This may seem like an odd confession coming from a person who
- writes for a fantasy football site that introduced the world to the concept of VBD, a pioneering new way to leverage projections,
- produces returner and upside projections for that same site, and
- creates dynasty player value charts that are heavily projection-based,
but I’m not entirely a fan of projections in fantasy football, and especially in dynasty. I think that the more uncertainty a system contains, the more likely projections are going to be wildly off for an entirely unforeseeable reason. I also believe there’s a real risk when projecting to gain a false sense of precision.
Instead of projections, I like to rely on something called “heuristics”, which are tools that boil down incredibly complex concepts into very simple rules of thumb. I don’t do this because it’s easier, (although as a happy coincidence, it is). I do this because, in highly uncertain environments, heuristics can perform as well or better than projections. This article from the Harvard Business Review lays out the case for heuristics.
As an example, my dynasty value charts build on two very basic heuristics. First, player performance does not decline gradually, but instead it falls off suddenly and dramatically. Second, the older a player is, the more likely he is to hit that sudden and dramatic decline.
Projections might try to understand the cause of that decline. Is it the player losing a step? Is it injury-related? Is it retirement? My heuristics are silent on that matter; I don’t know why a player suddenly falls off the cliff. A player can fall off for any number of reasons. All I know is that those various reasons go up as a player ages.
Calvin Johnson retired. Torry Holt wore down all the cartilage in his knees. Terrell Owens became too big of a headache for teams to bother with anymore. Joey Galloway finally lost a step. The end can look different, but regardless of the form it takes, it’s more likely to come the older you are.
Those two heuristics allow me to do something simple, but powerful. I can estimate a player’s “true” performance level, (for this I rely heavily on Bob Henry’s projections— just because I don’t do them doesn’t mean I don’t recognize their value, and Bob is among the best in the business). I can estimate a player’s remaining career, (using historical data to create league-wide “mortality tables”, much like those used by life insurance companies to set your premiums). And then I can just multiply “true” value times years remaining to get career value remaining.
But that’s not the only heuristic I use in dynasty. There are plenty of others. “Value upside over floor in backup quarterbacks”, for one example. “Buy injured players” for another.
And then there is one of my bedrock heuristics, one I’ve become known for over the years.
No one is worth four firsts.
To clarify and spell out my rule of thumb: by “four firsts”, I’m talking about four random 1st round picks in the next rookie draft. The “random” part is important because a top-end star could conceivably be worth the 1.09 + 1.10 + 1.11 + 1.12. Part of the value of “random” firsts is that there’s a chance you’re getting a top-3 pick out of it, because top-3 picks are disproportionately valuable. (In fact, there are very few players who would be worth even two top-3 rookie picks.)
Also, the “in the next rookie draft” part is also crucial because future seasons become progressively less valuable the further away they are. I would pay four firsts to acquire anyone… provided those four firsts weren’t coming due until 2117.
I’m also not saying that there’s never been anyone who was worth four firsts with the benefit of hindsight. Last year I created historical value spreadsheets of every player over the 30 years from 1985 to 2014, (you can peruse those spreadsheets for yourself at the provided link). My preferred player value metric is EVoB, or “Estimated Value over Baseline”— essentially points per game minus baseline points per game times games played.
By my estimates, a random 1st round pick is worth about 250 points of EVoB. Ignoring time discounts, Jerry Rice’s career would, therefore, be worth just a hair shy of twelve first round picks.
Twelve.
The problem, though, is that I’m saying this with the full benefit of hindsight. Yes, after the fact, it’s easy to say that Jerry Rice was worth twelve first-round picks. But at the time? Jerry Rice was the best receiver in the NFL when he was 28 years old, but nobody in the world would have guessed he would still play for another fourteen years after that.
There are a lot of wide receivers who people thought might be the next Jerry Rice. After his age-27 season, Randy Moss was 200 points of value ahead even of Jerry Rice to that point, (1167 EVoB to 937 for Rice). Many thought he was going to challenge all-time records.
Was Randy Moss worth four firsts at that point? No, he was not; he produced just 618 more points of EVoB, not quite two and a half firsts worth of value, and that’s even including his magical 2007 season in New England. By age 33, Randy Moss’s career as a fantasy-relevant option was over.
After his age-27 season, Calvin Johnson wasn’t even worth two perfectly random 1st round picks, as he wound up retiring in his prime.
Marvin Harrison, on the other hand, had 1371 EVoB after his age-27 season, the equivalent of about five and a half firsts worth of value. The problem? Nobody would have paid four firsts to acquire him; at the time he was coming off the only top-20 finish of his career. If you waited a year to make sure it wasn’t a fluke, he would only have about 4.5 firsts worth of value left. If you waited two years just to be extra sure, he would have fallen under four firsts worth of value, and it would be too late.
And trying to be extra aggressive rather than waiting for a second or third year to confirm things is no help, either; after 2007, Braylon Edwards was a 25-year-old former 3rd overall pick in the NFL draft coming off of a 16-touchdown season. He’d produced 160 EVoB in his age-24 season alone. If you’d aggressively traded four firsts for him, you would have gotten killed; he produced less than 100 more EVoB over the rest of his career.
This is the balancing act, and why the heuristic works. If you bid on a player who has a short track record, that track record is more likely to be a fluke, and you’ll lose your shirt. (See also: Roy Williams, David Boston, etc.)
On the other hand, if you wait a couple of years to make sure a player isn’t a fluke before you trade four firsts for him, half of his career value will already be in the rearview window, and he won’t have as much time left to make up the cost.
The net result is a paradox that Joseph Heller would love: the more certain we are that a player is worth four firsts, the less that player is worth four firsts. If we buy too early, we’re likely to wind up with fools gold. If we buy too late, we’re likely to wind up with leftovers. No one is worth four firsts.
One Final Thought on the Problem With Heuristics
While they are great tools for getting your arms around incredibly complex problems, heuristics’ simplicity is also their weakness. Because creating a simpler framework doesn’t eliminate the inherent complexity, it just allows you to navigate it.
The idea that no player can be simultaneously (a) good enough, (b) young enough, and (c) proven enough to be worth four perfectly random first round draft picks, for instance, is based on history. When something comes along that history hasn’t seen before, heuristics can break down.
Consider, for instance, the case of Odell Beckham, Jr. Beckham had about 180 EVoB at age 22. He had about 180 more at age 23. Since 1985, (as far back as my data goes), there has been one season of 150 or more EVoB at age 21, (Moss), four more at age 22, (Moss, Fitzgerald, Gordon, and Beckham), and seven more at age 23, (Moss, Bruce, Boldin, Calvin, Rison, Boston, and Beckham).
As you can see, the list of players with one truly phenomenal fantasy season at that young age still contains a pair of outright busts, (Gordon and Boston) and several players who, while long-time quality contributors, fell just shy of four firsts, (Bruce, Boldin, and Rison). Larry Fitzgerald probably barely returned positive value over four firsts on a theoretical trade at age 24. Moss, on the other hand, easily returned positive value.
But Beckham isn’t on that list just once. He’s on it twice, the only receiver other than Moss who can make that claim. And just because Fitzgerald wound up in quarterback purgatory and Gordon ran afoul of the substance abuse policy doesn’t mean Beckham will, either; there’s real danger in drawing sweeping conclusions off of small sample sizes like this.
What I’m getting at is that we cannot automatically assume that Beckham *is* worth four random first round draft picks. But on the other hand, given the uniqueness of his age/talent/production combination, we can’t definitively say he’s not, either. There’s only one other 24-year-old in history with this kind of pedigree and production profile, and that’s Randy Moss.
But extreme historical outliers like Odell Beckham, Jr. are extreme historical outliers. And so, beyond possibly tacking on a hedge at the end, our heuristic stands.
No one is worth four firsts. (Except possibly for Odell Beckham, Jr.)