Double-edged sword

Age is a double-edged sword. With each passing year, we gain knowledge, wisdom, and experience. That's the good edge. The bad edge is that turning the calendar erodes our physical abilities. If you're, say, an NFL running back, then at the beginning of your career, the wisdom comes in large quantities and the physical deterioration is fairly slow. The net result should be improvement from year to year. Eventually though, the weakening of the body overtakes the strengthening of the mind, and the result is the inevitable decline that is only stopped by the end of your career.

So a typical RB should expect to see his overall performance improve steadily for a few years, then start to tail off. Consider for example Brad Muster. Here are his fantasy point totals at each age:

                                   ----------- AGE -----------
                                    23  24  25  26  27  28  29
--------------------------------------------------------------
Brad Muster's fantasy points        49 107 148 112 110  59  15

Brad broke in at the age of 23, then improved until he peaked at 148 fantasy points at the age of 25. He then headed downhill until he was out of the league by age 30 (note: I'm defining a player's age for a given season as his age as of December 31 of that year. As always, I'm defining fantasy points as yards/10 + TDs*6). Graphically, that looks like this:

What you're seeing there is age on the horizontal and fantasy points on the vertical. You can see that he exhibits the pattern described above: improvement, peak, decline.

Every back's career looks different of course. Walter Payton, for example, peaked early, but sustained his excellence for a long time with just a few blips along the way:

Ironhead Heyward peaked late:

and Ickey Woods peaked very early, and then quickly faded away:

In order to get a handle on a "typical" aging pattern, what we need to do is to look at the curve exhibited by every RB we can find, then average those curves together. So I cracked open the ol' database, which consists of complete data for every player who was active anytime after 1984. I found all RBs who:

retired in 1998 or earlier;
peaked at 100 fantasy points or more.

The first condition is to ensure that the back's career actually is over so we know for sure when his peak was. The second is to eliminate players who were never a real fantasy factor. Given those constraints, we have 145 RBs to look at. You can see them all here.

Before we average the curves together, we'll make two adjustments. First, while I made no attempt to adjust players' fantasy point total for time missed to injury (because that may well be related to age), I did adjust their fantasy point totals to correct for the strikes of 82 and 87. Secondly, what I care about (for the purposes of this study) is that Walter was the best Walter could be when he was 23 and Brad was the best Brad could be when he was 25. It's not particularly relevant that the best Walter could be is a lot better than the best Brad could be. So before averaging the curves, I'll normalize them so that each of them peaks at 100. Specifically, I'll divide each player's production at each age by his peak production. So Brad Muster's new line would look like this:

                  ----------- AGE -----------
Name               23  24  25  26  27  28  29
---------------------------------------------
Brad Muster        33  72 100  76  74  40  10

OK, let's do it. After averaging together everyone's scores, this is the result:

AGE 22 23 24 25 26 27 28 29 30 31 32 33 --------------------------------------------------------- 0f peak 45 60 62 63 58 54 52 43 40 30 32 28

What this says is that running backs tend to peak at age 25, with a sort of plateau from about age 23 to age 26. Following that is a fairly gentle decline until about age 28, after which the decline becomes steeper.

Note that when averaging, I didn't count players who didn't play at that age. So the "40" at age 30 means that the players who were still playing at age 30 averaged 400f their peak at that age. Not counted are the many backs whose skills declined to the point that they were forced out of the NFL at age 30. So if anything, these numbers might underestimate the steepness of the decline RBs experience in their late 20s and early 30s.

NOTES:

First, a reason to be suspicious of the data presented here. All the players in this study retired between three and 16 years ago. It is conceivable that current players are able to stay healthy and productive longer by taking advantage of improved training techniques and medical procedures not available to the players studied here.

The first thing I noticed when looking at this chart is that the peak is earlier than I would have expected. Most RBs have seen their best season come and go before they turn 26. Here is a breakdown of how many of these RBs peaked at which age:

Age 21 22 23 24 25 26 27 28 29 30+ ------------------------------------------------------------------- RBs who peaked at that age 1 9 27 25 36 15 11 9 7 5

Over 670f the RBs peaked at age 25 or before -- that's a little surprising. (That's Rashaan Salaam who peaked at 21, by the way. John Riggins had the oldest peak, 34.)

Going back to the graph, the first very steep dropoff is from age 28 to age 29. This appears to be when age really starts to show for a lot of backs. Go check out the data page and notice how many backs looked very productive up until age 28, and then fell off dramatically. Running backs turning 29 this year: Charlie Garner, Jerome Bettis, Jamal Anderson, Tyrone Wheatley. I am not predicting the demise of any of these backs, mind you. But historically, running backs do tend to lose about 170f their fantasy production at that age. That's a strong trend to fight against. Betting on a player to buck a strong trend is fine, as long as you have a specific reason for doing so.

Let's check out the other positions. The drill is the same, but for QBs, we require a peak of 150 fantasy points. For TEs, 60. For WRs, it's the same 100 used for RBs.

Here is the "normal" career path for a WR:

AGE 22 23 24 25 26 27 28 29 30 31 32 33 --------------------------------------------------------- 0f peak 41 50 58 60 64 64 57 57 51 50 46 46

This differs from the typical RB career trajectory in a couple of ways. First, the peak is later -- age 27 compared to 25. But also note that the decline is much more gentle. WRs certainly do age more gracefully than RBs. This is also borne out by the number of WRs who peaked at each age:

Age 21 22 23 24 25 26 27 28 29 30+ ------------------------------------------------------------------- WRs who peaked at that age 1 3 11 14 13 24 21 7 5 14

120f all the WRs peaked at age 30 or later, compared to just 30f the RBs.

Now the TEs:

AGE 22 23 24 25 26 27 28 29 30 31 32 33 --------------------------------------------------------- 0f peak 32 46 47 60 57 51 51 45 46 39 36 35

They show a peak at 25 (same as the RBs), but the decline is more gradual (like the WRs).

Age 21 22 23 24 25 26 27 28 29 30+ ------------------------------------------------------------------- TEs who peaked at that age 0 2 8 5 10 9 6 6 4 6

Finally, the QBs:

AGE 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 ------------------------------------------------------------------------- 0f peak 20 30 44 53 52 52 55 57 55 46 41 52 47 42 39 28

Age 21 22 23 24 25 26 27 28 29 30 31 32+ --------------------------------------------------------------------------- QBs who peaked at that age 0 0 2 4 13 8 4 6 3 3 4 9

As expected, the QB data looks very different from all the other positions. As we all know, QBs can be productive deep into their 30s (although generally not quite as productive as they were in their late 20s).

There's an interesting oddity in the QB numbers. The graph shows that a typical QB should peak at age 29 or so, but the most common age for peaks among actual QBs is much younger -- 25. Why the discrepancy? If you go to the data page and look at the guys who peaked at 25, you'll notice lots of Scott Brunners, Scott Campbells, Randy Wrights, and the like. What this suggests is that 25 is a popular age for QBs to be given their first starting job. The ones that don't do well never play regularly again (so age 25 will have been their best year), but the ones that do play well go on to play better in the coming years.

The big dip at age 32 is also somewhat mysterious. Possibly that's the age at which marginal QBs are most likely to lose their job. I don't think it suggests that age 32 is anything to be concerned about for a productive QB.

Conclusions

Here, in very broad general terms, are the typical aging patterns for players at each position:

Running backs: they tend to peak between 23 and 25. From 26 to 28, they generally experience slight declines (around 50er year). After 28, the declines can become much more severe.
Wide Receivers: they tend to peak around age 26 or 27, then decline gradually (about 50er year) after that.
Tight Ends: peak at 25, decline by about 5-100er year after that.
Quarterbacks: the data suggests that 28 or 29 may be when QBs are at their best. It also suggests that you shouldn't worry too much about how old your QB is.

As usual, I urge strong caution when applying these trends to actual human beings. There are lots of factors that can blow these general tendencies completely out of the water.

But again, it's useful to know what the general tendencies are. As I've written before (even in this article), betting against a trend is fine, if you have a specific reason for doing so. Consider Jamal Anderson for example. I flagged him earlier in the article as someone who might start to show some serious age-related decline this year. Now ask yourself: is Jamal Anderson different from a "typical" RB in any meaningful way? Maybe he has a better offseason workout regimen. Maybe he his knee is finally really back to full strength. Maybe his offensive line was just upgraded. Maybe there's a new coaching staff in town who will use him more effectively (I'm obviously no longer talking about Anderson in particular any more). There are a million reasons why someone might be better equipped to deal with age 29 than the typical RB. If you can think of one, great. If not, well, he still may beat the odds, but it's not a good bet.

Another thing to remember here is that a 5-10 0ecline doesn't mean as much for some players as it does for others. Marshall Faulk can lose 100f his production and still be an elite player. So can Eddie George and Marvin Harrison and many others. From one year to the next, age shouldn't be a very large consideration when evaluating these kinds of players. The guys that you should keep an eye on are the borderline fantasy starters who can't afford to take a 50r a 10 0ip in production -- guys like Troy Brown and Bill Schroeder (to pull two names out of a hat; I don't intend to suggest anything in particular about these two).

Data