Over at TheGridFe, I just finished the single season portion of my series on the (statistically) greatest regular season quarterback performances in NFL history. I’ve discussed the stats, as they are, which always seem to paint modern quarterbacks in a much better light. I’ve prorated for season length, which can sometimes produce a few curious results. I’ve also applied both hard and soft inflation adjustments to account for the evolution of the position and increase in its usage rate. ^[1]I also presented TAY/P+ scores, but that’s not particularly relevant to this discussion. Check it out anyway, though. After talking with Adam Steele and agreeing that maybe even the most moderate approach still left seasons like Sid Luckman’s 1943 or Dan Fouts’s 1982 getting far more credit than they probably should. So I went ahead and made an even weaker era adjustment, which I will discuss briefly in this post, to try to mitigate the effects of the original modifications.

My main purpose for writing today isn’t to give you another list of great quarterback seasons, although I will do that as well. My goal is to solicit the opinions of the Football Perspective readers, whom I respect for their thoughtful and reasoned nature. I have two primary questions:

1. Is it necessary, or even right, to make adjustments for changes in both schedule and QB usage rates?
2. If we believe it is right to reward older players with schedule and inflation corrections, should we also penalize them for playing in weaker leagues? ^[2]This doesn’t mean the AAFC and the AFL; it includes those, but it also includes the pre-UFA, pre-big-money, pre-AFL-merger, pre-integration, pre-AFL-merger, pre-T-Formation, and WW2 eras of the NFL.

I will provide two tables below with several examples of what it looks like to make the changes I mentioned in the first question. I have no such table for the second question. In fact, this study from Jason Lisk is the only attempt I have seen at comprehensive comparison of the relative strengths of the NFL and AFL. ^[3]I used that article as the foundation for my own research and came up with some multipliers for AFL stats, but they have received mixed fanfare from the few to whom I have showed them. The more … Continue reading I’ll talk a little about both after we look at some numbers.

The Basics

Since the NFL began running a standardized schedule (1936), there have been 94 professional seasons between the established league and its two greatest rivals. In those years, we have seen schedules range from a mere nine games to the current sixteen-game format. We have also seen a range of quarterback plays per game from 19.5 to 40.9. As you’ll see, this can create pretty extreme numbers when we attempt to put these seasons on a level playing field.

The table below probably isn’t for everyone, but I think many among you will find it instructive. It displays the 94 seasons mentioned above, along with the numbers we get after using different methods to prorate and adjust for inflation. Read it thus: The 2015 NFL featured a 16-game schedule and saw quarterbacks engage in 40.8 plays per game. ^[4]Note that when I say a year, I actually mean the three year span with that year in the middle. In order to increase my data set, I aggregated each three year span to find my averages. The exceptions, … Continue reading The multiplier to prorate the schedule is 1.00 (16/16). To find the lower multiplier, just take the average of the higher multiplier and 1.00. In this case, the lower multiplier is still 1.00. To find the hard inflation adjustment, take the season’s plays per game and divide it into the historical plays per game (34.7). For 2015, that number is 0.85. To arrive at the soft inflation adjustment, find the average of the hard adjustment and 1.00. For 2015, that’s 0.93. To find the weakest adjustment, divide the sum of the hard adjustment and two by three [(0.85 + 2)/3]. The quotient is 0.95 for 2015. Once you have those figured out, you can used any combination of schedule and inflation adjustment to modify a quarterback’s yearly value. The last six columns show us what we end up doing to a season once we apply these combinations.

Depending on which combination of adjustments you implement, your results can vary drastically. The last few seasons receive the most significant possible penalties, with their values only given 85% credit in some cases. On the other hand, quarterbacks in 1936 receive as much as 238% credit for their output. I know this isn’t important to some of you, and it might not even be that helpful in the abstract, but I think it’s a useful tool to see how much credit we are adding – or taking away from – players based on the era they happened to play football. If you’re more interested in a practical example of what this means, you may prefer the next table.

The Results

To set up the table below, I calculated every quarterbacks Total Adjusted Yards per Play since 1932. ^[5]Those of you familiar with the stat from reading about it here or at my site should be warned that this is a stripped-down version of the original. Proper TAY/P includes first downs and removes … Continue reading I then found their efficiency relative to the rest of the league. This is similar to how Chase calculates RANY/A, but for a few small differences: first, I back out the player’s stats from the rest; second, I aggregate years n – 1, n, and n + 1 prior to finding the average. The result is what I have called value per play. All we have to do now is multiply Val/P by total plays to find total value over average. ^[6]We could try to look at replacement value as well, but I think a higher baseline is better when trying to determine the very best individual seasons. We’ll save replacement value for the career … Continue reading

Once we have the actual, unadulterated value, we can apply our combinations of era modifiers to see what happens. I have done just that. Read thus: Peyton Manning, playing in the 2004 NFL, had 535 plays at a rate of 9.16 TAY/P. His marginal value of 4.32 leads to a total value over average of 2312, tops in history. However, that value changes to 2171 with a ProHard modification, 2241 with a ProSoft modification, and 2265 with a ProWeak modification. Because he played in a 16-game season, his LowHard, LowSoft, and LowWeak modifications are the same as his Pro ones.

The most glaring issue that arises is what happens when we sort by the ProHard column. Not only does Luckman’s 1943 season jump to the top, it does so by an astronomical margin. Some might argue that he was a dominant force that year and deserving of the top spot. I think 1943’s 2.19 total multiplier gives him way too much credit. I’d say the same for Fouts in 1982, as well as for every quarterback to play in a 12-game-or-less season. ^[7]Luckman’s season ranks 15th all-time without any adjustments, while Fouts’s ranks 99th. The harshest adjustments see Luckman rank as high as first and Fouts rank as high as seventh. You may agree … Continue reading It isn’t until I apply the weakest of all era adjustments that Manning and Dan Marino jump back to the top two spots. Luckman still outranks Tom Brady’s blitzkrieg 2007 campaign. It’s clear to me that seasons with low attempts, whether because of a shorter schedule or lower usage rate of passers, seem to receive undue credit. Perhaps that’s a nice balance to the fact that those seasons rarely receive proper credit from fans in real life. I don’t know.

Maybe the issue isn’t that I am adjusting for era so harshly. Maybe the real issue is that I am doing so without discounting for the relative weakness of the league that Luckman (or Sammy Baugh, or Cecil Isbell, or Otto Graham, or George Blanda) dominated. That’s where the second question I asked at the beginning comes in. Is it a good idea to start applying some sort of discount to these eras? If we do, should it be based on rigorous analysis, or should it be based on what feels right? I don’t have the answers to these questions. If I did, I wouldn’t crave your input. What do you, esteemed readers of Football Perspective, think about the idea of era adjustments and discounts?

Let me know in the comments. Thanks for reading.