MAVEM: Conference Based Drafting

Love or hate the draft, you know you want your team to draft well or it will face the consequences of drafting the wrong players. To that end, Ethan Young seeks to apply his innovative player evaluation model to college conferences themselves to see which conferences consistently produce the best NFL talent.

As I mentioned in the original article, Modified Approximate Value Evaluation Model (MAVEM) can be used to test any player acquisition hypothesis. As such, I decided to take a look at how each college football subdivision conference stacked up using MAVEM to see if I could find any trends that could be useful for NFL draft strategy. So let’s jump into it.

First, lets see how each conference ranks in average MAVPY (Modified Approximate Value Per Year) since 1999:

wdt_ID Total MAVPY
Total MAVPY

The surprising result here is the Mid-American Conference (MAC), but it’s easy to forget that players like Pittsburgh’s Ben Roethlisberger and Antonio Brown, Oakland’s Khalil Mack, San Francisco’s Joe Staley, New England’s Julian Edelman, and Green Bay’s T.J. Lang all hail from the MAC. And because of the smaller sample of the MAC, these players help disproportionately buoy the conference’s MAVPY up into the range of the Power 5 conferences. But it’s clear that the MAC is no little fish in the NFL draft landscape.

Just to be clear, these averages are set by what conference a team is currently in. So players coming from BYU all count toward the Independent average, even if they went to BYU while they were still in the Mountain West. For reference, the distance between a 2.43 and a 1.41 MAVPY is the difference between the 99th and 142nd pick in value. And that is over an entire conference population.

The SEC has consistently put out the best NFL talent, and while it’s interesting to see that quantified, you likely already thought that. The real question: How is each conference perceived and drafted compared to their actual output? By looking at the draft capital spent on the average prospect from each conference, we can judge just that.

Total Average MAVPY Draft Capital RODC DCAR Total Picked
Total Average MAVPY Draft Capital RODC DCAR Total Picked

While Return On Draft Capital (RODC) may seem like the best way to evaluate how each conference is valued, it’s important to remember that RODC is an efficiency stat, and efficiency is harder to maintain at higher volumes. In this case, the higher volumes are the varying draft capitals between conferences.

For example, the average SEC player is picked around the 84th pick, while the average C-USA prospect is selected near the 139th selection. Mid-third rounders and early fifth rounders have very different expectations and very different average RODC’s attached to them. So by subtracting the average RODC associated with the picks closest to each conference’s draft capital averages from their actual RODC totals, we get Draft Capital Adjusted Return (DCAR). While Return On Draft Capital (RODC) may seem like the best way to evaluate how each conference is valued, it’s important to remember that RODC is an efficiency stat, and efficiency is harder to maintain at higher volumes. In this case, the higher volumes are the varying draft capitals between conferences.





Basically, DCAR removes variables attached to different draft capital amounts, and allows us to isolate actual conference efficiency.

Using DCAR over RODC changes some things. The SEC is head and shoulders over the rest of the power five. The Mountain West fell behind several conferences. And then there is the MAC. While its sample size raises some questions, it is clearly undervalued. Now, I’m not going to advocate for a team to draft only MAC prospects; that would be ludicrous. On average the MAC only has 5.8 prospects drafted per year, so there are not even enough prospects to do that. But, if a team were to maintain that level of efficiency with every pick they made over all seven rounds, the value created would be equal to having an extra early fourth-round pick. The inverse is true of Conference USA (C-USA) of course: If you lost that much value in all seven rounds it would be like losing an early fourth-rounder.

Putting players into buckets provides some interesting data, although not as concrete as DCAR. “Stud” players are defined as having a MAVPY of 7.50 or higher; only 3.6% of players drafted since 1999 are in this tier. “Starters” are players with a MAVPY over 4.00, and “0s” are players with a MAVPY of, you guessed it, zero.

“Busts,” however, are a little different: They are defined as players who have a return on draft capital of -3.00 or below. So players must have a draft capital of 3.00 to potentially qualify. This limits the sample size for the Sun Belt, MAC, and C-USA, so keep that in mind. That means the latest you can be coined a bust is in the third round and, of course, if you fail to meet expectations as a first or second rounder.Putting players into buckets provides some interesting data, although not as concrete as DCAR. “Stud” players are defined as having a MAVPY of 7.50 or higher; only 3.6% of players drafted since 1999 are in this tier. “Starters” are players with a MAVPY over 4.00, and “0s” are players with a MAVPY of, you guessed it, zero.

Now, “busts” can be functional players, especially at the top of the first round, but they signify a sizable loss of investment – Leonard Davis (University of Texas) and Donté Stallworth (University of Tennessee) are good examples of this.

The C-USA has the lowest starter percentage and highest bust rate, which is concerning. It really struggles throughout the entire draft.

Let’s break it down by round:

[dt_divider style=”thick” /]

MAVPY Round 1

So, the Big 10 has been extremely inefficient in the first round. Not only does it have the lowest stud percentage among all conferences, the worst starter percentage among Power 5 conferences, and the highest bust rate among Power 5s, but it’s DCAR is twice as bad as anything we have seen so far. Since DCAR is adjusted for draft capital, we can use it to compare specific round performance across different rounds. DCARs that bad are rare, especially for a Power 5 conference with the third biggest sample in said round. On average over the last 15 years, selecting a Big 10 in the first-round has cost that team a fifth-round pick in value.




The small conferences have done pretty well as a whole, albeit in a pretty condensed sample. While that is pretty interesting, please keep in mind the Sun Belt result is insane and unsustainable (thanks to Troy’s DeMarcus Ware and Idaho’s Mike Iupati).

[dt_divider style=”thick” /]

MAVPY Round 2

The second round has been pitiful for both the C-USA and Independents the last 15 years. Between their 30 selections, they have 12 busts and 0 studs. But check out the MAC! The sample is really small, but let’s not ignore the fact that they actually have the best MAVPY of any conference in the second round.

[dt_divider style=”thick” /]

MAVPY Round 3

With no conferences having a DCAR of at least positive or negative 0.50 in the third round, there is not much to glean here. Although interestingly enough, this is the first time the Big 12 has been efficient so far.

[dt_divider style=”thick” /]

MAVPY Round 4

It’s not possible to “bust” anymore by our working definition, so 0% starts to become a bigger factor. The MAC just smacks everyone in fourth-round MAVPY, even surpassing its third-round total. The volatility of the American Athletics Conference (AAC) is also something to note here, and its percentage splits really showcase that. It’s weird to think the AAC has had three more fourth-round studs than the Pac-12, Big-10, and Big-12 combined, but here we are.

[dt_divider style=”thick” /]

MAVPY Round 5

[dt_divider style=”thick” /]

MAVPY Round 6

[dt_divider style=”thick” /]

MAVPY Round 7

We haven’t talked about the conference a ton, but it’s worth pointing out the consistency of the SEC in these late rounds. It is only one of two conferences to not be inefficient in any of these last three rounds, the other being the AAC.

Let’s talk about the AAC a little more.

If you scroll through these results again, you’ll notice the AAC is always hanging right around the top five. The notable exception is the second round, but it is consistently slightly undervalued through the draft.

And just for fun, here is how the football championship series conferences stacked up.

Unfortunately these sample sizes are too small to really analyze in depth. The Big South has performed at an unsustainable rate, which makes sense considering it’s laughable sample size. Even if we were to find something interesting, there are just not enough players from each conference being drafted to apply any findings in a worthwhile manner.

Maybe we can use this data in another way, though.

Let’s look at all the players from “small schools,” i.e all the FCS conferences, C-USA, MWC, Sun Belt, Independents, and, heck, we’ll even throw the MAC in there. Basically, everybody but the Power 5 and AAC.

I analyzed a couple aspects of this group, but one approach that gave some interesting results was applying Positional Slaytic Thresholds to this group. Since 1999, only one “small schooler” that failed to hit the lower bound PST has ended up reaching the MAVPY starter level described above (former Pittsburgh Steeler defensive end Aaron Smith). It appears having bad measurables and playing at a smaller school is not a good blend for finding NFL starters. This is obviously bad news for small school players that failed to hit the PST this year – like Baltimore’s Matt Judon (Grand Valley State), Arizona’s Harlan Miller (Southeastern Louisiana), Minnesota’s Willie Beavers (Western Michigan), San Francisco’s Ronald Blair (Appalachian State), New England’s Cre’von LeBlanc (Florida Atlantic), and Seattle’s Rees Odhiambo (Boise State). Baltimore’s Kamalei Correa (Boise State) would be in this group as well if he stayed at EDGE, but it appears he is moving to ILB, where his measurables tested in an acceptable range.

Anyway, that is a lot of conference-based intel for you. It’s important to note that I am not advocating anyone to go crazy with this information. Draft picks are micro events, and an NFL team shouldn’t take what it views as an inferior player based on macro conference data. What we did identify (avoid the Big 10 in the first round, avoid Conference USA in the second round, small schoolers with bad measurables rarely become starters, and that the MAC has been a lot better than we think, with many other smaller notes) can be implemented in an overarching draft approach, but this is ultimately an informative aid for a decision maker rather than a decision maker itself.

Follow Ethan on Twitter @NFLDrafter. Also check out his work on a new, more accurate and more versatile player evaluation chart and a better way to predict QB sacks.

Want more Inside the Pylon? Subscribe to our podcasts, follow us on Twitter, like us on Facebook or catch us at our YouTube channel.

Leave a Reply

Your email address will not be published. Required fields are marked *