Photo: USA Today Sports

Welcome to Learning from History, where I dig into every fantasy season since 1985 to find historical comparisons for modern players.

Today, we look at Travis Kelce. Kelce is exactly the kind of player for whom a rigorous historical analysis like this is most valuable, in my opinion; he's relatively old and extremely productive. When we try to think of comparable players, names like Tony Gonzalez, Jason Witten, Antonio Gates, and Shannon Sharpe all come to mind. Each of those four players maintained very strong production well into their thirties, so perhaps we expect Kelce to do the same.

There are two mistakes going on right here, though. The first mistake is something called the "availability heuristic", which suggests that we mentally judge how likely something is to happen based on how easily we can think of examples of it happening. It's really easy to think of examples of tight ends playing a long time because the tight ends who do so invariably become especially famous.

The second mistake is something called "selection bias". We judge Kelce as a great tight end and want to compare him to other great tight ends. But we're comparing Kelce at the midpoint of his career to other tight ends at the end of their careers; there are other tight ends who, like Kelce, might have seemed great at 29, but who we don't judge as great in hindsight precisely *because* they could not sustain their success.

So what do Kelce's historical comparables look like if we ignore our faulty memories and judge him strictly based on the cold, hard numbers?

Comparison #1: Superstars Through Their 20s

First, a quick sanity check. Yes, Travis Kelce really has been on a tremendous pace through his 20s. Of all tight end seasons between 1985 and 2018, only eight players have accumulated more fantasy value over replacement during that decade. Here is every tight end with at least 400 points of value over replacement in his 20s, along with his average per-year (APY) production from ages 20-32 and his total remaining career production.

A few quick bookkeeping notes. Any total marked with an asterisk is excluded from the average below, (so, for instance, the fact that Zach Ertz— who is currently 29— hasn't produced any value after age 30 isn't weighing Kelce down here). In a couple of cases, judgment calls were made. I excluded the "after-30" values of Jimmy Graham and Greg Olsen because I felt they might meaningfully add to that total still. I included the after-30 value of Jason Witten because I felt the overwhelming majority of his fantasy value was probably already accrued. Thankfully, since all players in question were reasonably close to the group average, this does not meaningfully impact the analysis. (Including Graham and Olsen would have raised the "after-30" average by just over a point.)

Before analyzing these numbers, it's helpful to provide a sense of scale. Tight end production isn't as consistent from year to year as running back production, (where the top 1-2 players will consistently get over 200 points of production over replacement). Over the last decade, the #1 tight end has averaged 160.6 points of value. Travis Kelce's 208 points of value last year were the second-most of that span (behind Rob Gronkowski's 219 in 2011). On the other hand, Travis Kelce's value of 101 points over replacement in 2016 was the "worst" season by a #1 tight end.

In 2009, seven tight ends gave at least 100 points of value. The next year, only two did. Position value tends to swing wildly from year to year. With that said, there are some general guidelines. Typically 40 points of value will rank you within the ballpark of 12th overall. The 100-point level usually gets you somewhere in the Top 5 at the position, and the 150-point mark is around where you start to find the genuine league-winning assets.

With that in mind, the first thing worth noting is that the comparable players to Travis Kelce were barely top-12 tight ends (on average) over their age 30-32 seasons. Tony Gonzalez was a true difference-maker. Witten, Gates, and Sharpe were solid starters. Jimmy Graham, Greg Olsen, and Steve Jordan were low-end TE1s. And the other ten players— nearly 60% of the total sample— were either fantasy backups or out of the league entirely.

Even worse for Kelce, the comps show a clear relationship between success before 29 and success after 29. The four players to top 250 points of value in their thirties all topped 800 points of value in their 20s. Conversely, 80% of the players who topped 800 points of value in their 20s also topped 250 points of value in their 30s, with Gronkowski as the lone exception. This is bad news for Kelce because, while he has been terrific through his 20s, he's simply not in the same stratosphere as Gonzalez, Gronkowski, Witten, Gates, and Sharpe.

If we exclude those five from the list, the average production of Kelce's comps falls to 546.4 points of value in their 20s (still quite close to Kelce's 596.1), 26.9 points of value per season from age 30-32, and 85.3 points of value remaining in their entire careers, (adding Olsen and Graham back into the comparison to help pull that average up as much as possible). That is a catastrophic set of historically comparable players and a devastating critique of the availability heuristic.

Fortunately for Kelce, "finding historically comparable players" is quite difficult and there's reason to believe that some of those comps aren't quite as good as they would first appear.

Comparison #2: Production in the Back Half of the 20s

One interesting point in Travis Kelce's favor is that he didn't accrue a single point of fantasy value until age 25. In fact, he's been steadily picking up steam as he goes along, with each season being more valuable than the season that came before. He was worth no points in 2013, 72 points in 2014, 74 points in 2015, 101 points in 2016, 141 points in 2017, and 208 points in 2018.

While Kelce has only been picking up steam, many names on that first list were already effectively done well before they celebrated their 30th birthday. Kellen Winslow Jr., Jeremy Shockey, and Todd Heap were worth 98, 93, and 91 points per season from ages 22 through 26. They were respectively worth just 36, 43, and 24 points from ages 27 through 29.

So here's a similar list as above based strictly on fantasy value in the second half of a player's 20s, from ages 25 through 29. This should clear out the Winslows, Shockeys, and Heaps from the list and leave us with players who were genuinely still going strong as their 30s began.

At first glance, this list of comps might not seem that much more favorable than the first group. The per-year average over ages 30-32 only rose from 42.4 to 49.2. The total career value remaining rose from 199.4 to 254.2, but that's only about the equivalent of one extra TE10 type season.

The reason this comp list is much better for Kelce is not that there are fewer busts, but that the players who were most successful look a lot more similar to Kelce himself. In the first list, the closest success (Shannon Sharpe) was 250 points of value ahead of Kelce himself. In this list, both of the guys immediately ahead of Kelce went on to have stellar production through their 30s. (Of course, both of the guys immediately behind Kelce fell off precipitously immediately after their 30th birthday, so again, it's a mixed bag.)

Another point worth noting is that Kelce's production was slightly below the group average in the first list, but is slightly above the group average in the second. In both cases, how well a player did before their 30th birthday is highly correlated with how well they did after, so looking stronger than historical comparisons is a good thing for Kelce.

Comparison #3: Superstars In Their Very Late 20s

If I wanted to make the absolute strongest case possible for Travis Kelce, I'd point out once again that the most notable thing about him isn't that he has done quite well over the last five years, it's that he's only picking up steam as he ages. So here's one final list of comparable players based on total fantasy value from ages 27 through 29.

If you're a Kelce owner who is a bit freaked out about this analysis so far, this last chart should be a welcome relief. Not only does Kelce find himself at the very top of the list of comps this time (which is especially good for him given how strongly value before 30 has correlated with value after 30 to this point), but the quality of the comps continues to improve. The whole group is now averaging 56.2 points of value per season from ages 30-32, and the five players immediately behind Kelce himself averaged a whopping 92 points of value per season over those ages.

The total group averaged the equivalent of nearly three more top-5 tight end seasons or two more elite season-winning performances. The five tight ends directly behind Kelce averaged over five more top-5 or three more season-winning performances, with Graham potentially able to add to that total. This is as rosy of a picture for Kelce as the historical data can possibly paint.

Final Thoughts

The biggest problems for Kelce is that even the rosiest picture possible is still somewhat limited. There's only been one tight end since 1985 to top 100 points of value per season from ages 30 to 32, and that's Tony Gonzalez, perhaps the biggest outlier relative to his positional peers this side of Jerry Rice.

Four more players averaged 90 points of value from ages 30-32, and that list includes two more names that Kelce owners might find comforting: Todd Christensen and Mickey Shuler. Christensen didn't even record his first career reception until age 25 but was arguably even more dominant than Kelce from ages 27-29. Shuler set career highs at age 27 with 26 receptions for 272 yards before blowing up with top-five finishes every year from age 28 through 32 (plus finishing 4th in points per game during an injury-shorted age-33 season). Both players were already 29 in 1985, the first year in my data set, so they fail to show up among Kelce's comps despite being arguably the two most similar players.

At the same time, many people forget how comparable players like Keith Jackson and Ben Coates really were. Jackson was the only tight end since Mike Ditka to finish first at the position as a rookie. Through his 20s, he finished 1st, 2nd, 2nd, 5th, 2nd, 8th, and 4th at the position. Coates didn't start with quite the same bang, but from age 24 through 29 he ranked 3rd, 1st, 1st, 3rd, 2nd, and 2nd. He essentially matched Shannon Sharpe point-for-point and award-for-award through the '90s. Together, the two players combined for just one startable fantasy season after their 30th birthday, which is why neither is remembered alongside the Sharpes and Wittens and Gateses of the world.

In my analysis of Saquon Barkley, I found lots of fantasy superstars and very few outright busts. Kelce's list is just the opposite. History provides plenty of evidence that Kelce could remain a startable fantasy asset going forward, but I doubt his dynasty owners would be thrilled with a handful of TE8 type finishes over the next five years. On the other hand, we have more examples of tight ends who look like Travis Kelce falling off the map entirely than we do of them remaining high-end difference-makers deep into their 30s.