Last offseason, I began to play with NFL data in many of the same ways I’ve been doing for years with college football. I was most curious, not about the differences in the two, but in the similarities. Turns out there were a lot of them.

For instance, the difference in success rate — an on-base percentage style of football efficiency measure and the primary building block of my S&P+ ratings — between NFL and college games is about one percentage point. Over the last decade, the college success rate has been about 42 percent, the NFL success rate 41 percent.

Admit it, you expected more of a difference between the pointsy college game and the most stodgy pro game. But efficiency levels are the same. There’s about half an extra big play per game in college, and turnovers are a hair different, but most of the scoring difference between the two levels comes from tempo differences and total plays. I found that surprising.

Knowing this, I was able to knowingly compare apples to apples with some primary advanced stats, and I was able to come to one primary conclusion when it came to the quarterback and skill positions:

College stats aren’t perfect predictors of pro performance, but they do tell us a lot about your ceiling. Why? Because you aren’t going to top your college stats in the pros.

This is most directly evident at the quarterback position.

Last year around this time, I wrote about both this ceiling assignment relationship and what it said about the 2018 draft prospects. The bullet-pointed version of the quarterbacks piece:

1. It’s really hard to figure out who’s going to be a good pro quarterback before they become good, even with assistance from stats. But we can indeed figure out what QBs are capable of on the high end. As I put it then, “Stats will never tell you everything about what a player can do. In this case, though, it tells you what certain players probably can’t.”
2. Using stats like success rate (or its down-, distance-, and field position-adjusted cousin, marginal efficiency), or yardage-based measures like AY/A (adjusted yards per attempt) or ANY/A (adjusted net yards per attempt), we find that no recent quarterback has exceeded his college stats over the first four years of his pro career. Most don’t come anywhere close.
3. We can therefore begin to at least establish tiers of potential for upcoming prospects. Though the guys in the top tier won’t automatically produce top-tier stats, the bottom-tier guys definitely won’t.

I loosely built three tiers of 2018 prospects based on each prospect’s career success rate. Explosiveness is random and often unsustainable, and it really doesn’t translate from college to pro. But efficiency does.

(Note: as I do not have play-by-play data for FCS or lower levels of the sport, I am only drawing references to quarterbacks who played at the FBS level below. Sorry, Easton Stick.)

There were four quarterback prospects in last year’s top tier.

Evan Habeeb-USA TODAY Sports

• Baker Mayfield, Oklahoma (54.8 percent career success rate in college)
• Sam Darnold, USC (52.0 percent)
• Mason Rudolph, Oklahoma State (50.0 percent)
• Logan Woodside, Toledo (49.5 percent)

Success rate is unadjusted for opponent, and playing in the MAC almost certainly boosted Woodside’s stats a bit. But what this said was, of all the 2018 draft prospects, these were the only four who cleared the statistical bar required to produce high-efficiency numbers in the pros during their rookie contract.

Of the 38 players in the NFL sample, only Jameis Winston (57.1 percent) and Sam Bradford (55.4) had higher career success rates in college than Mayfield. Winston’s first three seasons in the NFL have generated a 45.8 percent success rate; Bradford battled injury and a porous offensive line, generating a 37.2 percent success rate in his first two years before rising to 42.3 percent, near the league average (which is usually between 42.5 and 43 percent), in his next two.

Since both were drafted so high, there was no sitting — they threw a combined 1,181 passes in their respective rookie seasons. Mayfield and Darnold will potentially be thrust into action just as quickly. Rudolph and Woodside, perhaps less so.

Indeed, neither Rudolph nor Woodside threw a pass in a regular-season game last year, though Woodside did end up on the AAF’s San Antonio Commanders, producing average-at-best numbers (58 percent completion rate, 5.16 ANY/A) in the offensively challenged league. He recently signed with the Tennessee Titans.

Mayfield and Darnold, however, got thrown into the deep end as expected. And of the five quarterbacks drafted in last year’s first round, they produced the best success rates, just as they had in college: Mayfield at 43.7 percent, Darnold at 41.1.

The middle of the three tiers was what I called the inconclusive one.

Tommy Gilligan-USA TODAY Sports

There were successes and massive failures here, and the data was too messy to be definitive. Lamar Jackson, with his 47.4 percent career college success rate, landed here. (Reminder: I’m not looking at rushing figures here, only passing.) He also produced the third-best pro success rate as a rookie, at 40.5 percent. Convenient.

With an excellent defense at his disposal, plus a sturdy run game that he very much contributed to (749 rushing yards), he was the only rookie to quarterback a playoff team. The Ravens improved drastically down the stretch after he took over for Joe Flacco; at worst, he didn’t prevent the improvement from happening, and at best, he played a key role in it.

Then there was the bottom tier. A simple measure like success rate threw up major red flags for both Josh Allen and Josh Rosen.

Timothy T. Ludwig-USA TODAY Sports

Of the 38 QBs in our pro sample, only one (Brock Osweiler) managed a league-average passing success rate in the NFL over his first four years after producing a college success rate [below 47.1 percent]. The only two QBs in the lower-efficiency range who were drafted in the first round, as Rosen and Allen will be: Blaine Gabbert and Jake Locker. Not the greatest of role models.

If you’re likely to finish, at best, two to three percent below your college success rate, that means [Allen’s] ceiling is around 40.5 to 41 percent. That’s Ryan Mallett territory (40.8 percent). As a ceiling. Are we sure we’re willing to spend a top-five pick on a guy who might, with some good breaks, become Ryan Mallett?

Correction: Allen might become a more athletic Ryan Mallett.

Allen’s best pass play was to drop to pass for a second and then take off running. He was quite good at it — in 12 games, he rushed 89 times for 631 yards and eight touchdowns. Most of his yardage came on scrambles, as opposed to more designed runs for Jackson.

When Allen actually threw the ball, mostly bad things happened, just as the numbers predicted.

Each of the five rookies basically produced a success rate around 10 percentage points lower than their college numbers, and Allen was no exception. After producing just a 43.3 percent passing success rate in college (more than 11 percentage points behind Mayfield), Allen hit just 34.9 percent in 2018. He completed only 53 percent of his passes with 10 touchdowns and 12 INTs.

Granted, he had a sketchy line, no run game, and no receiving corps. Buffalo has worked to address these things in free agency and will likely continue to do so in the upcoming draft. As a top pick, Allen will get extended more opportunities to prove himself than a late-rounder or undrafted free agent would, and with a more worthy supporting cast perhaps his stats will improve. They started out, however, every bit as poor as his college stats suggested they would.

For all his flaws, Allen built far more of a case for optimism than Rosen.

Joe Nicholson-USA TODAY Sports

I wasn’t sure what to make of Rosen’s college stats since they were dragged down by the fact that he was a starter as a true freshman. Most college QBs aren’t. He improved his rate stats each year he was in college, so I thought that maybe put him in the clear. And hey, maybe it will when all is said and done.

His rookie year couldn’t have gone worse, though. His passing success rate was a horrifying 33.7 percent, nearly 13 percentage below his college success rate (46.6 percent). On a listless roster with a lost coaching staff, he completed 55 percent of his passes, but at only 10.5 yards per completion (Allen was at 12.3) and with no rushing output (under two carries per game). And while Allen’s organization has tried to figure out how to get him help this offseason, Rosen’s has been looking to potentially trade him and draft another QB in the first round.

It’s obviously been just one season. We’ll find out all we need to know about the five first-rounders, plus potentially others like Rudolph, in the coming years. Still, these guys started off in the order that their college stats predicted. I’ll take it.

So now let’s flip to 2019. What can college stats tell us about the new crop of quarterback prospects?

Photo by Jamie Sabau/Getty Images

As tends to happen, a consensus has formed regarding this year’s draft class. Oklahoma’s Kyler Murray will likely be the top QB (and overall player) taken, followed by Ohio State’s Dwayne Haskins, Missouri’s Drew Lock, and Duke’s Daniel Jones. After that, things get blurry. We don’t know the teams associated with each player, but it does appear that’s the order most expect.

Here’s what the stats have to say about everybody.

There obviously won’t be 22 QBs drafted, but those are the (FBS-only) players I’ve most frequently seen on prospect lists. Based on the ones I think are the most realistic potential draftees, I see basically four tiers forming this year.

Tier I: The highest of ceilings

Photo by Brett Deering/Getty Images

• Kyler Murray
• Dwayne Haskins
• Will Grier

There is no doubting who the most statistically impressive quarterback in college football was last season. Murray somehow managed to trump Mayfield’s un-trumpable stats at OU, joining 2015 Deshaun Watson as the only QBs to throw for 4,000 yards and rush for 1,000 in the same season. He did it in one fewer game than Watson, too.

(Less than two decades ago, it was mind-blowing when Clemson’s Woody Dantzler hit 2,000/1,000. Now that’s passé.)

Better yet, Murray pulled this off with minimal draft talk. He had already been selected by baseball’s Oakland A’s, and playing pro football appeared to be off the table.

We didn’t have to spend the entire fall debating whether Murray was tall enough, whether his hands were big enough, or whether his off-the-script speed bursts might translate at a higher level. We could just relax and watch as he dragged Oklahoma to another College Football Playoff appearance despite a defense far worse than the one that dragged Mayfield down in 2017.

Of course, we’ve made up for that lost debate time this offseason. Murray decided he enjoyed football too much to give it up and elected to explore his NFL Draft options. And Arizona, the team with the No. 1 overall pick, hired a head coach who had just months earlier said that he would take Murray with the No. 1 pick if he had the chance. So now it looks like he’ll be the No. 1 pick.

The drawbacks are obvious here. He is tiny by pro quarterback standards, and he does rely on his running speed to bail him out sometimes. Any player is a risk, really, and Murray’s potentially damning traits are more out in the open than most.

He’s also got the highest ceiling of anyone in the damn draft, and it’s not particularly close.

Mayfield’s initial success — not to mention the fact that 6-foot-nothing Drew Brees continues to play at an otherworldly level — is helping to shoot down all of the old assumptions about needing a 6’5 statue to win in the pros. And while Murray does appear to stray from the script a bit more than Mayfield ... I mean ... have you seen how fast he is?

There are two other QBs in this tier, as well.

Photo by Tim Warner/Getty Images

Haskins’ presence is like Darnold’s last year, well below the OU QB of choice but ahead of most others. He has his own set of questions to answer over time. He really didn’t face much pressure in 2018 and didn’t always respond well when he did, and a lot of his yardage and efficiency came from having maybe the best WR screen game in the country in his back pocket.

Still, this exercise is simply about clearing a bar, and Haskins does so. So does Grier.

The whole “system QB” debate has become less useful with former “system QBs” like Patrick Mahomes II and Jared Goff thriving in the NFL, but Grier has seen that label quite a bit. He spent his last two seasons in Dana Holgorsen’s QB-friendly system, with two QB-friendly wide receivers (David Sills V and Gary Jennings Jr.) lined up wide.

When Grier sat out, though, WVU’s production plummeted, suggesting this production was not merely about the system at hand. And he’s got experience and proven pocket presence that neither Haskins nor Murray can match.

He’s proven far more than Daniel Jones, too. Just saying. But again, he clears the statistical bar. The rest of the scouting report is up to you.

Tier II: You should know, there are red flags

Photo by Jamie Squire/Getty Images

• Drew Lock
• Ryan Finley
• Jarrett Stidham
• Gardner Minshew
• Kyle Shurmur

If the NFL Draft were purely about picking the most fun players, Lock would be the No. 2 overall pick behind Murray, maybe No. 3 behind Ed Oliver, too. The dude throws a gorgeous deep ball, invents new arm angles from which to throw, and never fails to remind you that he’s enjoying himself.

For what it’s worth, he also improved over each of his four years at Missouri, and like Rosen, his career stats were dragged down by the simple fact that he was asked to play for a miserable 2015 Tiger offense. Take out his 2015 stats, pretend that was a redshirt year, and his career success rate rises to 44.5 percent, his ANY/A average to a nearly Grier-like 8.37.

A 44.5 percent success rate still isn’t all that high, however. It’s only about one percentage point higher than Allen’s, and even if you adjust in your head for the fact that he was facing SEC defenses — or the fact that his No. 1 receiver, Emanuel Hall, was hurt during a three-game senior slump against Georgia, South Carolina, and Alabama — that’s still going to be a sticking point. He misses on some easier passes from time to time, and that negates some of that deep-ball buzz. Or maybe a lot of it.

Tier III: Maybe a later-round steal?

Photo by Loren Orr/Getty Images

• Jordan Ta’amu
• Brett Rypien

Murray, Haskins, and Grier were the only three who clear the top-level bar, but a couple of other potential draftees came close. Rypien started for four seasons at Boise State and ended up throwing for 13,578 yards and 90 touchdowns as BSU averaged 10 wins per year. He gets high marks for his ability to process the game and work within the pocket, but that compliment, plus his mid-level draft status, tells you a lot about his perceived mobility and other traits.

Still, his numbers were strong. So were Ta’amu’s, though those stats are both supplemented by pretty strong “yeah, but ...” qualifiers:

• Yeah, but ... Rypien faced a lot of bad Mountain West Conference defenses. Like Woodside, you probably have to adjust for that a bit. Plus, while he started as a freshman like Lock or Rosen, his freshman numbers were basically the same as that of the rest of his career. Removing his 2015 stats doesn’t really change anything.
• Yeah, but ... look at Ta’amu’s damn receiving corps. In his season and a half as Ole Miss’ starting QB, he had at his disposal both all-time freak athlete D.K. Metcalf and A.J. Brown, who showed quite a few flashes of being the best receiver in college football.

Still, Ta’amu produced. And it is somewhat intriguing that in 2017 he produced at a higher level than Shea Patterson, who got hurt midway through the season. Patterson was an all-world recruit, and Ta’amu was a no-name JUCO transfer, and Ole Miss’ offense didn’t miss a beat when Ta’amu took over.

As for Rypien ... well ... few penalized Allen for facing those defenses. In fact, they turned it around as a compliment — he was working with MWC receivers!

If nothing else, they are worth considering as options later in the draft, especially as an alternative to picking someone with much higher risk in the first round.

Tier IV: Yeah, let someone else take the risk

Photo by Mark Brown/Getty Images

• Tyree Jackson
• Daniel Jones

I get the draw of Daniel Jones. I really do. He’s 6’5, he’s mobile, and he looks like a Manning when he steps away from center. Since he was coached by the guy who coached the Mannings in college (David Cutcliffe), that makes sense.

Plus, Jones’ high-level moments were extraordinarily high-level. He torched a solid Temple pass defense for 30-for-41 passing and five touchdowns in the Independence Bowl in December, and the month before that he did a really impressive Lamar Jackson impression against North Carolina, throwing for 361 yards and rushing for 186 in the same damn game. He also torched Northwestern (impressive) and Pitt (less so) earlier in his junior season.

• In these four games, he completed 65 percent of his passes at 8.6 yards per pass, with a college passer rating of 165.3.
• In his seven other games last year, he completed 57 percent of his passes at 5.6 yards per pass. Passer rating: a ghastly 108.8.

Jones’ best moments were quite a bit better than Josh Allen’s in college, so again, I get the draw. But his arm isn’t as strong as Allen’s, and ... my goodness, the inconsistency.

It wasn’t a one-year thing, either. In 2017, he produced a passer rating over 150 twice and under 100 four times. His passer rating never topped 131.7 over the course of an entire season.

Jones will likely be a first-round pick, and he will thus be given extra opportunities to figure things out. Maybe eventually things will click, and the high points will outweigh the lows. But if I were an NFL general manager, I’d be letting someone else take that risk. Trade down a few spots and try to get Grier. Maybe aim for quantity and try to get both Rypien and Ta’amu later on. (Just don’t get Clayton Thorson. His stats are like Jones’ without the good games.)

Or, hell, if you’re a risk taker by nature, just nab Tyree Jackson further down on the draft board. He’s taller, his highlights are even more fun, his upside is just as high, and his downside is even lower. Ride that roller coaster, baby.