Projecting the most likely Pro Bowl players in the 2019 NFL draft class

There's an almost one-in-four chance that Kentucky pass-rusher Josh Allen, who's likely to be picked in the top five of the 2019 NFL draft, doesn't develop into a starting-caliber player.

When teams select a player in the draft, they aren't getting what that player will become. They're getting a range of outcomes. That's where projections come in.

ESPN's NFL draft projections take Scouts Inc. grades, incorporate testing from the scouting combine, then translate the data into probabilities of career-path buckets -- Pro Bowl caliber, starting caliber, bench/special-teams player, replacement level or nonfactor -- for the player's first three seasons in the league, all relative to the player's position. Positional value is not factored in: a tight end might have a higher chance to reach a Pro Bowl level of play, but that doesn't necessarily mean his projection is more valuable than, for instance, a quarterback with a lower chance to reach the same level. Full methodology can be found here.

These projections can help decision-makers in two ways. First, providing these range of outcomes can aid in choices surrounding trades and positional value (e.g., if the Jets trade down from No. 3 to No. 10, how much less likely are they to land a Pro Bowl-caliber player than if they stood pat and took Quinnen Williams?). In addition, a model like this can reveal players who might be overlooked or overrated and could be worth a second look.

Here are the six prospects who our model thinks are most likely to reach a Pro Bowl-caliber level of play in their first three seasons. The comps to past NFL draft picks are based only on combine measurements for similarly graded players.

Jump to a section:
Underrated prospects | Overrated prospects
Ranking Pro Bowl chances | What about the QBs?

1. Devin White, ILB, LSU

Pro Bowl-caliber chance: 70 percent
Athleticism comp: Mychal Kendricks

White ranks off the charts here when compared to every other prospect in this class. Luke Kuechly is the only other inside linebacker in our data set -- which goes back to 2006 -- whose projection would have compared to White's.

White's Scouts Inc. grade of 94 is one of the best in the class, but that isn't the only reason his projection is so strong. It's also because of his absurd athleticism. His 4.42 40-yard dash at the combine is the fastest time of any inside linebacker since at least 2006, and speed is an incredible asset in today's NFL.

Statistically, that speed also showed up in college as a blitzer: White recorded pressure on 24.5 percent of his pass rushes, which ranked second only to Allen among prospects with at least 100 pass rushes.

2. Devin Bush, ILB, Michigan

Pro Bowl-caliber chance: 57 percent
Athleticism comp: Haason Reddick

Bush is just behind White, according to our model. Both are well-regarded off-ball linebackers whose upside is more realistic given their speed, quickness and athleticism. Bush ran a 4.43 40-yard dash but posted the best broad jump among inside linebackers in our data set. The difference between them really only stems from a lower Scouts Inc. grade (Bush is 91). Bush is nonetheless one of the most likely players in this draft to excel in the league.

Like White, Bush also had success when rushing the passer, with a 21.9 percent pressure rate.

3. Dexter Lawrence, DT, Clemson

Pro Bowl-caliber chance: 53 percent
Athleticism comp: Not enough data

What happens in our model when a prospect barely participates in drills at the combine? The model works with what is available. Whatever information it has -- in this case, height, weight, wingspan, 40-yard dash and bench press -- to project how the player would have performed in the other events. While Lawrence's speed didn't stand out, his overall size does. And so when combined with the size, his speed actually is quite impressive. That combo is the real driver here.

Add that into a solid Scouts Inc. grade (88) -- which is the largest factor in the model -- and it thinks Lawrence has a high chance of becoming one of the league's top defensive tackles.

4. Andre Dillard, OT, Washington State

Pro Bowl-caliber chance: 47 percent
Athleticism comp: Joel Bitonio

Dillard showed incredible speed (4.96 40) along with solid but unspectacular strength (24 reps of 225 pounds on the bench press) for his position. So why does he get a bump from the combine?

play

NFL draft profile: Andre Dillard

Andre Dillard is an offensive lineman who earned first-team All-Pac-12 accolades at Washington State.

Counterintuitively, bench press does not correlate well with offensive tackle performance. But when combined with Scouts Inc. evaluations, speed reveals an aspect of athleticism for tackles that can be underrated by qualitative scouting methods. So when a player like Dillard -- who already has a strong grade -- drops an impressive 40 time, it indicates he could be even a better prospect than previously thought.

Our model considers Dillard the safest player in the draft. It gives him a 92 percent chance to become at least a starting-caliber player within his first three seasons, best of the entire class.

5. Noah Fant, TE, Iowa

Pro Bowl-caliber chance: 47 percent
Athleticism comp: Mike Gesicki

Our model is high on both Iowa tight ends: Fant and T.J. Hockenson. While the latter seems to be receiving more love in the pre-draft process -- and Scouts Inc. gives him a 92 grade vs. Fant's 89 -- the qualitative measurements are close enough that Fant was able to grab the top tight end spot in our model thanks to his athleticism.

Fant led all tight ends at the combine in the 40 (4.50), vertical jump (39 ½ inches) and broad jump (127 inches). He received the lowest blocking grade from Scouts Inc. that a tight end can get, so a team selecting him is really drafting a big receiver.

The Iowa tight end ran go routes 52 times last season -- more than any other route type -- and recorded the highest yards per route run on those go plays. He had 320 receiving yards when lined up in the slot, compared to just 183 starting from an inline position.

6. Nick Bosa, DE, Ohio State

Pro Bowl-caliber chance: 45 percent
Athleticism comp: Shaq Lawson

play

McShay: Bosa can be elite if he stays healthy

Todd McShay compares Nick Bosa and his brother Joey's game, and McShay sees plenty of production despite Nick playing only three games.

Unlike most of the other players on this list, Bosa's projection is driven by Scouts Inc. grade of 95, which is tied with Quinnen Williams atop the 2019 class. In Bosa, a team would get an edge defender who has a very good chance to become a high-end pass rusher -- he has a 77 percent chance to become at least a starting-caliber end.

Though we have only a limited sample of 2018 to look at for Bosa, his 21.2 percent pressure rate was among the leaders in college football.

Two underrated prospects to watch

The model appears more bullish than the general consensus on the two prospects below. That doesn't mean the consensus is wrong, but if we ran an NFL team, we'd take a close look at drafting them:

Marvell Tell, S, USC

Pro Bowl-caliber chance: 38 percent
Athleticism comp: Justin Simmons

Tell didn't run the 40 or do the bench press at the combine, but his other testing numbers -- such as his 42-inch vertical and 136-inch broad jump -- indicate high-end athleticism for his position. The only knock on him, athletically, is his weight -- he's only 198 pounds -- but the model still gives Tell a high chance of reaching his upside.

Connor McGovern, OL, Penn State

Pro Bowl-caliber chance: 32 percent
Athleticism comp: Ali Marpet

McGovern didn't participate in every combine event, but the former Penn State guard showed good athleticism in the broad jump (112 inches) and short shuttle (4.57 seconds). He trails Boston College's Chris Lindstrom there, but McGovern can be had later in the draft, as Lindstrom is expected to be picked in the first round.

Two prospects who might be overrated

These prospects have a worse projection than we would presume given that they are both expected to be selected among the top 40 picks.

Jonah Williams, OT/G, Alabama

Pro Bowl-caliber chance: 15 percent
Athleticism comp: D.J. Humphries

Williams' short arms, which made their way into the spotlight around the combine, are part of the reason our model is down on him, along with his size (6-foot-4, 302 pounds). He excelled with the Tide despite those shortcomings, and he still can succeed in the NFL too -- it's just less likely than his projected draft position might suggest.

play

Josh Jacobs can simply run people over

Former Alabama running back Josh Jacobs runs with a full head of steam, can catch the ball and isn't afraid to get out there and block.

Josh Jacobs, RB, Alabama

Pro Bowl-caliber chance: 24 percent
Athleticism comp: Mark Ingram

Jacobs is the running back in this class with the best chance to reach the top tier, but his 4.60 second 40 wouldn't have been close to the top of any of the other recent draft classes. For comparison, Saquon Barkley was at 49 percent to reach Pro Bowl-caliber by our model. Now, Jacobs will clearly go later in the draft than Barkley did, but he's only the 12th-best running back by chance to reach that top level in the past five seasons.

What about the quarterbacks?

Projections for the top quarterbacks are lower than many of the other top prospects at other positions, but that doesn't mean they don't justify an early first-round pick.

Kyler Murray -- 16 percent to be a Pro Bowl-caliber player -- has the best chance by our model. Murray has a 58 percent chance to become at least a starting-caliber QB in his first three seasons. For context, that is higher than the same model would have had Baker Mayfield last year (45 percent). The highest starting projection the model would have had on any quarterback since 2006 was Andrew Luck at 73 percent.

In terms of chance to be a starter in this class, Murray is followed by Drew Lock (55 percent) and Dwayne Haskins (53 percent) before a big drop-off to Tyree Jackson (23 percent), Jarrett Stidham (22 percent), Ryan Finley (21 percent), Daniel Jones (21 percent) and Will Grier (20 percent).

Methodology

ESPN Sports Analytics' new projection model for the NFL draft makes it easy for fans and analysts to assess the potential of prospective players. First, using grades from our partners at Scouts Inc., and testing numbers and metrics from the combine, it projects each prospect's chances at early-career success. Second, it matches each prospect with his five most athletically similar players from previous drafts. Ultimately, it provides two ways to understand and compare each prospect's potential.

The career projections estimate the chances each prospect's early career would achieve one of five levels of success:

1. Pro Bowl-caliber

2. Consistent starter

3. Reserve/special-teamer

4. Replacement level

5. Nonfactor

By describing the chances of achieving each of those five levels, the model is able to accurately express the uncertainty inherent in projecting the fortunes of draft prospects, something often overlooked. Career success is measured by Pro-Football-Reference.com's approximate value (AV) metric, which is based on position-specific statistics, games played, games started and awards, such as being selected to the Pro Bowl. Percentile buckets of AV (relative to position) were used to estimate the five levels above.

The projections are powered by something called a ridge regression. It turns out that the Scouts Inc. grades are dominantly more predictive compared to the combine metrics, because accurate scouting grades consider far more than mere athleticism and can include athleticism itself. But some combine events are sometimes over- or undervalued in scouting assessments, and prove to be helpfully predictive at the margins.

The player-similarity part of the model allows fans and analysts to assess prospects by comparing them to previous prospects whose pro fortunes are known. For example, a current prospect who is most athletically similar to several other recently successful players at his position should be someone to watch closely. If he is not rated highly by scouting reports, perhaps there is an "upside" to the player that might have gone unnoticed. This approach also allows fans to get a sense of a player's style; is he a bruising power running back, or is he a speedy and elusive runner?

"Similarity" is determined using a method called "t-SNE," and is based primarily on combine metrics, including height, weight, and performance on speed, agility and strength drills. For prospects who might not have completed every combine drill, performance on those metrics are estimated based on all of the other known metrics of that player. Scouting grades are factored in, but only in a way that helps the model compare top-ranked prospects to similarly ranked prospects within each position.