How Much Can You Dump on the Bazaar Before the Price Drops?

A Commerce Journey — Episode 16. Real economics, measured against live Hypixel Skyblock Bazaar data.

The production possibility frontier is the first diagram in most economics courses. Two goods on two axes, a curve showing every combination you could produce, and the whole point of the picture is that the curve bends.

The bend carries the lesson. A straight line would mean giving up the same amount of one good for each extra unit of the other, forever. A curve means the trade gets worse the further you push, which is called increasing opportunity cost and is treated as a fact about the world.

Skyblock lets you draw a real one. When I did, it came out mostly straight, and working out why took me somewhere more interesting than the diagram.

1️⃣ Why the textbook curve bends

The standard explanation is about resources being unequally suited to different jobs. A country moving from making butter to making guns first shifts the workers and land that were bad at dairy anyway, so the early trade is cheap. Push further and you start converting good dairy land into weapons factories, and each additional gun costs more butter than the last.

That story is about production technology. It says nothing about markets, and it is where the diagram gets its shape.

It is worth noticing how strong an assumption that is. The curve is drawn before any buyer appears. Whoever produces the guns is assumed to be able to sell all of them at the going rate, and the same for the butter, so the only thing that can make the trade worse is the difficulty of making the next unit. Every introductory course teaches the shape and almost none of them mention that a market has to exist underneath it.

2️⃣ Building one from real data

A Skyblock player faces the same problem in a cleaner form. Time is the constraint, two items are the outputs, and the question is what you give up.

The complication, and the reason this is worth measuring rather than assuming, is that producing goods is only half of it. You have to sell them. Episode 11 established that a large sale walks down the buy-order book, taking a slightly worse price at each level, so the coins you get for your hundred thousandth unit are not the coins you got for your first.

That is a source of increasing opportunity cost that has nothing to do with production and everything to do with who is standing on the other side. So I measured it directly across 743 items with real weekly volume.

Enchanted diamond is the flat line. Selling a thousand units gets you 100.00% of the top buy order. Selling a hundred thousand gets you 99.88%. The book holds 19,647,089 units of standing demand, so a hundred thousand barely registers.

Enchanted cobblestone is the one that curves. At a thousand units you keep 99.98%, at fifty thousand 97.55%, and at a hundred thousand 94.33%. Its book holds 3,890,225 units, so a serious harvest is a real fraction of the whole market.

Those are the two shapes, and the difference between them is not how they are made. It is how many people want them.

3️⃣ Most frontiers are straight

Three items is an illustration. Here is the population.

Selling 50,000 units at once, the median item gives up 0.05% of its top price. That is not a bend. That is a straight line with a rounding error on it.

At the ninetieth percentile the cost is 16.4%, which is a real curve. At the ninety-ninth it is 93.2%, which is a cliff. The worst item in the set gives up 99.7%.

So the textbook shape is not wrong here, it is rare. For something like nine items in ten, at volumes a serious player actually produces, the frontier is effectively linear and the opportunity cost of shifting production is constant.

4️⃣ Where it does bend, it snaps

Shard Pangolin gives up 99.7% selling 50,000 units. You would receive, on average, under a percent of what the screen quotes as its price.

These are not gentle curves. Nothing in this tail resembles the smooth arc in the textbook diagram. The books are shallow enough that you exhaust genuine demand in the first few thousand units and spend the remainder selling into lowball orders that have been sitting there for days waiting for exactly this.

Notice too that only 46% of items can absorb a 50,000 unit sale at all. For the rest there is no frontier to draw, because the axis runs out before you reach the quantity.

The same pattern holds at every size I tested. At 10,000 units the median item gives up 0.01% and the ninetieth percentile 10.0%. At 100,000 the median is 0.29% and the ninetieth 14.0%. The middle of the distribution stays flat as you scale up, while the tail gets worse, so the gap between an ordinary item and a thin one widens with every extra unit you produce.

Key characteristics of a frontier that will bend on you:

• Weekly volume smaller than what you can produce in a session
• A buy-order book whose total is a modest multiple of your batch size
• Most standing demand concentrated in one or two orders near the top
• A price that looks perfectly normal on screen, because depth is never shown

5️⃣ What the bend costs in coins

Percentages under-sell this, so take a real run. You have farmed 100,000 enchanted cobblestone and you want it gone today.

The top buy order is 472.00 coins, so the screen tells you the pile is worth 47,200,000. Walk the book and you actually receive 44,521,424, an average of 445.21 per unit. The bend has cost you 2,678,576 coins, and none of it was visible before you pressed the button.

Split the same pile into four sales of 25,000 across the week and the book has time to refill between them, which is the standard fix and it works. It also means holding inventory for days, which Episode 13 priced as capital sitting still. There is no free version of this, only a choice about which cost to pay.

Run the same arithmetic on enchanted diamond and the loss is 149,328 coins on a pile worth 121,170,000. On that item the question does not arise.

6️⃣ A straight frontier changes the right answer

There is a consequence of linearity that took me a moment to see, and it is more useful than the shape itself.

When a frontier curves, the best plan is usually a mixture. The trade gets worse as you push, so you stop pushing partway and produce some of both goods. That is why the textbook diagram leads to balanced production.

When a frontier is straight, the best plan is a corner. If every unit of item A costs the same amount of item B forever, you should produce whichever one pays more and produce nothing else, because there is no point at which the trade turns against you.

Since most Bazaar frontiers are effectively straight at realistic volumes, the correct strategy for most items is total specialisation, not a balanced island. That lines up with what good players actually do, and it is a conclusion the standard diagram would have argued against.

7️⃣ The reason the diagram is right for the wrong reason

Here is what I took from this, and it changed how I read the original diagram.

In Skyblock, production genuinely is linear. A minion produces at a constant rate whether it is the first hour or the hundredth. Your farming speed does not decay because you switched crops. There is no unequally-suited land, no retraining, none of the machinery the textbook uses to justify the curve.

And yet the curve appears anyway, on some items, sharply. It arrives entirely through the market. The increasing cost is not the cost of making the next unit, it is the cost of finding somebody to take it.

That distinction matters because the two have different fixes. If the bend came from production you would solve it with better tools. Since it comes from demand, better tools make it worse, because they let you produce further down a curve that was already falling.

8️⃣ Your own frontier

The practical version is a single question to ask before scaling anything up. Not how much can I make, but how much can I sell before the price I am counting on stops being the price I get.

For enchanted diamond that number is enormous and you can ignore it. For enchanted cobblestone it is real and you should size your batches against it. For the tail on that last chart, the honest answer is that you should not build a production line at all, because the market cannot take the output at any speed.

Check the total of the buy-order book, compare it to a day of your production, and if the second number is a meaningful share of the first, your frontier bends and you should plan for it.

Real-world parallel: Paul Samuelson put guns and butter into his 1948 textbook Economics and fixed the shape of this diagram for every course since, with the curvature explained by resources being specialised. The market version shows up somewhere else entirely: agricultural commodities, where a large harvest arriving at once pushes the price down against the farmers selling it, so a bigger crop can earn less money than a smaller one. Economists call that inelastic demand, and it is the same effect as a thin buy-order book. The frontier bends because of the buyers, not the soil.

9️⃣ What I could not measure

I have measured the selling half of the frontier and assumed the producing half is linear. That assumption is fair for minions and rough for manual farming, where you slow down as you get bored, and I have no way to measure the second thing.

The visible book is also only the top thirty levels a side. Whole-book totals come from a separate field and are reliable, but the price I walk to is computed from the levels I can see, so an item recorded as unable to absorb a large sale may manage it further down. That makes the bend figures a ceiling on the damage rather than a measurement of it.

This is also one snapshot. A buy-order book that looks thin now may have been restocked by tomorrow, and the tail items in particular move around a lot precisely because they are thin. What holds is the distribution: most books are deep enough to ignore, a few are not, and nothing on screen distinguishes them.

Before your next big production run, add up the amount column on the buy side for whatever you are making. If a day of output is more than a tenth of it, you are on the curved part of your own frontier and the price you planned around is optimistic.

Coming next: Sunk cost. The minion you already paid for, why it should not affect what you do next, and how much that mistake costs in practice.

Order books from the Hypixel Bazaar API on 2026-09-05, covering 743 items with meaningful weekly volume on the sell side. Prices given up are measured against the top buy order at the moment of the snapshot.