Re-Re-Grading Chris Ballard’s 2017 Draft Class

With 1 full season in the rear-view mirror, Chris Blystone recently took a look at the 2017 Colts draft and re-graded the players, providing an overall appraisal of GM Chris Ballard’s first selection efforts. In that article, Chris lowered Ballard’s draft from an A to a B- and while I have no qualms with that assessment, I am not an X’s and O’s guy. I am a 0’s and 1’s guy.

So, while grades will satisfy many, I need numbers to satiate my curiosity. I need to multiply stuff and make charts and tables and try to wedge the word heteroskedasticity into the conversation. As such, I am supplementing Chris’ evaluation by utilizing my net AV(1) methodology to judge Ballard’s draft class and make a numerical comparison to the rest of the league.

Now before we dine with the data, I will make the caveat that AV isn’t the greatest stat in the world. It has a lot of weaknesses that make what I am trying to do questionable.

Also one year is waaaaaaaay too soon to judge a draft class.

Also I’m not a statistician.

But if you close your eyes, it’s almost like those issues don’t exist. So let’s do it!

---

THE ROOKIES

As far as actual performance, it was not a very good year for the Colt rookies. Malik Hooker played well when he was on the field, but missed about half the year due to an ACL-MCL tear, while Zach Banner missed the whole year due to being on the Browns. Anthony Walker was limited because of injuries while Quincy Wilson was limited because . . . reasons.

With only 4 fully active players, 3 of which were from the back half of the draft, the Colts rookie class wasn’t likely to set any records. And they didn’t.

Marlon Mack was the only player to accumulate a respectable AV total but his touches were restricted as Frank Gore collected the lion’s share of the carries.

After subtracting the expected AV “cost” of the players(2), Ballard’s class earned 4.7 AV less than expected. That means that based on their position and where they were picked in the draft, the Colts rookies performed at a lower level than the historical average player at the same position from the same pick in the draft.

Historically, that is around the 20th percentile and for 2017, it places the Colts in a disappointing 27th place.

The Saints had a phenomenal rookie class, curb-stomping everyone in net AV led by an outstanding year from 4th pick Alvin Kamara and solid hits on their first 3 picks (Marshon Lattimore, Ryan Ramczyk and Marcus Williams).

At the other end of the spectrum, was CAR who had decent output from Christian McAffrey but their only other productive pick was 7th round kicker Harrison Butker who did well . . . for the Chiefs.

---

LUCK ERA

Expanding the view back to 2012, the Colts rookie classes have fared much better overall. As bad a GM as Grigson was, his rookie draftees ranked higher than most teams. He had a bad 2013, but his 2012 and 2016 drafts performed well enough to lift the Colts to a cumulative 3rd place.

This really is not a very meaningful stat though. If you look at the table closely, you may notice that a lot of good teams have poor rookie performance and a lot of bad teams got good play out of their 1st year draftees.

The correlation of total rookie net AV to total wins is -0.11. In other words, a coin flip tells you almost as much about wins as rookie net AV.

But that isn’t true when expanding the measure beyond their first year. Using career AV(3), the correlation of net AV to wins increases to 0.39, which is in the semi-respectable region. On that scale the Colts rank 14th, which is pretty close to their reg season win total rank of 12th.

---

SO WHAT?!

AV doesn’t measure a player’s potential or front office strategy or any of the really important stuff. It just measures what happened and what happened wasn’t good. There’s no real sugar-coating that part, Ballard’s first draft class had a bad first year.

Fortunately, the data also tells us that this one year performance isn’t very relevant to future wins. So while the first step was a little shaky, there is plenty of time for this draft class to prove itself.

---

FOOTNOTES

1) Net AV is the difference between actual AV earned and an expected AV based on draft pick number and position. There is always an expected AV cost for each player, but AV is only accrued if the player remained with the drafting team.

2) Expected AV is a log-linear regression of 2012-17 average first year AV values by draft pick number and with separate regression curves by listed draft position.

3) Career AV was calculated by taking the best 3 years of a player’s AV performance. Players from 2015 - 17 drafts utilize all of their earned AV.

As always thanks to Pro Football Reference and Armchair Analysis for being excellent data sources.