Dynasty, in Theory: No One Is Worth Four Firsts (Redux)

I used to think that models of how NFL players age were fundamentally flawed. In 2015, to test this, I created a new way to measure player value (estimating the points a player added over starting the best available replacement off the waiver wire every week) and then calculated the value of every fantasy season over thirty years (from 1985 to 2014). Armed with that data, I was able to determine that we were, in fact, probably thinking about age the wrong way. Neat!

Afterward, I was left with spreadsheets with thirty years of fantasy value to play with, so I could ask questions like "just how much value is there in the first round of a typical rookie draft". Based on my best estimates, it was about 3,000 points over a waiver-wire replacement—meaning if someone offered you a perfectly random first-round pick from a perfectly random season, you should expect it to be worth about 250 points of value.

And based on that number, I evaluated one of my favorite and oldest heuristics in dynasty: that no one is worth four firsts. We just passed the 10-year anniversary of that article—certainly long enough that it's worth revisiting.

Is Anyone Worth Four Firsts?

If your first question right now is "what's a heuristic?" you should probably read last week's piece. The short explanation is that they're simple rules of thumb designed to help us make better decisions. If your next question is "why should we care about heuristics?" then again, you should probably revisit last week's piece, which has evidence of how simple rules often outperform much more complicated models in highly uncertain environments.

I ended last week by noting that "...simplicity is not in itself a virtue... To the extent that we use heuristics, it's imperative that we match them to an environment that is conducive to them, and then monitor and test to ensure that they're leading to better decisions in the long run." I want to acknowledge up front that we're going to be failing that test from the get-go! But I still think it is interesting and useful to revisit the rule.

A Few Caveats Before We Begin

My goal isn't to rewrite the previous piece from scratch, so I'll briefly touch on two key points and then move on. First, when we talk about first-round picks, we're talking about hypothetical "perfectly random" firsts, something that doesn't exist in nature. Plenty of players might be worth four firsts... if we already know those four firsts are the 1.09, 1.10, 1.11, and 1.12. This exercise requires us to imagine that we're trading from behind a Rawlsian veil of ignorance.

Second, for the sake of simplicity, we're hand-waving away the idea of a time discount and assuming all production is equally valuable. In truth, I'd gladly trade eight firsts for virtually any player in the league... if those firsts come due in the 2346 NFL draft. The further in the future a pick resides, the less likely I am to benefit from it, and the less valuable it becomes. With those two objections raised and settled...

Why Is No One Worth Four Firsts?

To start with, my own numbers put the value of a perfectly random first-round pick at 250 points, and there are plenty of players in my dataset who were worth significantly more than that. The biggest example was Jerry Rice, who over his career added 2,936 points over a waiver-wire replacement; if you had traded every single first-round pick from 1985 through 1994 to acquire Rice as a rookie (ten picks total), you still would have come out ahead.

But who (outside, possibly, of his own mother) would have traded a decade of first-round picks for Jerry Rice before he played a single NFL snap? More likely they would have waited until he'd already established himself as the best receiver in the league. Rice was a first-team All-Pro in his sophomore season, and set the single-season receiving touchdown record in his third year (in a 12-game, strike-shortened season, no less). After that point, he "only" had 2,399 points of value remaining—no longer worth ten firsts, but still returning a handy profit to anyone who paid four firsts for him.

So Jerry Rice, with the full benefit of hindsight, was worth four firsts. The problem for us is we can't manage our teams with the full benefit of hindsight.

What About With the Benefit of Foresight?

We could resolve to trade four firsts for any receiver who looks like Jerry Rice after three seasons. That approach would have netted us Jerry Rice (though we wouldn't have known at the time what that was worth, as Jerry Rice was without precedent).

It also would have netted us Randy Moss, A.J.​ Green, Anquan Boldin, Isaac Bruce, and Gary Clark, the top five receivers (after Rice) in fantasy value through three seasons. On average, those five players scored 858 more points than a waiver-wire replacement over the rest of their careers, worth about 3.4 random firsts in expectation.

What if we'd looked by age rather than years of experience? The five receivers most similar to Jerry Rice through the end of their Age 25 season were Randy Moss, Larry Fitzgerald, Andre Rison, Andre Reed, and Isaac Bruce; they were worth about 3.3 random firsts from Age 26 onward.

The two best fantasy receivers after Rice were Terrell Owens and Marvin​ Harrison—who produced 1,897 and 1,821 points over a waiver-wire replacement, respectively, worth more than seven firsts apiece. But at what point in their careers would you have traded four firsts for them? Both had slow starts to their career, ranking 46th and 47th in fantasy value through three seasons.

Harrison had a huge 4th-year breakout, but since he was such an old rookie, he was entering his age-28 season. Would you have given four firsts for a 28-year-old receiver with one good year? How about a 29-year-old with two great seasons? If you'd waited any longer than that, he would have had less than 1,000 points of career value remaining.

Owens' career tells a similar story. He didn't post a high-end fantasy finish until Age 27, and he didn't confirm that it wasn't a fluke until Age 28. By the point we were confident enough that he was a star, he was about to turn 29 years old.

This Is the Rule, not the Exception