But there is a very, very simple measure that will give you a better sense of whose draft will be good and which ones will be bad: just count how many picks each team makes. It may seem too simple to be true, but in looking back at the 2004, 2005 and 2006 drafts with the benefit of our current hindsight, you would have done a better job of ranking the drafts from No. 1 to No. 32 by simply counting each team's number of picks, rather than listening to the draft experts' instant grades.
We'll include all the statistics later for those who want to check the math, but we ran the drafts through two formulas to determine successful vs. unsuccessful drafts. The first gave points for rankings for games played (by draftees), games started and career approximate value with bonuses for Pro Bowl appearances (five points) and first-team All-Pro appearances (10 points). The best team in the league in each category received 32 points with descending scoring so that the worst team in a category received one point. It was the same formula that we used to grade the drafts for our post-draft grade report yesterday.
Using that formula, the stats showed a statistically significant result. But we were worried that the formula was weighted to help teams will large numbers of picks because the rankings included games played and games started. Theoretically, a team who drafted one or two stars may not be given enough bonus over a team with no stars but many backups.
So to try to make sure the formula wasn't skewing the result, we also ranked the team's drafts solely by approximate value, a formula that was developed by Pro Football Reference. Career AV should ensure that a team who drafted few but very successful players would be just as credited as a team with lots of mediocre draftees. The results? No real difference than the first formula -- it again shows that there is a small but significant correlation between the number of draftees and a draft's success.
In fact, among the top 25 drafts over those three years, not one came from a team that had less than the normal seven draft picks in the seven-round draft and 17 of the top 25 were by teams that had more than seven picks. At the other end of the spectrum, nine of the worst 25 drafts were by teams with less than seven picks; seven teams had seven picks and nine had more than seven picks.
|BIGGEST 2010 DRAFTS|
|SMALLEST 2010 DRAFTS|
|New York Jets||4|
Here are some examples of the best draft bonanzas from 2004-2006.
San Diego 2004: 11 Picks. Drafted Eli Manning (1), Igor Olshansky (2), Nate Kaeding (3), Nick Hardwick (3), Shaun Phillips (4) and Michael Turner (5).
Yes the Chargers immediately traded Manning for Phillip Rivers, but that's proven to be a near even swap. Whether you count it as Manning or Rivers, the Chargers picked up six NFL starters in one draft. Whenever you do that, you're doing a great job of laying the foundation for a playoff team.
Tennessee 2005: 11 Picks. Drafted Pacman Jones (1), Michael Roos (2), Brandon Jones (4), David Stewart (4), Bo Scaife (7), Reynaldo Hill (7).
Jones has proven to be a major headache but, before the Titans cut him loose, he was productive. Roos and Stewart are the two anchors on the NFL's best offensive line while Scaife has been a very productive tight end. Getting a pair of Pro Bowl-caliber offensive tackles in the same draft would have been enough, getting additional help is just a bonus.
Tennessee 2004: 13 Picks. Drafted TE Ben Troupe (2), DE Travis LaBoy (2), DE Antwan Odom (2), DT Randy Starks (3), OG Jacob Bell (5), C Eugene Amano (7).
This is an example of what you sometimes get when you have 13 picks. Tennessee didn't have a first-round pick and they didn't land any future Pro Bowlers (although Antwan Odom was on track before his injury last season), but the Titans landed six players who started at least two seasons in their career. Pair this with the 2005 draft and Tennessee picked up the guts of a 2006/2008 playoff team.
And now, I'll defer to my partner on this project, Greg Stonerock, Eagles fan and stats whiz, who ran the numbers to explain how we determined that this is statistically significant:
Keeping it simple, a correlation analysis tells you if one variable tends to change when another changes. For example, say highest yards per pass attempt (YPA) usually comes with more wins, and low YPA to fewer wins. Wins and YPA would be positively correlated. If teams with fewer penalties consistently win more games, they're negatively correlated. In the end, we get a "correlation coefficient," a number ranging from -1 to +1. The further away you get from 0, the stronger the relationship. If it's near 0, then one variable tells you next to nothing about the other.
The correlation between number of picks and the draft's final value was about +.33. With three years' worth of draft data, you'd get a number that high by chance alone about 1 in 1000 times. (If you're stats savvy, N = 96 and p = .001.) Given those odds, it's pretty safe to say that there's an actual relationship here.
Squaring the correlation coefficient tells you how much of one variable is explained by the other variable. Here, squaring +.33 gives you about .11. So, you can account for 11% of the variation in the final value of a draft class just by knowing how many picks they made. 11% ... is that big enough to matter? Well, the same stats found that instant post-draft grades accounted for less than 1% of actual performance in the end. That makes 11% look pretty good to me.
Correlation isn't causation, so this draft doesn't lock the 2013 Eagles into hoisting the Lombardi Trophy ... but it's a good start, and a man can dream.