Fighting the Conventional Wisdom about the Draft

Ajit Kirpekar sent in this piece last week.

 I am perfectly happy with the Costanzo pick.

Regardless of the final outcome, as of now, I believe it is certainly the right move. With that said, some of the arguments I’ve heard before the draft propelled me to write this.

A friend and I were discussing the 49ers and who they should select. He, along with seemingly the rest of the country, was in love with Patrick Peterson and thought Peterson should be the no-brainer pick. He felt this way even if the Niners had had the first overall selection. I felt differently, however. I knew CB was a need, but to me, the 49ers are a talented team in a bad division who are fundamentally weak at some very important positions (Wr, QB, and 3-4 LB). 

As a result he and I started to look back at some of the numbers and play around with things to determine relative value.

 The age old question is ‘need versus talent’. I decided to run correlation studies for each position and its impact on that team’s given output. For instance, I ran a test to see how much quarterback improves a team’s passing game (a lot as you would guess) or how much a FB improves a team’s rush stats (not a lot actually). My data uses a combination of FO’s DVOA charts along with three years of Profootballfocus stats. I want to briefly acknowledge that there is certainly an argument to be made about the potential conflicts between the data. For instance, FO and PF both grade differently and have differing criteria. For instance, FO uses standard statistics and puts them over an average. PF uses subjective grading but takes many more mitigating factors into account. Even still, I thought it would be a fun exercise.

I’ll be using the following abbreviations: QB (duh), WR (all wide receivers), CB (all corners), S (for both safeties), DT (4-3 and Nose), TE, RB, OT (offensive tackle), OL (Center and Guard), LB (meant for 4-3 linebackers or 3-4 linebackers that aren’t necessarily pass rushers), MLB (inside linebackers), DE (3-4 Defensive ends) and PR (all 4-3 and 3-4 attack pass rushers).

Because both sites are commercial, I’m not comfortable pasting the exact numbers. But here’s what I found. The positions that tend to have very high impacts are QB, WR, DT, and PR. Some of these are no-brainers, as great pass rushers are generally on teams with high pressure per attempt totals. Ditto for a quarterback on a team’s passing game. Wide receiver is harder to parse because so much of their statistic is determined by the quarterback, but I came up with my own stat that I’m calling DYAR plus/minus (which takes a wideout’s DYAR and adjusts it based on the context of his offense). Aside from the Colts and Patriots, generally, teams which had good receiver DYAR plus/minus were also on good pass offenses.

All other positions (CB, S, TE, LB, MLB, OT, OG) tended to have less correlation. Yes, even OT and CB.

Why? Let’s examine: For a corner, I ran a correlation for Profootballfocus’s corner coverage stats against Footballoutsiders’s defense against number 1 2 and 3 wide receivers. I got a very weak relationship, signifying that covering wide receivers is more than just the responsibility of a cornerback. It tends to incorporate the linebackers and safeties as well as having an overall solid core of players in place.

This is also true for offensive linemen. Yes, good offensive linemen usually belong on good pass blocking or run blocking teams. But the problem is, each position, by itself, had no correlation. For example, teams that had a good pair of tackles might still have poor pass blocking or run blocking metrics simply because the team’s guards and centers are not very good. The reverse is also true. The point is not that these positions are meaningless; it’s just that they have less impactl than the one’s mentioned above.

I have two additional points. You may notice that I didn’t list either DE or RB in either category. For 3-4 defensive ends, it’s a position that had players that had high impact, but typically had low snap counts. Thus, while their pressure rates were good, they didn’t have an overwhelmingly large share of the team’s total pass rush number. This is also true for their run stopping numbers.

As running backs, it did come out as significant, but I wouldn’t necessarily call it a position teams should aim to draft high. This is true mainly for three reasons. First, running backs tend to have a very short shelf life and suffer very much from year to year regression. Think about every great season an running back  has and invariably, the next season is usually something of a slump. Second, it’s relatively easy to find competent backs than other positions. Again, think about how many top runners this year did not come from the first round (Forte, Foster, Charles, Hillis, MJD, Bradshaw, Blount). Finally, the evolution of the NFL has made great run games less important to team success than ever before. A run game now has become just another important component, but definitely pales in comparison to a good passing game or a great pass defense.

So what does it all mean?

We believe that an awful team, any team that has severe talent deficiencies that show up on both sides of the ball (think Denver, Buffalo, Arizona, Washington), ought to be drafting need with the four big positions: QB, WR, DT, PR. This is usually what happens, except teams also draft offensive tackles very high. It is our opinion that very bad teams are making a big mistake drafting offensive tackles in the top 10. Think about all the years those bad teams took tackles in the top 5 and how much they have to show for it. The Dolphins and Browns both hit on their left tackles and their Offensive DVOA has been no better than mediocre. And there are a slew of other teams that have taken tackles in the top 5 and are perennially atrocious.

Now, for teams that are decent to good to even great, at this point, we feel you should draft the best available player. This is where the rest of the positions should go in order of value (unless of course you are absolutely stocked at that position): CB, LB, S, RB etc.

As for the Colts, they took the best available player which is why I was happy overall with Costanzo. He was the best available tackle while also the best overall player left to them. What frustrated me, however, was how pundits all over felt regardless of who was available, tackle had to be the pick. If Amukamara had fallen and Costanzo was off the board, should the Colts still have drafted Tackle? And would Costanzo have been more valuable if say, Von Miller had somehow fallen to us?

The ultimate point is, Costanzo was the right pick, but it has less to do with his position and more to do with his value at that point.