Finding the best value on Day 2 of the NFL draft

Tom Brady, a three-time Super Bowl winner, is perhaps the greatest find ever on Day 2 of the draft. Al Bello/Getty Images

If you watched Day 2 of the NFL draft from beginning to end this year, raise your hand and join your local football addiction support group. Day 2 doesn't have any of the glamour or glitz that the draft's first day gets, but it could very well be the guts that your favorite team is built on. Ask the Super Bowl champion Giants, who got key contributions from 2007 Day 2 picks such as running back Ahmad Bradshaw, tight end Kevin Boss, linebacker Zak DeOssie and safety Michael Johnson. Their opponents from New England last season started six Day 2 picks, including some names you might be familiar with: Asante Samuel, Adalius Thomas, Rodney Harrison and Most Valuable Player Tom Brady.

Of course, teams are less likely to find stars when picking through what's left on Day 2 than they are in the first round. That doesn't mean, however, that the later rounds are a crapshoot. A close analysis of how players drafted on Day 2 from 1995 through 2007 have performed in the NFL reveals trends and truths. (Day 2 picked up the third round in 2008, but because we're looking at the past, we'll limit our analysis to picks from the fourth through seventh rounds.)

Defining how successful a pick is a difficult thing to do in and of itself. The simplest way of comparing value across positions is to use games played and games started; if a guy keeps lining up Sunday after Sunday, he has to be worth something. Likewise, if a player lasts only a year or two before exiting the league, his team suffers and is forced to take another player to replace him. That's why the metrics we'll use, games played and games started, are going to be divided by the number of seasons since the player's been selected, the number of seasons he's been eligible to play. This provides an effective way of comparing players who have played 12 seasons to those who have played only one or two.

Another issue with value is that we have to measure how successful a player is relative to other players at his position. To do that, we compared each player's games started per season eligible to players selected within 10 selections on either side of his draft position. (Players taken with the first 10 picks were compared to other top-10 picks.) Subtracting that from their own games-started data reveals their position-adjusted games started per season above average. We'll call it GSAA for short.

It sounds way more complicated than it actually is. Take Tom Brady. The 199th overall pick of the 2000 draft, Brady has started 110 games in eight seasons, or an average of 13.8 games started per season. On average, the other quarterbacks taken from 1995 to 2007 with picks ranging from the 189th to 209th picks have started only 0.5 games per season eligible; therefore, Brady has averaged 13.3 GSAA per season.

That figure ranks in the top 10 games started per season above average of all players in the timeframe, but it's not No. 1.

As one of the last picks in the fifth round in 2003, David Diehl barely had any business even making a NFL roster, let alone becoming a starter. Despite all that, the Giants left tackle's versatility and work ethic has led him to start every game since he's been drafted. Brady finishes sixth on this list.

Obviously, we're not saying that Diehl was a better pick or more valuable than Brady, since we're not adjusting the games each player started for their level of play. Guys such as Jason Taylor, Jared Allen, Zach Thomas and Marques Colston would all be in any discussion of the best Day 2 picks of this time frame, and have GSAAs ranked in the top 10 for their given year; this metric isn't designed to settle arguments about one pick being better than the other, but, instead, to help identify inefficiencies in the marketplace and player/player types who are more or less likely to produce similar talents in the future.

Take a look at the conferences in that top-10 list, for example. Nine out of the 10 players come from a major college conference, with only Houston defensive back C.C. Brown coming out of a small school. Using GSAA, we can analyze whether picks from certain schools are more or less effective than others. Separating the draft pool into a group of players from the "major" conferences and a group of all others reveals that players from major schools selected on Day 2 slightly outperform their brethren:

If we separate that group of major conferences into its six parts, we can also see that two conferences are much better at producing Day 2 talent than the others.

The SEC and the Big Ten produce, by far, the best Day 2 selections on a pick-by-pick basis. In the past three years, the Big Ten has produced players such as Texans TE Owen Daniels, Giants DT Barry Cofield and Cowboys All-Pro RB Marion Barber in the later rounds. In that same span, the SEC churned out Titans T David Stewart, Cowboys DL Jay Ratliff, Eagles LB Omar Gaither and Bears sack artist Mark Anderson.

Finally, when ignoring conference and looking just purely at the position of each player, we can identify which players are more likely to actually become starters when chosen in each round. Since GSAA already accounts for the effects of position, this table is strictly each player's games started per season eligible.

The huge drop-off in games started for players taken in the fourth round as opposed to the rest of the draft illuminates how important those picks are. Tight ends, offensive linemen and linebackers tend to be the most successful players, while defensive backs see the widest variation in value. While there is the rare skill-position player who succeeds (you'll note a "Tom Brady" spike in sixth-round games started for quarterbacks), fliers taken on those players tend to be for naught.

Next week, we'll highlight ten Day 2 players from the 2008 Draft who our research projects to be successful in the NFL.