Asset Valuation Pt. 4: Messing with the Draft

After a detour writing a couple articles with cheeky little bits and also a much more extended effort to develop a model for dynasty rankings, I wanted to return more explicitly to my bread and butter: the Asset Valuation series. After my first teaser article, my next three in this series showed how scoring settings can affect player value, and how you can exploit knowledge about these dynamics for your gain.

Those analyses, though, held one thing constant: the value of draft picks. To get those points across, I assumed a standard 12-team redraft setting, and flexed the value of players based on scoring scheme. Here, I’m doing the opposite: I’m messing with the timeline draft pool while keeping scoring scheme constant to show you how the value of a draft pick itself can change based on league size, keepers, and whether you use projections or actual realized dollar value.

League Size

This is something I think we all know intuitively: as league sizes increase, talent becomes scarcer. In-season, this is reflected in the free agent pool. In a 20-team league, where nearly 500 players might need to be rostered, the pickings are slim. Consider the players surrounding 500 on the WAR leaderboard in 2026:

I, uh, haven’t heard of a few of these guys. Maybe that’s on me, but I would think that I could be forgiven for not knowing who Eric Orze is going into this exercise.

Compare this to a 10-team league, which would roster closer to 250 players:

The names speak for themselves. These aren’t players having world-beating seasons, but they’re all making meaningful contributions to their real-life teams and probably did so on your fantasy squads, if only intermittently, as well.

Where understanding league size also comes into play is in the value of draft picks. Accordingly, as league size increases, so does the dollar value of a specific pick:

These dollar values are created based on historical Steamer projections from 2015 to 2026, calibrated based on escalating league size. For each size, I fit a spline model to the full history of data based on the projected dollar value for a given draft slot. I used a standard 5×5 roto, 23-man roster setting.

Mathematically, the reason that picks in a deeper league have a higher dollar value is because there is a higher total budget of dollars spent ($260 x n teams) on the same pool of players. All that money has to go somewhere. Philosophically, though, it makes sense too. In deep leagues, even picks far down in the draft are worth more relative to the same place in shallower leagues because acquiring above-replacement players matters much more when replacement level is so low.

Translated to round-by-round value, the picture looks a bit different:

Here, we see a few patterns: one, shallower leagues have a more linear decline as rounds increase. In other words, as league size increases, the slope of the drop gets bendier. In addition, the typical value of a pick in early rounds of deeper leagues is more valuable. By the sixth round or so, these lines converge then flip directionally. This may seem unintuitive at first – the sixth round of an 8-team league is within the top 50 players off the board while the sixth round of a 16-team league is the back end of the top 100 – but remember the earlier graph as well. The typical dollar value of the 50^th pick in an 8-team league is about the same as the value of the 100^th pick in a 16-teamer:

The above table would be a useful cross reference if you are in multiple different leagues and need help calibrating how to understand pick values in trade offers. Maybe you’re used to 12-team leagues, but joined a 16-teamer last year and someone offered you a player for your 2026 4^th-round pick. It would have been helpful to understand that that pick is about as valuable as a typical third-rounder in leagues you’re used to playing in. So whatever mental or emotional weight you held toward Francisco Lindor, Ketel Marte, or Christopher Sanchez going into 2026, you would have needed to similarly apply to Logan Gilbert, Brice Turang, or Max Fried.

Keepers:

Here is where things start to get a bit hairy. In order to properly account for how keepers effect the value of your draft pool, you need to remove kept players from the draft pool, remove spent picks from the draft, map the remaining picks back onto the ranking pool, and then calibrate your pick value using your preferred methodology.

In the exercise below, I use a 12-team baseline and have assumed no special pick cost associated with keeper; in other words, they are simply kept “for free” and mechanically would be assigned to the last k*n number of draft picks (where k equals the number of players kept and n represents the number of teams, 12 in this case). This matches how most default keeper settings (e.g. Yahoo!) work.

To simulate a pool of kept players, I wrote an R function to dig through each year’s projected player pool and grab k * n players that combine to equal between 90 and 95 of the projected value of the first k * n players. I restricted the eligible pool to not have ADPs below (k+2)*n. In other words, to use a keep-4 example, the function couldn’t look below the sixth round to grab 48 players whose combined value is between 90 and 95% of the value of the top 48.

Now, you might ask “this is overly complicated, why don’t you just exclude the first k * n players and call it a day?”

And to that I’d say “first, who uses the term ‘k * n’ in a sentence?” (Me, apparently). But second, and more seriously, I think this setup more accurately mirrors how keeper pools end up in real life. Rarely will a keeper pool reflect a perfect representation of the best k*n players; usually at least one team will have more worthy players than they’re able to keep, and other teams will have fewer than would meet that bar. I designed this set of rules to mimic the notion that some really good players might get cast back into the pond, while some other less than ideal fish might be brought home.

Once I had those players, I removed them from the available pool of “draftable” players for each year, recalibrated the order of the ranks to reflect this latest pool (meaning, for example, if my top ranked remaining players were the original 14th, 25th, and 32^nd ranked players by ADP, they are now one, two, and three), and recalculated the expected value of each pick based on a spline regression.

The implications here are obvious: the more players are kept, the less valuable your draft picks will be across the board. I call this pick inflation and if (when) I refer to this in subsequent articles, this is what I mean. This chart also provides a handy bar to clear for evaluating potential keepers – if they project to be more valuable than a first-round pick, it makes sense to keep them. If not, then it doesn’t, because either they (or a similarly ranked player) will be available for you to choose.

It’s conceptually straightforward (if intricate practically) to customize a framework like this to your own league. If you don’t know who your league’s keepers are going to be, you will need to make educated guesses. If your cost mechanism is more complex, you will need to account for that. For example, if you have an escalating pick cost each year (e.g. a player kept in round 16 in one year escalates and is eligible to be kept in round nine in the following year) you will need to make an assumption about whether the player projects to be worth a pick in that range. Comparing their projected value against the typical value of a redraft pick in that round will get you most of the way there, but edge cases might require you to know the tendencies of your league mates and also account for the fact that picks will be less valuable than they would in a redraft. Once you have this information, you just need to repeat the process I described above.

Projections vs. Actuals

Before diving into a practical application for putting all of this together, I wanted to touch briefly on the difference of pick values when derived from actual production in a given season versus projected production. Values based on actual production are lower across the board. Why?

In short: injuries and surprise performers. Each of these calculations uses the same auction budget spread out over different pools of players. Projections don’t know very well who’s going to get hurt or which poorly projected players are going to break out. As a result, more value will be assigned to higher-drafted players who end up hurt and less to players who aren’t on draft radars. This consolidates value spread out from auction budgets into a narrower player pool.

Conceptually, there’s nothing wrong with either – they’re simply measuring different things – but which should you use when valuing draft picks? Ultimately, it’s up to your values as a fantasy player. If you want to down weight how you value picks compared to expected player production, then you should trend towards a pick value framework based on actual production because the values are lower.

I personally think that using projected values for picks is more apples-to-apples with the information available to drafters in a given preseason – we also don’t know very well who will be injured or break out – and therefore represent the most reasonable set of assumptions based on the information we have. It’s also more conservative because being a bit stingier with your picks is going to help you avoid mortgaging the farm when it might otherwise be more tempting.

Putting it all together

In my very first article, I made this promise:

In draft leagues, to properly value a player for your league context, you need to marry league rules, league size, the mechanism (or lack thereof) to retain players season-to-season, and what that cost is. With this information in hand, you can create dollar values for draft picks, which can be used apples-to-apples when considering trades involving both players and picks. If you introduce the ability to trade draft picks across multiple seasons, yet more layers are added.

This is what we’ve been building toward over the last three-odd months, and what you are hopefully equipped to start doing now. Let’s sketch out a couple of hypothetical trades for Yordan Alvarez based on a few settings tweaks. We’ll use his rest-of-season ATC projection from the Auction Calculator and pretend that we’re in June with about half of the fantasy season left.

• Scenario 1: 16-team redraft
• Scenario 2: 12-team keep 8

In Scenario 1, you know that you won’t be keeping him in future years, and he projects to be worth $42 rest-of-season. Let’s say you’re trying to buy to bulk up for a playoff run. Since half (wink) of the season is left, you are essentially buying half of that production relative to what a draft pick would be, given that you would theoretically have the player you chose for an entire season. Based on this, the value we need to make up in pick cost is $21. Looking up at the chart above, we can see that a fifth-round pick is typically worth that same $21, so that would be a pretty fair offer. If I were rostering Alvarez, I might be discounting the value of future picks relative to in-season production – a topic I discussed in my initial dynasty ranking explainer – so maybe I’d ask for a fourth-rounder instead. If the acquiring party wanted to avoid sending such high value picks, she might instead offer an eighth- and ninth-rounder since they combine to $23. Again, that’s a fair offer, but usually if I’m trading a 10-dollar bill, I might want more than two fives plus interest in return, especially if I’m discounting the value of future picks somewhat.

It will be much tougher to strike a trade in Scenario 2. For one, the value of picks is substantially lower across the board. For another, since it’s a keeper league, you aren’t just buying production for this season, but also for every subsequent year in which you want to hang onto Alvarez. Let’s say you want to factor this season plus three more as what you want to value in the acquisition. Some quick back-of-the-envelope calculation based on my aging curve research indicates that a player loses about $11-12 of value between his age-30 and 32 seasons, relative to his age-29 year. Injuries notwithstanding, if we think Alvarez is a $42 player now, we might expect him to be a $39 player in 2027, a $35 player in 2028, and a $31 player in 2029. If you discount 20% like I did in my dynasty rankings, that equals $33, $24, and $18 of value you’re purchasing in each year. Add those up with the $21 you’re getting for his current half-season and you will need to pony up $96 of value to acquire him. That exceeds the entire expected value of a typical draft in this setting. There is probably no way it would be feasible without sending actual MLB talent back in return (just like in real life!).

Conclusion

While I acknowledge that the specific numbers I’ve provided are only directly applicable in leagues whose settings match them, directionally, these concepts will be the same in any league. Over time, I hope to publish reference material that is a bit more varied in scope but for now, you can make little adjustments to get most of the way there. For example, if you have a keep-nine league, you can probably take the midpoint values between keep eight and keep 10. If you have a keeper league with a different number of teams, you can probably apply slight scaling factors based on the ratio of the values in redraft rounds articulated above. This lines up with the, ahem, value I have preached throughout these analyses: don’t let the perfect get in the way of the good. If you’re applying these frameworks to how you think about fantasy transactions, you’re already miles ahead of someone rolling off vibes.