Since 2008, former NFL scout Dan Shonka of ourlads.com has been capturing “ball velocity” of quarterbacks throwing at the NFL combine, and posting it online. Some in the media have taken notice of the results and drawn conclusions from the historical data. “There has never been a QB who threw less than 54 mph with any long term success in NFL,” draft guru Benjamin Allbright noted in 2014:
There has never been a QB who threw less than 54 mph w/ any long term success in NFL. http://t.co/v0iw6Q9XoK
— Benjamin Allbright (@AllbrightNFL) March 2, 2014
I have several issues with this analysis. One, there are many reasons to think the emphasis on a quarterback’s physical tools is misguided. That especially applies to arm strength, which comes into play less often than commonly assumed. My Inside the Pylon colleague Mark Schofield charted every NFL starter for the 2017 seasons and found that around 86% of passes were within 20 yards. And while we think of arm strength as an innate characteristic, former Bears Director of College Scouting Greg Gabriel notes that such luminaries as Peyton Manning, Tom Brady, and Drew Brees improved their arm strength with mechanics work and weight training.
But even granting that arm strength does matter in some circumstances, it is not clear that the radar gun measurements are an accurate indicator of arm strength. The quarterbacks are not competing in a “velocity drill” where they are trying to throw the ball as hard as possible. The measurements are taken incidentally during general throwing drills. While Shonka attempts to capture readings during sideline throws that demand the greatest velocity, speed is only part of the picture. Quarterbacks are also being judged on their accuracy, footwork, spiral, and timing. A drill that focused purely on velocity might produce different results. Ourlads timed rocket-armed Patrick Mahomes, for example, at 55, but he later threw 62 in a TV spot where he was trying to maximize his radar gun reading. Accuracy and consistency is also a question; the Athletic’s Dane Brugler got different readings than Ourlads for the 2017 class, with discrepancies of up to six miles-per-hour:
Ball velocity numbers (MPH) from the Combine:
60 – Mahomes
59 – Webb
56 – Kizer
55 – Trubisky, Evans
53 – Peterman, Kaaya
49 – Watson
— Dane Brugler (@dpbrugler) March 10, 2017
The idea of a minimum amount of necessary arm strength does make some sense, but the idea that there’s a hard threshold at exactly 55 doesn’t pass muster. Many of the most successful quarterbacks who actually threw at the Combine—more on that in a minute—barely cleared the 55 threshold. Joe Flacco, Russell Wilson, and Jameis Winston measured at exactly 55, while Cam Newton, Andy Dalton, Jimmy Garoppolo, and Marcus Mariota were at 56. Prior to 2016, it was arguably as difficult to find a successful passer with a 57+ as it was to find one who threw under 55. If so many of the quarterbacks who went on to have productive careers barely cleared the 55 mark, it’s absurd to think a passer who registers a 53 or 54 can’t succeed.
Then there’s the problem that many of the top passers don’t throw at the Combine, or at least didn’t used to. Of the 19 first-round quarterbacks from 2008-2014, only nine threw at the Combine. Worse yet, we only have radar gun readings for one of the six Pro Bowl quarterbacks in this set, Carolina Panthers star Newton. That means the sample of quarterbacks for whom we have radar gun readings at all, good or bad, skews heavily towards failures.
Looking at any individual year makes this plain. In 2008, third-overall pick (and future MVP) Matt Ryan didn’t throw. Flacco did, hitting exactly the 55 threshold. The other quarterbacks contributed little, with the biggest names long-term backups Chad Henne (53), Matt Flynn (50), and Josh Johnson (49). In that context, with one of the two good quarterbacks in the class not throwing and the other hitting exactly 55, what can this metric tell us? The 2009 class had one good QB, top pick Matthew Stafford, who didn’t throw (fellow first-rounders Mark Sanchez and Josh Freeman did, both registering 57s; both flopped after a little early success). In 2010 none of the top three quarterbacks chosen (Sam Bradford, Tim Tebow, and Jimmy Clausen) threw, so the radar gun readings are all for guys like Levi Brown, Max Hall, Jevon Snead, and Dan LeFevour. In 2011, five of the top six picks threw, but in 2012 it was back to low compliance, with Andrew Luck, Robert Griffin III, Ryan Tannehill, and Brock Osweiler all sitting it out. The universe of players for whom we have radar gun readings is heavily skewed towards players with little NFL success.
In recent years, this has changed somewhat. Most quarterbacks of significance in the past five drafts have elected to throw, with notable exceptions in 2018 third-overall pick Sam Darnold and presumed first-rounder Kyler Murray this year. With a few more years of data, we might be in a better position to draw inferences based on these radar gun readings, though the methodology flaws will still prevent any firm conclusions.
More quarterbacks throwing means we are starting to see some exceptions to the “rule,” however, though as with any young players, the fate of these passers is not set in stone. The 2015 season saw the surprise emergence of fourth-round pick Dak Prescott, who earned Pro Bowl honors while helping the Dallas Cowboys to a 13-3 record. The gun measured Prescott at 54, just under the threshold. In 2017, second overall pick Mitchell Trubisky registered a 50 and 51 on two throws at the Combine; after an uneven rookie season, he showed excellent promise in his first year under offensive guru Matt Nagy, completing two-thirds of his passes and making the Pro Bowl.
It is fellow 2017 draftee DeShaun Watson, however, that threatens to blow the “ball velocity” framework to smithereens. Watson registered a 45, one of the lowest scores recorded in the decade of measuring ball velocity. That hasn’t stopped him from establishing himself as one of the most effective and exciting young passers in the league. Watson’s 8.4 yards per pass attempt as a rookie would have led the league were his season not cut short by a knee injury, and his 8.2 figure in 2018 ranked sixth. He’s far from perfect—his 10.9% sack rate was far too high—but those struggles have little to do with arm strength. It does not appear his low radar gun reading is limiting him in any way. Combined with the methodology and statistical concerns noted above, Watson’s success may prove to be the nail in the coffin for this measurement.
The crazy thing about Watson is, before the ball velocity measurements came out, he was widely reported to have helped himself during the Combine. USA Today rated him the #3 player (and #1 QB) who helped himself most in Indianapolis. Kansas City Chiefs head coach Andy Reid described Watson’s performance as “Unbelievable. He had a great day. Every throw was on the money.” Whether insiders like Albert Breer, outsiders like Doug Farrar, or former scouts like Daniel Jeremiah, everyone had praise for the Clemson quarterback, including the spin, accuracy, and zip on his passes. Then the numbers came out, and the naysaying started.
I consider myself a statistics and analytics guy. I have no problem with numbers. But part of being a statistics and analytics guy is understanding how numbers are measured, how valid the methodology is, what flaws and biases exist in the methodology and data, and how strong are the inferences we can draw from the numbers. For the reasons I’ve described above, any person who is serious about statistics and analytics should reach a simple conclusion: ball velocity is bunk.