No metric is perfect.
Every advanced statistic we have ever developed has flaws that prevent it from being used without proper context.
Unsurprisingly, passer rating has many flaws. It is extremely outdated, it applies too much weight to touchdowns, interceptions, and completion percentage, while completely ignoring sacks. (For a more thorough description of passer rating’s deficiencies, read our article here.)
That’s why I created Adjusted Passer Rating.
My goal was to simply repair some of the well-known shortcomings of passer rating. The concept of passer rating isn’t bad; the execution is where it went wrong. I didn’t want to entirely stray away from the foundation upon which Don Smith’s committee created passer rating in 1971. Other metrics (ESPN’s Total Quarterback Rating and Pro Football Focus’ grading system) attempt to more accurately quantify quarterback ability by accounting for “clutchness,” receiver drops, etc. This is more useful in many situations. However, the intention of Adjusted Passer Rating is to create a better version of passer rating while sticking with our traditional objective passing stats (yards, touchdowns, etc).
The traditional formula for passer rating is difficult to digest at first. As you can see, it is made up of four main components — completion percentage, yards per attempt, touchdown percentage, and interception percentage.
At first glance, the formula for Adjusted Passer Rating is no simpler.
Let’s break it down.
NY/A is net yards per attempt. Instead of being calculated as passing yards divided by passing attempts (which is used in passer rating), it factors in sacks. The formula is (passing yards – sack yards) / (passing attempts + sacks).
Completion percentage was removed because NY/A already rewards passers for completing a higher percentage of their passes, so there’s no point in double counting completions. Also, passers are never rewarded for completing passes that don’t gain any yards (like passer rating).
The weighting for touchdowns and interceptions were altered to be more accurate. Due to studies done by statisticians, passing touchdowns are worth 20 yards while interceptions are worth approximately -50 yards. Passer rating gave these plays a value of 80 yards and -100 yards respectively, a massive overweight.
The x, y, z variables represent the league average for the stat in a certain season. For example, the average NY/A was 6.42 in 2018, so 6.42 is the x value this year. This would be subtracted from Patrick Mahomes’ NY/A of 8.13 to see how much he is performing above average.
1.5 is added to the part of the equation within parentheses so that if a player is performing at an average level, the result of the calculation would be (0+0-0+1.5)*(100/3), or 50. Therefore, 50 is an average rating.
Like passer rating, there is a minimum and maximum score a player can receive. The sum of the values within the parentheses cannot be greater than 3.0, so the maximum rating is 100. Also, the minimum rating is 0, meaning the sum of the values within the parentheses must be positive. A 0-100 scale with an average of 50 is probably more intuitive than a 0-158.3 scale in which 66.7 is average.
I applied Adjusted Passer Rating to all 1,604 qualifying quarterback performances since 1967: check it out here. There’s an infinite amount of interesting things to observe, but I’ll point out what sticks out to me.
(1) Peyton Manning’s 2004 season is rated as the greatest passing season in NFL history. That’s not very surprising — it ranks second all-time in passer rating (121.1) behind Aaron Rodgers’ 2011 MVP campaign (122.5). It’s also not the most unpopular opinion to say it’s the best quarterback season of all-time.
(2) Steve Young’s ratings from 1991 to 1994: 76.51, 78.71, 70.49, 77.68. That’s incredible. Since 1990, there have been just 11 single-season quarterback performances which earned a rating of at least 75 and Young accomplished this feat thrice in a four-year stretch. In this span, Young was also a 3x First-Team All-Pro and he won Most Valuable Player honors twice.
(3) While Patrick Mahomes’ incredible MVP season is described by some as a top-5 passing performance in NFL history, Adjusted Passer Rating ranks it at a mere 17th … in the past 18 years.
(4) The median career values of the four quarterbacks who are most often compared in the modern era: 1. Peyton Manning (65.92), 2. Aaron Rodgers (62.83), 3. Tom Brady (61.1), 4. Drew Brees (60.38). Not the most surprising results. Steve Young’s 70.39 median rating (average of 66.78) is amazing, but he only played 9 qualifying seasons. Dan Marino’s 62.08 median rating lags slightly behind Joe Montana’s 62.64.
(5) Jared Goff’s Adjusted Passer Rating of 20.32 in his 2016 rookie campaign is the second lowest rating of the past 18 years (JaMarcus Russell, 20.14). In his two seasons since, he has posted two ratings between 60 and 65, which is above average. Was his rating that bad because of Jeff Fisher? Or is the increase because of Sean McVay? Both?
Other Problems With Adjusted Passer Rating
While it’s an improvement from traditional passer rating, Adjusted Passer Rating is still very much imperfect.
One of the objectives was to tackle the problem of modern passing inflation. The result was that while modern passing inflation was fixed, the new issue is the opposite. Consider that there have been 67 single-season passing performances with a rating of at least 70. That’s a little over 1 per year, so it’s a very elite grade to receive. 21 of these 67 performances were in the 1970s (3 in the 60s, 9 in the 80s, 11 in the 90s, 13 in the 00s, 10 in the 10s). This means that an annual average of over 2 passers received a rating of over 70 in the 1970s.
Why? It must be a result of a greater skill-gap between the best and worst quarterbacks in the league that existed in the 1970s. The bad quarterbacks brought down the league averages, letting the good quarterbacks stand out even more. There have been 37 single-season passing performances with a rating that was less than 30. 15 are from the 1970s. There was essentially no “middle-class” of quarterbacks. The great quarterbacks had inflated ratings as a result. For example, Roger Staubach’s 85.52 rating is the second-highest of all-time. He didn’t even win MVP, though — defensive tackle Alan Page did. Unfortunately, there’s no real viable way to fix this new inflation.
Additionally, Adjusted Passer Rating does not tell the full story. There are many qualifiers to passing stats that are relatively new. For example, some modern metrics factor in wide receiver drops and, “garbage time” situations, and yards after the catch. Both the old and new forms of passer rating have no way to account for these nuances without having to disregard all of NFL history before the 1990s. That wouldn’t work — the goal was to create a rating that could be used even with the limited statistics of the early Super Bowl era.
Passer rating is also exclusively a way to judge passing efficiency. It is not a full grade for quarterbacks. This has always been the case, but it is often misunderstood. Rushing yards are intentionally not accounted for, nor are intangibles that make players like Tom Brady great (clutchness, leadership, football IQ, etc). Remember, the initial purpose of passer rating was to crown a passing leader every season. It wouldn’t make sense to crown a passing leader based on rushing yards.
The fact that this metric is based on league-averages over a single-season makes this very susceptible to large shifts in rating over a smaller sample size. A player may have a massive touchdown percentage in a single game in an outlier performance. Their touchdown percentage would be compared to the league-average over the course of the entire season, meaning they would be far more likely to score a perfect 100 (or close to it) even if their touchdowns don’t tell the full story. This is why Adjusted Passer Rating is not best suited for being used to grade single-game performances. It’s best to rate quarterbacks over a full season.
Like any other statistic, Adjusted Passer Rating is not perfect. Its best use is to compare passing performances of different eras. It’s not the best to determine whether one quarterback is better than the other. Once somebody understands a metric’s strengths and weaknesses, it can be used effectively.