NFL Teams Know How To Draft - Why Don't They?

So every year, hundreds of athletes are drafted to play in the National Football League. This isn't some draft for war. These athletes dream of being on the stage. These organizations dream of finding their next star. Billiondollar entities deploy small armies of scouts and analysts, sorting through thousands of college players.

But despite these multi-million dollar staffs and an infinite ocean of data, teams are still making numerous drafting mistakes—more drafting mistakes than perhaps any other major sport. It's not that these teams lack intelligence. We never look at a situation and say, "Oh, it looks like we're we're all set at this position." We need to evaluate our roster and know our players so we can make our roster better moving forward. That's not what surprised me. What surprised me was they know how they should draft, but they don't.

I spoke with a former GM, a former analyst with PFF, and a former NFL scout. And there is much more to this than I initially thought. I want to show you how your team should draft and, importantly, why they probably won't.

First choice in the first round: John Matuzac, defensive end, University of Tampa. "Who'd want to watch that?" That was the response from the NFL commissioner, Pete Rozelle, when a little-known network, ESPN's Primetime, asked if they could televise the NFL draft. Things have changed since then. "What's up Detroit, let's go baby!" Today, the public not only spectates—listen to this crowd go wild for you—they are far, far more informed.

"It's infinitely more data than there's been in the past." That's Kevin Cole, a former data scientist for PFF, a company that analyzes player performance and sells information to NFL teams. You may have seen their famous player rankings while watching a game. How they're graded is something in consultation with NFL coaches and, if it's for college, you know, scouts and others. They try to have semi-rigid frameworks for what particular players are doing on particular plays based upon their role in that play. This data has only compounded, starting first with player tracking in the NFL and now player tracking in college, giving these organizations an even larger workload.

"You're asking a football team to essentially be a research institution. They don't have the compute power. They don't necessarily have the data infrastructure for it." Sean Clement was nice enough to take time with me as well. His resume includes NASA, the Miami Dolphins, the Ravens, the Cincinnati Reds, and now Sumer Sports, who sells their services to NFL teams. "Just how big are these staffs? Probably in the ballpark of 50 people." There's a lot of variation between teams. Some teams run very lean on analytics or very lean on pro scouting. Teams are fully aware of the importance of the draft. In many ways, it's more important than any other major sport.

It's no secret that tenure is a rare thing in the NFL. Losing a step that often means losing your job. With such high player turnover, organizations are forced to replace chunks of their team every single year with new talent. And by far the most cost-effective way to do that: the draft. Rookie contracts are like college grads—cheap labor. You talk about the real money in sports—that's the second contract. And in a league where there is a hard salary cap, unlike the MLB or the NBA, you can't just outspend your opponents or recruit players to South Beach. This is a socialist enterprise. It's not as simple as just being the Dallas Cowboys. If you want to win, balls back knocked out of his hands, Cowboy linemen, football follies, and maybe Philadelphia has it.

But because the NFL is so dynamic and contracts are so confusing, very few people understand how this cheap labor works and what teams should do with their draft picks. About a year and a half ago, I put together a video detailing out the NFL draft, highlighting why I thought the Bryce Young trade was a sizable overreach by the Panthers. My judgment was largely based upon Richard Thaylor and Cade Massiey's work, the losers' curse, which quantified how teams value picks based upon historical trades. Essentially, teams significantly overvalue early first-round picks, often giving up far too much to trade up. Due to the randomness associated with finding good talent, they should be doing the opposite: trading down and/or trading to the future to get more at-bats and more picks. Other charts illustrate similar findings. If you haven't seen or heard of this, it might surprise you.

Let's take a look at Kevin Kohl's here real quick. Player value. First, we see the obvious: on average, the earlier you are picked, the more valuable you are to your team. For value on our y-axis here, we have value in percentage of salary cap. And what that allows us to do is compare across different years because it's a percentage. And since the salary cap changes over time, next we see contract cost. Even more obvious: players who are picked earlier cost more. But critically, it is the third line here, this green line, surplus value, where we start to see something truly confusing. There's a curve suggesting that early first-round picks are actually less valuable due to the spike in cost, in this case, salary. This is even more evidence that early first-round picks should trade back or trade to the future. If we look at the peak here, it's in the middle of the first round. It's really the difference between how much value that you're getting with a draft pick versus what you're having to pay for it. And those draft picks have this fixed cost associated with them.

So to recap the initial evidence from Thaylor and Kevin Cole's graph: number one, teams overvalue early pick performance in the form of player salary; number two, teams overvalue early pick performance in the way they trade with other teams; and number three, teams overvalue the present. This doesn't seem like how NFL teams draft. Here's the thing with all that though: everybody knows it. It's conventional wisdom in the NFL. I can assure you every NFL franchise is aware of what I just shared with you. So if everyone knows about it, why is nobody taking advantage of it? Well, as Kevin mentioned to me, there are pieces to it that even the nerds get wrong.

Although Thaylor and Massiey's work did a good job of illuminating the inefficiency of the draft, there are some caveats. For example, even though you might have a higher average value of players by trading back and maybe just getting more at-bats, there can be maybe an overlooked premium to truly high-end elite talent. Reason being: roster constraints. You can only play 11 guys at a time. Depth is important, but if you want to win, you need stars. As illustrated by Michael Lopez, the probability of finding a superstar player rises much more steeply at the top of the draft than that of finding an average player. Importantly, you aren't just more likely to find a star at the top of the draft. That's obvious. You are exponentially more likely.

To give you an example, let's say a team trades pick number five for two late first-round picks in the same draft, number 31 and number 32. Using Lopez's work, here's what that would look like: the number five pick has a 30% chance of finding one star, a 17% chance of finding one average player. We can assume that the rest is basically a bust. For the second scenario where you have two late first-round picks, number 32 and 31, you have a 31% chance of finding one star and a 23% chance of finding one average player. You have a 3% chance of finding two stars and a 2% chance of finding one star and one average player. About a 1 and a 1.5% chance of finding two average players. Looking at historical data, the number five pick is about 36.1 million. The two late first-round picks would be about 26.7 million in total. So in this example, the odds of getting a star aren't all that different, but it would still point to doing one thing: trade down. Reason being: salary is quite a bit less and you have two players. But this assumes something really important: that you always have someone to trade with. Surely there are more trades proposed than than actually happen. Who knows how many uh, get proposed that that aren't actually accepted. We can just look at this year's draft as an example. After Sanders, there isn't really any elite talent that teams are probably willing to trade up for, meaning there's a pretty sharp cliff. This varies dramatically year to year and can certainly affect the trade markets. Last season was different. Chicago Bears select Caleb Williams. There were six quarterbacks taken in the first 12 picks. Why was this? Well, we need to look at the surplus curve again. This surplus curve is an average. Positionally, it varies much, much more. All the way at the top, one position is dwarfing the others. The Kansas City Chiefs no longer have a first-round pick. Wow, they just traded it. Even averaged out for all of the draft picks. With the 10th pick in the 2017 NFL draft, the Kansas City Chiefs select Patrick Mahomes, the second quarterback. In 2021, we again had a pretty substantial draft with a lot of different quarterbacks through M. Jones—I think it was five quarterbacks in the first 15 picks. So yeah, teams are starting to realize this, that this is who they want to go after.

At first, this is just intuitive. Quarterback is the most important position, but we have to remind ourselves of something here: it is the most difficult to predict and the most expensive. So how is there still this big of a difference? While it's true that the majority of first-round quarterbacks don't really pan out, for those that do, it is a massive, massive bargain to the team—nearly double what all positions lumped together are. Take for example Jaden Daniels. He was the second highest paid player in his class: 37 million over four years. That averages out to about 9 million a year. For reference though, Gino Smith, not a bad quarterback, he's the 16th highest paid quarterback. He comes in at a whopping 37.5 million a year. In other words, a middle-of-the-pack quarterback will be making 4x what a top five QB will make. And that will continue for the next 3 years. With this centerpiece, the Commanders can now build around Daniels in ways that very few organizations can. And it doesn't just end when that rookie contract expires. If you have a quarterback, unless there's a real, you know, weird type of blow-up between ownership or coaches, you should be able to have that quarterback for as long as you want. And if they do leave, normally it's some sort of mutual decision late into their years. Whereas in other sports, you see superstar players who would be the equivalent value of a quarterback on an NFL team be able to leave all the time or be able to get traded because they have the threat of being able to leave. From day one, you draft a quarterback, you say, "We got you for five years and we can realistically franchise you another two years after that where you're getting paid out on a year-by-year basis. You're taking the injury risk." You're taking all those types of risks. So you know you're going to come to the table and you're going to eventually say, "Okay, I want the huge payout. I want the huge contract. I'm not going to keep on playing this out forever." So you can keep that one particular player the entire time. Thus, lumping quarterbacks in with other positions doesn't really make sense. And this highlights our misunderstanding of risk. Although taking a quarterback is viewed as a risk because of the high failure rate, you also have the highest upside potential. By far, not taking a risk is still a risk.

Looking back at the curves, we can see why teams don't take linebackers or safeties in the first round. It's not pretty for him. There just isn't the same value in comparison to the other positions. In other words, you can find them for cheap in free agency. The same cannot be said for offensive tackles or defensive linemen, which is why we find the first round littered with these positions. These are the facts. Certain positions, they're more important than others. We all know this. And because of this, the risk-reward justifies taking these players in the first round, even if you're more likely to miss. But this is not a hall pass for rational behavior; we still haven't fully answered the question of why don't teams trade down more often or to the future.

I want you to imagine that you've just been hired by the Jets. That might seem like a curse, but you're actually hungry and excited, and well, the money ain't bad. Your job is clear: build a team that gives the organization the best chance to win. Knowing what I have just shared with you, you immediately put together a proposal that suggests a very rational approach. Predicting talent is hard, so let's give ourselves the best odds by trading our current picks for more future picks. The owner agrees. Quick rule of thumb on how this works: if you trade a second-rounder this year, you can expect a first-rounder in return for next year's draft. It's usually about a one-round difference for the year. You go to the extreme with this. You trade all of your current picks for future picks, piling up two first-round picks for the following year and twice as many expected draft picks. Additionally, you trade a few older players since you're not trying to win now. The New York Jets are going in a different direction, and you gather three more draft picks. After comp picks, you've amassed a total of 19 selections. You're sitting pretty. The current team isn't though. They finished dead last. The fan base isn't thrilled, but they can see what you're doing. It's obvious a rebuild is okay for the first year on the job. You get the number one pick and you draft your franchise quarterback. Following the same rationale, you trade out and up again, but this time only three mid-round picks to next year, still moving up in the process for the following year's draft. You build out the rest of your team with your pile of picks at a ridiculously inexpensive venture. The team improves but still struggles, finishing fifth to last in the NFL. That's okay though. We knew this was going to take some time. I mean, these are rookies after all. Year three rolls around, a new hall of players, and your team finishes 15th—an improvement, 500, narrowly missing the playoffs. Unfortunately, you didn't make the end of the season because, well, you were fired when the team started 4 and 6, as was the rest of the coaching staff. Three straight losing seasons. That's unacceptable. Although your team showed considerable improvement, your pile of picks that you traded to the future for would be entering their golden surplus years of their rookie contracts, year two and three. You won't get to see that. The public and owner lost their patience and trust with the process. A new GM rolls in with a loaded cheap roster, fills gaps in free agency, picks up a few players on the waiver wire, trades future picks for now, and well, takes your $2.5 million salary. They go on to win the division. Everyone credits the new coach for changing the culture.

The incentives: NFL GMs have the shortest tenure in all of major sports. Despite the fact that there's a lot of luck and randomness involved with draft picks and even winning football games, the incentives question: what are they optimizing for? They're they're optimizing for the moonshot, like they are trying to get the best possible outcome, even if if it's a higher risk—most hits are still misses according to their incentives. Convince a general manager to make the quote-unquote optimal decision if they think the optimal decision will get them fired. Well, that's why you see those teams kind of build themselves back up in a way where teams that change general managers or change head coaches all the time never quite seem to hit that stability because you at least provided better incentives. You're not constantly fighting for your job every single year. "Hey man, here you go. No one is safe."

"Hey, Howie Roseman," widely considered the current GOAT of GMs, had his name echoed and chanted throughout the city. He crushed it in free agency, with trades, through the waiver wire, and no doubt about it, in the NFL draft. But not in a positive light. These people were calling for his neck at a different sporting event. There is no patience for the tank. Even if your team could see short-term turnaround, you need to win now. I mean, just take a look at these guys. Father Time, well, he's already in the group chat. This is the primary driver of the flawed environment: the incentives. What if the Eagles had done what a lot of organizations do and fired Roseman when they fired their head coach? Would we still have seen the three Super Bowl appearances from the man who trades down perhaps more than anyone? The best organizations, they're the most stable organizations. And I'm not talking about Jerry Jones stability there either, as is the case in the corporate world. A good leader lets others do their job. A good leader recognizes that if they're evaluating short-term performance, there is a fair amount of randomness involved with both success and failure. This is where you can see a company like Sumer is attractive: outside consulting with a longer-term vision minus all of the normal incentives. They don't necessarily have the appetite for that. And then even when they do, there's a big capacity limitation. They'll go after a couple of projects here and there. "You know, the May-June time frame is kind of your exploration time as a data scientist within an NFL team. That's not a lot. You know, if you only have like two months to explore and maybe put something novel and new into production, you're kind of racing against the clock there." So we've done a significant amount of tracking data work.

So how does this tracking work for practice data for colleges? You do have some GPS use, but by and large it's computer vision based. So you know, folks processing the game footage and then trying to come up with the locations of all the players. There's a couple of things that make computer vision a little harder. You think about a deep pass, the broadcast angle will sweep from the quarterback to the safety. Well, in the early part of the play, you may be losing the safety, and so you don't have his exact location. And then at the end of the play, you'll have the coverage and the wide receiver, but you won't necessarily have the quarterback. So you have to, you know, maneuver your way around that. So it can be challenging to work with. In the NFL, there is just too much luck and chaos to act irrationally. But that is what makes the sport so fun and the draft so special. Somebody bring some ibuprofen to the commissioner's green room.

Take for comparison the NBA. How accurate is its draft? Well, far more than the NFL. Last year's All-NBA teams, they had one player selected in the second round. 14 of the 15 players were drafted in the first round. There are still surprises in hoops and busts, but largely we know every year who the stars are going to be. Football is a little different. The game is so dynamic that is far more challenging to isolate and predict talent, especially at the game's most popular position. Jaylen Herz is going to Philadelphia, and there are a million ways this can go. Oh, and there's a new archetype that has made it even more challenging to evaluate this position. These guys, they can run now. Lamar Jackson is gone. Daniels makes a man miss. Holmes going to walk in. Fer Murray clear on the afterburn. He's gone. Williams looking, dancing, hopping. Allen got some daylight, hurdles to the end zone. Herz on the run. He might go. Jaylen Herz, touchdown Philadelphia. In an age where everything has become optimized, football has remained stubborn: every move has its countermove, and every player is somewhat of a mystery. Yeah, it's frustrating when your team misses a pick or misses out in the playoffs, but they could be right back in it the next year. And that is the appeal: the uncertainty. It's like being a kid again, and you can see all the presents under the Christmas tree, but you don't know what they are. The beauty of the NFL and this draft lies in this uncertainty. Despite the data revolution, we still have humans making bad decisions, and we still have chaos. Hope you enjoyed this video. Make sure to subscribe if you haven't and take care.