977: Finding a Fantasy Football Edge With Mock Drafts

JJ Zachariason (0:02)

This is The Late-Round Podcast with your host, JJ Zachariason.

What's up, everyone? It's JJ Zachariason. In this episode, 977 of The Late-Round Fantasy Football Podcast. Thanks for tuning in. The Prospect Guide is out the door. At least the pre-draft version is. If you pre-ordered The Late-Round Prospect Guide, it's in your email inbox. If you didn't, that's no problem. You can still order it on lateraound.com. The Guide is 160 pages long. It gives you profile write-ups for every running back and wide receiver who are at the combine. You get year two profile breakdowns. I talk about what goes into the Prospect and year two models and so much more. In the post-draft version, you get tight end scores too. Check it all out over on lateraound.com. And while you're on the site, make sure you subscribe to the free newsletter. Why not? It's totally free, and it helps me out. Now, today's episode is kind of related to that Prospect Guide. As many of you know, in the various ZAP models, my Prospect models, I use draft capital as an input. Where a player gets drafted matters to his overall score. Pre-draft, so right now, this period, we obviously don't know exactly where a player is going to get drafted. To help with that, I use nflmockdraftdatabase.com to get projected draft capital. That's based on a bunch of different mock drafts. One of the things I've questioned on this show in the past is, what if that projected draft capital is actually better and more predictive than actual draft capital? What if the wisdom of the crowd is actually a better input into these models than the draft capital that we're actually seeing via the NFL draft? NFL teams are just one mind, right? I think we all remember just a handful of years ago, the surprise of 2-2 Atwell getting drafted in the second round by the Rams. That one input is telling the ZAP model and telling everyone else that 2-2 Atwell is a second round quality player. But what if the Rams were the only organization on planet Earth who thought that he was better than a fourth rounder? As a result, that draft capital input might not be as trustworthy as we think, and maybe an aggregate average would have been better. But then even beyond just the predictiveness of draft capital, I've always wondered if we could learn more with aggregate mock draft data. If there's a player who's beloved by analysts and mock drafters, but he slips in the NFL draft, should we consider that player a value in our rookie drafts? Or what if there's a prospect that no one liked heading into the draft, or didn't love, like 2-2 Atwell, but then that guy gets drafted in round one? Is he someone that we should avoid? I finally tested all of this, and I want to share some of those results.

Now, I do want to preface that some of what I'll be explaining today is super nerdy. I'm going to do my best to be as clear as possible, but just be prepared to maybe focus a little harder than usual, like you're watching an episode of Severance. If only I was as brilliant as those writers. Now first off, the ZAP model databases at Running Back and Wide Receiver, they date back to 2011 Data from NFL Mock Draft database goes back to 2016 So for everything that I'm going to be talking about today, I'm focusing on players who were drafted from 2016 onward. I'm also going to focus on players who have had the ability to play three seasons in the NFL. To test all this mock draft data and see if it's meaningful, we can't really test against like the 2024 rookie class. They've only played one year. We don't know who's good and who's not good yet. Instead, I'll go through the 2022 class, because the 2022 class has now played three seasons.

So, I've got two samples, one of running max, one of wide receivers, and they're all players in the ZAP model database from 2016 through 2022 So, these are guys who were either drafted or at the NFL combine during that time frame. Now, in the ZAP model, I measure against what I call B2S. It stands for Best Two Seasons. I look at the first three seasons and points per game in a player's career. I take his top two seasons and points per game, and I average those top two seasons together. That's a player's B2S. And for a season to qualify, the player had to play at least eight games. Now, I was curious to see if projected draft capital or big board data, this is all from NFL mock draft database, I was curious to see if that was any more or less predictive than actual draft capital. There are a lot of ways to test this, but I just did it via B2S. I compared where a player was drafted and where a player was projected to get drafted versus his top two seasons and points per game across his first three years in his career. I not only did this with raw draft capital numbers and raw projected draft capital numbers, but I used Chase Stewart's draft value chart as well. His draft value chart essentially takes an overall draft selection and assigns a value to it. As we know, the NFL draft is not linear. The second overall pick to the sixth overall pick shows a four pick difference, but so does pick 100 to pick 104 But we all know that there's more of a difference in value from pick 2 to pick 6 Stewart's value chart helps account for this. Now, I'm going to be referencing that chart a lot on today's show. You can simply Google Chase Stewart draft value chart if you want to learn more. But it's basically just assigning different weights and different values to each slot in the NFL draft. Now, when looking at the correlation between draft capital and projected draft capital, versus B2S, no matter how you analyze it, whether just the raw numbers or through the lens of Stewart's value chart, actual draft capital seems more helpful than projected draft capital. Meaning actual draft capital helps explain a player's best two seasons during his first three years of his career better than projected draft capital does. We see a little less of a gap when looking at running back than we do wide receiver, and honestly, the gap isn't that dramatic at either position. We're talking like a 50% R-squared for actual draft capital at wide receiver within this sample versus it being 44% with projected draft capital, and at running back, the difference in R-squared is even smaller than that. Now, I don't think the fact that actual draft capital is better is that surprising. Teams are putting a lot of effort into evaluating players, but they're also the ones who are specifically drafting players based on their individual needs. So if the masses view a player one way, and a team views them another way, it could be that the team has a specific way that they want to use that player. And if that's the case, then that player might have a better chance to thrive in the league and score fantasy points. So that's part one. And yeah, kind of a boring result, which happens sometimes. In the end, actual draft capital does seem to be more helpful when looking at fantasy success than projected draft capital, even if it's not by much. At least we know that projected draft capital isn't that bad of a thing to use before the draft actually happens. But what I wanted to do next is see if there were any instances where we could use projected draft capital to our advantage. Like the 2-2 Atwell example earlier, we know that Atwell has not been that great of a fantasy asset for a second round pick. Could we have projected that simply based on where aggregate mock drafts had him ranked? Because he was ranked well below his actual draft capital. Could that have told us something? In other words, if a player is perceived to be a reach by a team, is that a bad thing for fantasy football? And then on the flip side, what if that player is a huge value? Let's start this discussion by looking at running back. I used the following process to see if an NFL team reaching for a draft selection mattered or not in fantasy football. First, I converted every player's projected and actual draft capital into a value via Chase Stewart's value chart. So instead of some raw draft pick, I now have a value to work off of. Next, I found the difference between actual draft capital value and projected draft capital value for every player in the sample. If a player had really low projected draft capital value, but then a much higher actual draft capital value, then he'd be considered more of a reach. If a player had good draft capital value, and then he slipped in the draft, then he was considered a value. Just by doing that at running back, I found that the biggest reach at running back in any draft from 2016 to 2022 was when the 49ers took Ty Davis Price back in 2022 The second biggest reach was AJ Dillon. The third biggest reach was actually Ezekiel Elliott. Now, on the flip side, the best value picks were Kenneth Dixon, Paul Perkins and Dalvin Cook. Now, my goal is to analyze and see if there's anything to this idea that a reach is bad and a value is good. I can't use raw fantasy output to do this, though, because players are going to score a wide variety of points based on draft capital alone. Here's what I mean by that. If I'm trying to see if reaches are bad, and I group all of the biggest reaches up and see how they performed across their first three seasons, what if all of those reaches were first round picks? They're all going to perform well, even if they were reaches during the NFL draft. So to help with that, I calculated expected max season for every running back in this data set based on his draft capital. I told you this was going to get really nerdy. Every running back, every wide receiver, they have some expectation at their draft capital. And I'm looking at an expectation for best season, best singular season in PPR points per game across the players first three years in the league. So let's go back to Dalvin Cook. He actually got drafted 41st overall. At that spot, we expected him based on what's happened over the last 15 or so years to give us a season of 14.1 PPR points per game. Dalvin Cook, though, ended up scoring 20.9 PPR points per game in one of his first three seasons. That was his best season. So we outperformed that max, that expectation by about 6.8. And remember, that was after being a value pick according to where he was drafted versus where he was projected to get drafted. So what I did for both Running Mac and Wide Receiver, we're just focused on Running Macs here first, I grouped all of these players up into different draft capital value buckets. Were they a reach or were they a value and by how much? That's all based on the Chase Stewart chart. Then I looked to see how well they performed against expectation during their first three years in the league. And I did that by simply looking at their best season in one of those first three years. And what I found was kind of interesting. There were a lot of players in the data set who saw a value of zero, meaning their projected draft capital value versus their actual draft capital value, they were the same. That's because they were projected to go undrafted, and then they went undrafted. And then every once in a while, you get players who were drafted exactly where they were projected to get drafted. So they had a draft value of zero as well since the NFL team didn't draft them above or below where the aggregate mock drafts had them. But we still had 143 running backs, not in that situation. We had 143 running backs that we can analyze. And what I found was, generally speaking, as the value went up, as NFL teams got perceived better value, those players outperformed expectation at a slightly higher rate. And generally speaking, as the value became lower, as teams had a perceived reach for players, fantasy output got worse versus expectation. I will say though, the correlation is not as strong here as it might sound. I wish the data was more convincing, but it's only a tiny bit convincing. Like, there were 15 running backs who ended up being a value by 2 or more points based on Chase Stewart's value chart. So these are the 15 running backs who the masses would consider big values by the NFL team that drafted them. Of those 15, they outpaced their max PPR point per game output by almost 1 point per game on average. When you look at the 35 running backs who were drafted with their NFL team losing 2 or more points of value, they ended up playing below expectation by almost 2 full PPR points per game. At each extreme end, you do see that running max will outperform expectation if they were valuable, and they'll underperform expectation if they were reaches. I wouldn't say it's super clear cut, but there's at least some signal there. I think it's more when players become big reaches. But that's even more evident at wide receiver than it is at running back. Overall, at wide receiver, you're not going to find a ton of signal when looking at players who are big values, or moderate values, or even moderate reaches. It's all about big reaches. There were 14 wide receivers in this data set whose NFL team lost four points of value when looking at pre-draft draft capital and actual draft capital. Since there's only 14 of them, I'll just read you the names. Again, these are the biggest reaches by NFL teams from 2016 to 2022 Ricardo Lewis, Chad Williams, Corey Davis, John Ross, Kenny Golladay, Deontay Johnson, Jalen Hurd, Jalen Reger, Des Fitzpatrick, Tutu Atwell, Jahan Dotson, Tyquan Thornton, Valis Jones, and Wondell Robinson. If you want to add 2023 and 2024 to the mix, we'd add one more player to the list, Trey Tucker. And the two players who just missed the cut were Darius Davis and Jonathan Mingo. Now, if you want to expand that to teams who lost 3.5 or more points in value, you'll add Curtis Samuel, Mike Williams, Dante Pettis, Michael Pittman, and Van Jefferson.