69	13
14	44	8
15	26	4

We see that a draft class typically peaks in terms of total number of players in the second season (the allure of potential and promising players and such), and decays from that point on. In terms of overall value, the peak is typically in years 3 and 4. Now, let's convert that value chart into a percentage of total AV for a season. After all, I'm going to want to look at, say, the 1966 season, and consider how much the 1964 draft class should be weighed relative to the 1960 draft class, or the pre-AFL draft classes of the 1950's. This chart converts the raw average AV totals for each year into a percentage of overall AV:

 Yr	Percent AV
1	0.078
2	0.114
3	0.125
4	0.124
5	0.114
6	0.102
7	0.088
8	0.073
9	0.058
10	0.044
11	0.032
12	0.022
13	0.014
14	0.009
15	0.005

55.5% of a season's value comes from players in the first 5 years of their career. However, we see that the average AV per player is lowest for the early draft class years, because far more of these players are serving as backups and part-time starters. Because I want to compare two separate leagues against each other, I'm as concerned about quality as quantity. The chart above is more of a quantity measurement, showing us in which years draft classes contribute the most raw numbers. So, I then created a "quantity + quality" chart, where I took the raw average AV for each year, and subtracted out [2 AV *# of "starters"], to try to account for value over "part-time starter". Here are the new percentages for each year of a draft class:

 Yr	Percent AV
1	0.041
2	0.098
3	0.122
4	0.129
5	0.122
6	0.110
7	0.096
8	0.080
9	0.064
10	0.049
11	0.036
12	0.025
13	0.014
14	0.009
15	0.006

With this new "quantity combined with quality" chart, year 4 is the peak year for a draft class, with years 3 and 5 being equal and close behind. Years 2 and 7 for a draft class are now of roughly equal importance. The rookie season (when there are a lot of contributors but only a handful of stars) is now of equal importance to roughly the tenth and eleventh seasons (when there are few contributors but most who remain are stars and future hall of famer types). This feels more right for what I'm trying to model. Let me know what you think. I will probably use some version of that last chart to model for the AFL versus NFL conclusions.