Photo Credit: Matthew Deery via Flickr Creative Commons

This is Part II of a three-part series (Part I).

Introduction

The goal of this three-part series is to provide some perspective on Cody Kessler.

Part I of this series attempted to evaluate Kessler’s 2016 season when compared to other rookie quarterbacks since 1970. Using a simple, five-category z-score measure to control for changes in the NFL since that time, the analysis suggested that Kessler posted elite rookie marks for completion percentage and interception rate, but also took sacks at an exorbitant rate. Summing across all categories, Part I of this series ranked Kessler as having the 33rd best rookie season out of 96 qualifying rookie quarterbacks.

While Part I attempted to provide a historical perspective on Kessler’s rookie campaign, this approach failed to offer any predictive insights about what his season might mean for his future in the NFL. In particular, the simple summation of z-scores in Part I allowed for similar rankings of clearly dissimilar quarterbacks; that’s how Kessler can be listed next to Cam Newton and Brandon Weeden depending on the scoring system.

But none of this answers the $64,000 question: Can Cody Kessler become a successful starting quarterback in this league?

Parts II and III of this series attempt to answer this question.

In Part II, this study will attempt to generate a list of Kessler’s closest comparables since the NFL-AFL merger. Given his physical profile and rookie-season statistics, it is hoped that a tangible list of comparables will provide all of us as Browns fans a realistic expectation of his ceiling and floor as an NFL quarterback. Part III of this series will examine the statistics of NFL rookie quarterbacks since 1970 and explore whether they have any predictive power in explaining the player’s future career value.

Methodology

As outlined in Part I of this series, any historical evaluation of quarterbacks in the NFL is limited to the “basic” statistics that have been available since 1970 (i.e., QBR and Pro Football Focus metrics are not options). The initial five rate statistics explored in this series have been:

  • Completion percentage (comp / att)
  • Yards per completion (yards / comp)
  • Touchdown percentage (td / att)
  • Interception percentage (int / att)
  • Sack percentage [sack / (sack+att)]

While this list was sufficient in evaluating past performance in the previous analysis, projecting future success requires additional information. Two quarterbacks may post similar rookie seasons, but differences in their respective physical tool sets may make for highly divergent future projections. Cody Kessler and Cam Newton, for example, were ranked back-to-back in one passing metric in Part I, but it would be reckless to project Kessler to have the same type of career as Newton given differences in body build and athleticism, among other factors. As a result, Part II of this analysis includes two additional variables:

  • Rush yards per game (yards / game)
  • Height (inches)

The inclusion of these metrics attempts to incorporate the commonality of a player’s physical characteristics to ensure more appropriate comparables between players. While these aren’t perfect measures, rush yards per game is included as a proxy for a player’s overall level of athleticism and playing style. Height is included to account for differences in physical build and, to a lesser extent, arm strength. While some thought was given to include a player’s draft round as a proxy for physical skill sets, it is reminded that players fall to lower rounds for all sorts of reasons. Quarterbacks with such vastly different physical skill sets end up in the same draft round—Andy Dalton and Randall Cunningham were both second-round picks, as one example—so using the two variables above to control for physical comparisons seems to be a preferred method.

In building a list of player comparables, however, a simple comparison of raw statistics between quarterbacks of different eras is inappropriate given alterations in NFL offenses due to rule changes and strategic innovations. As a result, this analysis normalizes player seasons as compared to the average value of quarterbacks with 150+ pass attempts in each year. The result is a set of seven z-scores for each rookie when compared to the league average; this step puts every QB since 1970 on a level playing field when it comes to evaluating their rookie-season statistics (for more, see the Kessler-Bernie Kosar comparison in Part I). The list of z-scores for Kessler is listed in Table 1 (below).

table-1-kessler-z-scores

Kessler’s values in Table 1 reflect the common narrative surrounding his career thus far. As discussed above, Kessler posted a completion percentage and interception rate that were substantially better than the league average (i.e., a positive z-score). As outlined in Part I, Kessler’s z-scores in these two categories were in the top 10 of all rookies since 1970. In all other categories, Kessler was below league average in 2016. This, by itself, is not surprising; most rookie quarterbacks feature negative z-scores in most, if not all, categories. Kessler’s worst scores were in sack percentage and height.

While Table 1 provides a statistical profile of Kessler’s performance and physical tools, additional steps are required to generate a ranked list of player comparables. To develop an index of similarity, this study will calculate a standard Euclidean distance measure by summing the squared differences between the seven z-scores between all pairs of quarterbacks in the sample. As an example of this process, Table 2 (below) provides the calculation of the “distance” between Kessler and Bernie Kosar’s 1985 rookie season.

table-2-kessler-vs-kosar-part-ii

As demonstrated in the table, Kessler and Kosar posted similar year-adjusted values for interception percentage, touchdown percentage, rush yards per game, and yards per completion. However, there are considerable differences elsewhere. While Kessler posted a better year-adjusted completion percentage than Kosar, he also took sacks at a much higher rate and is substantially shorter than his 6’5” predecessor. To measure these differences, the right-hand column represents the squared value of the distance between Kessler’s and Kosar’s z-scores. By adding these seven numbers up and taking the square root, the methodology suggests a 4.146 distance score between Kessler and Kosar. After similarly computing this distance score for Kessler’s when compared to all other quarterbacks in the NFL since 1970 with 150+ pass attempts, the next section will outline the players that emerge as the closest comparables based on this distance score.

Results

Table 3 (below) offers the best comparables for Cody Kessler based on height and rookie-season performance.

table-3-best-kessler-comparables

As one may have guessed from the title of this article, Table 3 reflects that Minnesota quarterback Teddy Bridgewater ranks as Kessler’s best comparison.

Hold that thought for now.

After Bridgewater, Table 3 features players that offer a mixed bag of career outcomes. Among the most successful QBs on the list, Jim McMahon (third) won the Super Bowl with the 1985 Chicago Bears and made the Pro Bowl in the same season. Neil Lomax (sixth) led the NFL is passing yards in 1987 and made two Pro Bowls with the Arizona Cardinals before being forced to retire due to an arthritic hip. Jeff George (tenth) compiled over 27,000 yards in a 12-year career; his inclusion was unexpected, but he also struggled with taking sacks and posted some rookie-season z-scores nearly identical to Kessler in terms of yards per completion and rush yards per game.

Not every quarterback on this list had such career success. Three players in Table 3—Charlie Batch, Colt McCoy and Nick Foles—have been, or currently are, seen as low-end starters or high-end backups. Finally, the list includes three others who either flamed out quickly (Charlie Frye) or had long careers as seldom-used backups (Tom Hodson, John Reaves). In other words, given the divergent outcomes offered in Table 3, this list does not provide immediate clarity about the fate of Kessler’s career.

In putting Table 3 in context, it is important to note that even the closest comparables to Cody Kessler are still quite far away in terms of the “distance” score; in other words, there are no ideal comparisons. In fact, Kessler is the only rookie quarterback in modern NFL history to exhibit the combination of signs (positive/negative) on the seven z-scores as reflected in his totals in Table 1. This may explain why it has been so difficult for many of us Browns fans to evaluate Kessler and his potential as a starting QB.

But let’s return to Bridgewater.

Yes, Teddy Bridgewater is the closest comparison to Kessler based on height and rookie-year statistical profile. But consider what happens when the list of comparables is re-estimated to include all quarterbacks, regardless of rookie status, in their age 23 season (Kessler’s age in 2016). The results are presented below in Table 4 (below).

table-4-best-kessler-comparables-at-age-23

Bridgewater’s name again appears at the top of the list of age-23 comparables to Cody Kessler. But what stands out here is that (a) this represents Bridgewater’s second season in the NFL and (b) the distance metric (1.16) indicates a far superior match than any other player-season evaluated thus far.

How good is that estimated match?

Bridgewater’s 2015 season represents the closest match to Kessler among all quarterbacks in modern NFL history, regardless of age and rookie status.

And it wasn’t even that close.

For some context, that’s a sample of 1,643 unique player-seasons. Across all ages. Since 1970.

Taking a step back, the comparison between Kessler and Bridgewater makes a lot of sense. Consider Bridgewater’s identified strengths and weaknesses:

Strengths

Weaknesses

Without posting comparable links from Cleveland sources, this sure sounds like the current perspective on Cody Kessler. Yes, there are differences besides their current state of health and skin color. But the general scouting report on Bridgewater would seem to affirm him as an excellent comp to Kessler.

Before making too much of these comparisons, it is important to remember that many comps between players often fail. And as Browns fans, we should be mindful of the last big Cleveland-Minnesota quarterback comparison: Mike Mayock comparing a pre-draft Johnny Manziel to Hall-of-Famer Fran Tarkenton.

So this whole conversation needs to be taken with a grain of salt.

That said, when looking past Bridgewater in Table 4, the list of comparable players offers mixed perspective on the future of Cody Kessler. The next two closest comparisons—Hodson and David Whitehurst—had only modest NFL careers. The next three players enjoyed long careers as NFL starting quarterbacks: McMahon, Chris Miller, and Dan Pastorini. Finally, the bottom of the list features two of the best QBs of all-time: Brett Favre and Drew Brees. That said, the distance score on these latter two is rather substantial, suggesting that these are not the greatest statistical comparisons.

Nevertheless, the inclusion of Brees on this list is eye-opening given that he was regularly mentioned as Kessler’s ceiling after the latter was drafted. And Brees’ statistical profile in his age 23 season has compatibility in all categories except one: sack avoidance. While Kessler was substantially worse than league average in that area (Z=-1.82), Brees—who stands 6’0” tall—was considerably better than the NFL mean (Z=0.73) at avoiding sacks during his age-23 season in 2002. In fact, Brees currently ranks fifth in NFL history in career sack rate.

Kessler’s disappointing sack rate was discussed at length in Part I. The concern here is that sack avoidance seems to demonstrate a positive correlation with career success. To provide some perspective on this issue, 15 qualified rookie quarterbacks have posted a z-score of -1.0 or worse in the NFL since the Browns’ return in 1999. Only one—Donovan McNabb—developed into a franchise QB. The list also features a solid starting QB (Alex Smith) and one who may become a franchise starter (Marcus Mariota). Other than that? Outside of two rookies (Kessler and Jared Goff), the list includes a lot of players who would qualify as busts (e.g., Tim Couch, Akili Smith, Blaine Gabbert) and one or two who developed into viable long-term backups (e.g., Colt McCoy).

Can Kessler improve at sack avoidance? I have no idea. It is not clear how much blame should be—or should not be—assigned to Kessler for his high sack rate in 2016. Nevertheless, I am not aware of research that examines how a quarterback can evolve in this area. Maybe improvements in the interior of the offensive line can fix everything. Maybe not. This is a topic that I may return to in a later analysis but, until then, it remains a mystery.

And what if Kessler improves at staying upright in the pocket? When one removes sack rates from the analysis above and instead only focuses on the z-scores from the other six categories, Kessler’s top comparables in modern NFL history among rookies and quarterbacks in their age-23 season improve dramatically, as demonstrated in Table 5:

Table 5 - Kessler Excluding Sack Percentage

This is obviously misleading; being able to avoid the rush is an important part of being a quarterback. But this list should offer some insight on Kessler’s primary weakness in 2016—it may not have been arm strength after all—and what his ultimate ceiling may be if there are vast improvements in this area.

Conclusion & Caveats

Part II of this series on Cody Kessler aimed to provide a list of historical player comps in order to provide us Browns fans some perspective on his likely career arc as a quarterback in the National Football League. While this study was successful in identifying Kessler’s closest comparables, I’m not sure how much clarity it offers any of us in trying to determine if Kessler can be a viable, playoff-caliber quarterback in the NFL.

This outcome is not terribly surprising. But in looking at the numbers in the analysis above, there are two reasons for this. First, the list of player comps featured an enormous range of potential outcomes. Each table seemed to offer equal parts optimism and pessimism when it came to projecting Kessler’s career. Consider the range of outcomes identified:

  • Future Hall-of-Famer: Drew Brees
  • Solid starter: Neil Lomax
  • High-end backup: Charlie Batch
  • Marginal backup: Tom Hodson

The second frustration is that the best historical comparison for Kessler—by far—is Teddy Bridgewater.

While I personally found the Kessler-Bridgewater comp enlightening, this comparison offers little help in establishing the most likely career arc for Kessler. Bridgewater’s gruesome knee injury, at minimum, will substantially alter his career trajectory. Could Bridgewater have become a championship-caliber quarterback had he stayed healthy? We’ll never know. And even if Bridgewater is ever able to come back at full strength, any insight we gain on his career may come after the Browns will have had to have made a decision on Kessler as a quarterback in Cleveland.

In other words, despite our best efforts, none of us can say with certainty how Cody Kessler’s NFL career will pan out. It’s simply too soon.

Before wrapping up, there are a couple of caveats to this study that need to be addressed. First, this analysis is based on Kessler’s 195 pass attempts in 2016. This is a rather small sample size, especially when using it to make inferences about the future. In contrast to Kessler’s half-season numbers, many of his comparables featured 400+ pass attempts. Would Kessler have become more acclimated to NFL game speed and grown as a passer with more starts? Would defensive coordinators adjusted their schemes to take away his strengths? We obviously can’t know what would have happened with more playing time, but the resulting small sample size does lead to increased caution about the applicability of these comparisons.

Second, the distance score used to generate a list of player comparables gives equal weight to each of the seven categories analyzed. But it is quite likely that these categories have dissimilar predictive power when it comes to estimating a player’s career value. For instance, it would stand to reason that yards per completion should be more heavily weighted than, say, rushing yards per game. And what about sack avoidance? In the case of Cody Kessler, do his struggles with sacks completely offset his high marks in completion percentage and interception rate? How should we weight these categories? What is the probability that Kessler can evolve into a successful starting quarterbacks?

These questions and more will be explored in Part III of this series.