Dynasty, in Theory: Trading for Future Picks

There's a lot of really strong dynasty analysis out there, especially when compared to five or ten years ago. But most of it is so dang practical-- Player X is undervalued, Player Y's workload is troubling, the market at this position is irrational, and take this specific action to win your league. Dynasty, in Theory is meant as a corrective, offering insights and takeaways into the strategic and structural nature of the game that might not lead to an immediate benefit but which should help us become better players over time.

Consistently Inconsistent

I wrote two weeks ago of my favorite Walt Whitman quote-- "Do I contradict myself? Very well then I contradict myself, (I am large, I contain multitudes.)" I'm also a fan of Ralph Waldo Emerson ("a foolish consistency is the hobgoblin of little minds"), Oscar Wilde ("I have always been of opinion that consistency is the last refuge of the unimaginative") and F. Scott Fitzgerald ("the test of a first-rate intelligence is the ability to hold two opposed ideas in the mind at the same time, and still retain the ability to function.")

(Given the heavy use of quotations, perhaps we should instead consider T.S. Eliot: "One of the surest of tests is the way in which a poet borrows. Immature poets imitate; mature poets steal.")

All of this is to suggest that consistency is not something you should expect from this column.

I wrote last week about how casually fantasy analysts tend to write about investing in the future in a way that ignores the reality of many dynasty leagues, where it's not guaranteed that any given team will still be around to reap the rewards of that investment. This is probably bad. In terms of practical applicability, dynasty analysis could perhaps benefit from a shorter-term focus.

With that in mind, I'd like to introduce this week's topic: Trading for 2025 draft picks. Should you be doing it? (You probably should.)

The Past is History; the Future is a Mystery

Now, I don't really know anything about the 2025 rookie class. I'm not sure if it will be an especially strong or especially weak one (I usually assume all classes will be roughly average until after the combine, at least). The most important feature of it, in my opinion, is that it's two seasons away. (Had this column been written last year, I'd instead advocate trading for 2024 picks. Next year, I'll suggest the 2026 class.)

In his October Trade Value Charts, Dan Hindery wrote the following:

Be very careful using the above Early 2024-1st blindly if valuing a pick in your league. Let's make some early value assumptions about non-Superflex pick values and do some basic math to illustrate the point. Assume the 1.01 ends up being worth 40, the 1.02 is worth 30, the 1.03 is worth 28, and the 1.04 is worth 26. If we assume equal 25% odds of an Early 2024 1st being one of those four picks, the value is 31 (124 divided by four). However, if the pick actually has a 50% chance of being 1.01 and a 50% chance of being later, it is worth more in the range of 35. If it only has a 10% chance of being 1.01 because there is another really terrible team likely to be 1.01, it may only be worth 28. That's a fairly large gap between two picks that would both be considered Early 2024 1st. Know your league rules about how draft order is determined, and try some back-of-the-envelope math to really drill down on projected pick value before making any trades.

This is a very important concept, and Hindery does a fantastic job reinforcing this every month with his trade values (which are one of my favorite features on Footballguys). Last month, he rated an "early 1st" as worth 55% more than a "mid 1st" (31 points vs. 20 points), but a mid 1st as worth just 25% more than a late 1st (20 points vs. 16 points). Picks get substantially more valuable the higher up the board you go. (Managers who just missed out on Bijan Robinson last year are no doubt keenly aware.)

The problem for our purposes is that managers are both aware of this fact and fairly good at estimating their team quality this year. Anyone who thinks their first could be high in 2024 is likely to demand a fortune to part with it. (A smaller but still significant problem when trying to acquire high picks: if you trade veteran players for future picks, those veteran players will increase the quality of the other team and likely push that pick later into the round.)

If managers were as good at estimating their team quality in 2024 as they are at estimating their team quality in 2023, it would be a bad idea to invest so far into the future. Are they? Let's investigate.

How Well Does This Year's Performance Predict Next Year's Performance?

(As a warning: we're about to talk about math. I'll do my best to explain the process to make it easier to follow if you're less mathematically inclined.)

To start, I trawled through the history of my dynasty leagues and logged each team's point per game (or ppg) average in each season. This is a fairly good measure of team quality. (Some might prefer all-play record, but I've found that both points per game and all-play record are essentially interchangeable when predicting future performance.)

Different leagues have different scoring and lineup requirements, so these values are not yet directly comparable. So I normalized these ppg values such that the league average in any given season would result in a score of 100, and every standard deviation above or below that average would add or subtract 15 points (standard deviation is a measure of how "spread apart" the data is). Thus, a team that scored one standard deviation below league average would score an 85, while a team that finished two standard deviations above league average would score a 130.

As an aside: I could have just divided points scored by league average to get a "points over average" value, which would also be directly comparable across leagues. Why did I use standard deviations above or below average, instead? Because it matters how tightly clustered or spread apart the scores are.

Imagine two different ten-team leagues where the average weekly score is 120 points. Now imagine both leagues feature a team that scored 144 points per game. In both cases, this represents 20% more than the league average.

Now imagine that the distribution of scores in the first league is: 88, 96, 104, 112, 120, 120, 128, 136, 144, 152. And imagine the distribution of scores in the second league is: 96, 102, 108, 114, 120, 120, 126, 132, 138, 144. It should be obvious that the team with 144 points is more likely to win a championship in the second league than the first.

The standard deviation of scores in the first hypothetical league is 20.7, so our hypothetical team would have a "normalized score" of 117. The standard deviation in the second league is 15.5, so our hypothetical team would have a "normalized score" of 123. Normalizing this way makes it clear that the second team is more likely to pick 12th than the first. (Similarly, a team averaging 96 points per game would be much more likely to earn the top overall pick in the second league than the first.)

(In case you're curious, this is the same process used to determine IQ scores, so a fantasy season that scores 115 on this scale should be about as rare as an IQ of 115, with both occurring in about 16% of observations.)

And What Did The Data Look Like?

After normalizing every team from every season I'd played, I compared each team's performance in one year with their performance in the next. This enabled me to find a "best-fit" equation, into which I could input a score for this year and get a "best guess" for a score next year. For my leagues, that equation was: Normalized Points in Year 2 = (0.363 * Normalized Points in Year 1) + 63.9.

Let's return to our second hypothetical league above (where league average was 120 points per game and the teams were more clustered together). Our 144ppg team had a normalized score of 123; entering that value into the "best fit" equation suggests that our best guess for their score the next year is 109. If the league's scores next year have a similar distribution, that would translate to about 132 points per game, a 12ppg drop. Or another way of thinking about it: 144 points per game was the best team in the league, while 132 points per game was the 3rd-best.

A team that scores 132 points per game is still quite unlikely to wind up with a high pick, but there's a second thing I can calculate from my data: the correlation between performance one year and performance the next. Correlation is a measure of how closely two values are related, with values ranging from zero (indicating no relationship) to 1 (indicating one season perfectly predicts the next). In this case, the correlation was 0.356, which is quite low.

If you square the correlation you get an estimate of how much of the variation in next year's performance is "explained" by the variation in this year's performance; in this case, that value is .127. In other words, how well a team does this year explains about 12.7% of the variation in how well that same team does next year. This leaves 87.3% of the variation still unexplained. That's a lot of uncertainty.

That uncertainty is then compounded by the randomness of fantasy football-- factors like schedule luck which ensure that occasionally even good teams lose a lot of games and pick early. Even if this is one of the rare instances that our best-fit equation is accurate and the team in question has the 3rd-best point per game total in the league, they still might miss the playoffs and pick in the top half of the 2025 rookie draft. (In my leagues, the team that finished 3rd in points still missed the playoffs 25% of the time thanks to tiebreakers and luck.)

What Is The Important Takeaway Here?

Teams that are good this year are likely to be much less good next year. Additionally, there is a wide range of uncertainty around that projection, meaning they could finish substantially better or worse than expected.

If they perform much better and their pick winds up later in the round, this likely doesn't reduce the value very much because the difference between picks is fairly small toward the bottom of the first round. If they perform much worse than expected and their pick winds up earlier in the round, this dramatically increases the value because the difference between picks is quite large towards the top of the round.

When buying rookie picks in two years from good teams, you're taking a gamble, but that gamble is not symmetric; the upside if you get lucky is significantly greater than the downside if you get unlucky. (For bad teams, this uncertainty cuts in the other direction; their pick is quite likely to be much later than they expect, which means there is much more downside than upside when trading for it today.)

(All of this assumes, of course, that managers whose teams are good this year assume their teams will be roughly equally good next year, which is not a given. It's worth looking into, though.)

As a bonus, if you trade currently productive veterans for 2025 picks, those veterans are less likely to impact those picks than they are with 2024 picks.

2025 is a long way away and there's no guarantee that you'll still be around to enjoy that extra pick you traded for today (especially because many rookies drafted in 2025 won't reach their prime until 2026, 2027, 2028, or beyond). But the sneaky thing about draft picks is their value timeline tends to be a year ahead. 2024 draft picks are hot commodities right now in dynasty; likewise, any 2025 draft picks you acquire today will be popular on the trade market next year, meaning it's only a one-year wait before that extra pick becomes valuable depth for your team.

(Searching for dynasty trades that were completed today, managers have managed to turn 2024 first-round picks into Keenan Allen, Alvin Kamara, Austin Ekeler, Tony Pollard, Michael Pittman, D.J. Moore, James Cook, Tee Higgins, Jaylen Waddle, Justin Herbert, Lamar Jackson, Tua Tagovailoa, and more. These are all real 1-for-1 trades from real leagues on a single day in October. That's a lot of points getting put directly into starting lineups by something generally considered a "future asset".)

If Two Years Out Is Good, Three Years Out Must Be Great, Right?

Assuming, of course, that your league even allows trading picks from 2026 already, it's probably a great idea to start buying those picks early, too, right?

Well... no. Calculating the best fit for performance in one year and performance two years later, we get the following equation: Normalized Points in Year 3 = (0.303 * Normalized Points in Year 1) + 70.3. Using our hypothetical 144ppg team from above, the expectation in Year 2 was 132ppg, a 12-point drop. In Year 3, the expectation would be 128 points per game, a 4-point drop from Year 2.

Additionally, the correlation did not decrease by much, falling from 0.356 to 0.297. If this year's performance explains 12.7% of the variation in Year 2, it explains 8.8% in Year 3.

Sure, the expectation for a good team two years out is slightly lower, and the uncertainty is slightly higher. But the timeline for you to expect a return on your investment is twice as long. (Wait, isn't three 50% bigger than two? Yes, but remember that picks reach maturity on the trade market a year before they come due.)

(Anecdotally, I also find that good teams tend to be quite confident they'll still be good next year, but much less confident that they'll still be good in two years. If this pattern is common, the available discounts might be much smaller, to boot.)

So Should You Trade for Picks Two Years Out?

Maybe! It's not going to be the right move for everyone all the time. But I've written about how some of the most lopsided trades, in hindsight, are when one party takes on short-term pain to realize a long-term gain. And since managers tend to be overconfident one year out, trading for picks two years from now is sometimes one of the best ways to get the maximum amount of long-term gain for the minimum amount of short-term pain.

But every manager and every team and every league is different. Maybe the managers of the best teams in your league are much more cautious than average, or maybe they're especially incautious and have already sold their 2025 picks. In general, though, I think this option should be considered more than it is.