Direct Derek - Alternative Markets

Expiring Country-Code Domains Nobody Monitors Globally

2026-07-28T00:00:00+00:00

A domain expires. The address remains visible from anywhere, but the machinery underneath it turns local immediately.

Which registry controls the lifecycle? Who is eligible to register the name? Does it drop cleanly, pass through an auction, or follow some other release process? Can a foreign buyer hold it directly, or is a local proxy required? A scanner can find the string the moment it goes dark. It cannot tell you who is legally allowed to own it next.

The appealing version of this trade is one system watching forgotten names across the internet and catching the valuable ones as they fall. The actual version involves registry calendars, eligibility rules, local relationships, recurring fees, and a buyer who may never appear.

Global visibility is not global eligibility.

At expiry, the clock that matters belongs to the registry.

One Namespace, 316 Rulebooks</h2>
At the end of 2024, there were 364.3 million registered domains across all top-level domains. Country-code extensions accounted for 140.8 million of them, spread across 316 delegated ccTLDs. The ten largest represented 58.2% of that volume: .cn, .de, .uk, .ru, .nl, .br, .au, .fr, .in, and .eu.
That looks like a large global market until you try to transact in it.
Each registry can set its own eligibility requirements, pricing, lifecycle, and drop rules. Drop-catching services cover major extensions, and heavily trafficked namespaces are watched closely. What remains impractical is comprehensive monitoring across all 316 regimes through one standardized process.
A platform attempting that coverage needs integrations, rule maintenance, compliance procedures, and exception handling for assets that often sell in the hundreds or low thousands of dollars. A local specialist needs to understand one registry deeply enough to know where the automated map is wrong.
The market is digital. The permissions are stubbornly territorial.
A Scanner Can Find a Name It Cannot Own</h2>
Eligibility friction begins after discovery. Canada's .ca has a Canadian-presence requirement. France's .fr requires an eligible European nexus. Germany's .de may require a German administrative contact when the owner is abroad. Other country-code extensions permit broader registration.
The inconsistency is the opportunity and the hazard.
Software can flag an expiring name and record its history. It cannot supply residency, create a qualifying legal nexus, or produce a reliable local counterparty. Nor can it override the registry's transfer and release mechanics.
A local-presence service may bridge the gap, but the correct way to model that arrangement is as a cost and a counterparty, not a box checked during registration. Someone must remain compliant and responsive for as long as the name is held.
The barrier also follows the asset out the other side. A rule that excludes competing buyers during acquisition may exclude potential end users at the point of sale. Restricted eligibility can create scarcity, but it can just as easily create a very private market in which the holder is the only person paying an invoice.
A two-letter suffix can generate more administration than economic value, which is an impressive amount of work for two letters.
Capacity Is Measured in Exceptions</h2>
Sedo's 2024 aftermarket data covered roughly 350 TLDs. The reported median sale price was $549, while the average was about $2,345. Sixty-nine percent of sales closed at a posted Buy-It-Now price.
The gap between the median and average matters. A portfolio can look attractive when modeled around occasional large sales, while most completed transactions occur at values too small to support much legal, administrative, or marketing work.
More capital does not solve this. It merely buys more small problems.
For a platform, capacity may appear to be the number of domains its systems can scan. The number that actually binds is how many jurisdictions it can monitor, validate, and administer without a compliance failure, weighted by the inconvenience of each rulebook. Nominal coverage is easy to advertise. Maintaining lawful control over a scattered portfolio is the expensive part.
For a specialist, useful capacity may be only one or two regimes understood properly: the registry lifecycle, the eligibility test, the available registrars, the likely renewal costs, and the local buyers who actually transact.
Every additional jurisdiction introduces another set of exceptions. Eventually the automated strategy starts hiring people, and the supposed software edge becomes an administrative business with unusually speculative inventory.
Catching the Name Starts the Invoice</h2>
Winning an expiring domain does not create demand. It creates inventory with an annual bill attached.
Industry guidance puts annual sell-through for a well-optimized, fairly priced domain portfolio somewhere around 1% to 1.5%. That is a rough portfolio benchmark rather than ccTLD-specific evidence, and it comes from industry commentary rather than a published marketplace dataset. It still describes the holding problem: in a given year, nearly every name reaches another renewal decision without selling.
The renewal arrives on schedule. The buyer does not.
Before acquiring a name, a patient holder wants a conservative sale value, credible end users with commercial intent, an honest estimate of timing, marketplace commissions, annual renewal costs, any local-presence expense, and the likely transfer friction. Most importantly, set the number of renewals to tolerate before walking away.
That limit has to be fixed before purchase. Once a domain sits in a portfolio, another renewal always feels cheap relative to admitting that the original thesis never contained a buyer. A small annual payment is excellent camouflage for a permanent mistake.
Pricing adds another layer of risk because the registry controls the recurring charge. Renewal costs vary widely by extension, and many non-legacy extensions carry no meaningful price caps at all — one registry operator raised wholesale prices across 231 of its extensions in October 2025, a median increase of roughly 12%, with individual extensions moving far more. In a country-code portfolio, the practical point is simpler: the holder controls neither the rulebook nor the invoice, and bears the full cost of both.
Friction Can Flatter the Hunter</h2>
The whole frame here favors markets that are fragmented, inconvenient, and too small for large capital to price efficiently. That instinct helps right up until procedural difficulty gets mistaken for economic value.
An eligibility barrier may reduce competition for a genuinely useful name. It may also leave an unwanted asset sitting undisturbed. From a distance, those conditions look remarkably similar, and the framework offered here does not reliably tell them apart — the test it proposes is "are there identifiable end users," which is a judgment dressed up as a screen.
So the discipline is narrow. Scarcity of the word, linguistic neatness, and difficulty of foreign acquisition are not an underwriting case. What is required is multiple identifiable end users and a carrying cost low enough to survive years of silence. Position size follows the period of zero liquidity you can tolerate, not the price you hope to post later.
The procedural protection works best inside a jurisdiction someone actually understands. Stretch it across dozens of regimes and the edge turns into overhead. Carry grows with every acquisition, while sales arrive unevenly and without regard for the renewal calendar.
At that point the honest question is whether the local rules protect the value or merely protect the asset from ever being priced honestly.
Where the Edge Survives</h2>
The trade works when local knowledge identifies a genuine expiry, eligibility restrictions reduce competing bids, carrying costs remain tolerable, and several credible end users exist. It requires little acquisition capital but considerable procedural confidence.
It fails when the profitable part becomes standardized, a registry changes its rules, a local relationship breaks, or the buyer pool exists only in a spreadsheet. Expansion is particularly dangerous because every added name creates certain carry while adding only contingent revenue.
Spotting the expiring ccTLD is the easy part. What determines the outcome is lawful access, verified local demand, and the number of renewal invoices you can absorb before conceding that nobody else wants the name.
Until an eligible buyer appears, the annual payment buys nothing but a globally visible word tied to a local clock.

Sports Cards After the Tourists Leave

2026-07-19T00:00:00+00:00

Suppose a graded card sells at auction for $5,000. The result enters a database, and every owner of the same card in the same grade can multiply accordingly. The screen is precise enough to make the arithmetic feel earned.
Then someone tries to sell a second copy.
The first may find a collector who has wanted it for years. The second reaches a dealer who already owns one. By the fifth, the remaining buyers know inventory is coming and adjust their bids accordingly. Nothing has changed about the card, and quite a lot has changed about the market standing around it.
A completed sale proves one buyer existed at one moment. It says nothing about how many copies that buyer wanted, whether the underbidder is still interested, or what either would pay next Tuesday.
The first question about a comp is therefore impolite but useful: how many more copies would that bidder have taken?
The capacity number is smaller than the collection value</h2>
Price guides establish a range. Auction records show where transactions occurred. Neither says how much capital can enter a narrow card market before the buyer becomes its principal source of demand.
For that, less glamorous information is needed:

How many copies sell in an ordinary month?</li>
How many repeat buyers are known?</li>
How much inventory do active dealers already hold?</li>
How long does a sale take without a material price reduction?</li>
What is the bid if the seller needs cash rather than another listing?</li> </ul>
This is where small size earns its keep. A patient buyer can reject ninety-nine cards and take the one whose price compensates for the exit. A fund attempting repeatable deployment cannot be so selective. It eventually has to accept weaker condition, pay thinner discounts, or acquire enough inventory to manufacture the optimistic comps supporting its own marks — a fairly elaborate way to become your own customer.
Size against a stressed dealer bid and a realistic holding period, not the guide value of the collection. If the two most active dealers in a niche stop bidding, the spreadsheet can still show clean marks to the dollar while the cards sit in boxes, waiting for an audience that has already left.
What the departure looked like</h2>
The 2020–21 card market attracted buyers who treated recognizability as liquidity. If the player was famous, the grade was high, and the chart pointed northeast, another bidder was assumed to be nearby.
That assumption became expensive. Card Ladder's CL50 index fell about 23% in 2022 and another 9% in 2023. Its Ultra-Modern index, after rising roughly 639% from 2017 to its February 2021 peak, gave back more than 30% during 2022 alone.
The individual sales were less polite. A 1986–87 Fleer Michael Jordan PSA 10 peaked at roughly $738,000 at Goldin in January 2021; by May 2022 two copies sold at the same house for $344,400 and $288,000. A 2003 Topps Chrome LeBron James refractor PSA 10 sold for around $300,000 in March 2021 and, after a multi-year slide, changed hands in the low $20,000s by 2024. Those are single-sale snapshots from the trade press rather than a rigorous series, but the direction is not in dispute. The card issue and the grade were unchanged; the room simply had fewer buyers in it.
Those numbers describe a handoff between different kinds of capital. The speculative buyer pays for momentum, familiarity, and the expectation that another bidder will appear on cue. A dealer works backward from a realistic resale price, then accounts for holding time, transaction costs, and the possibility that no customer asks for this card next month. The dealer also knows what is already sitting unsold in the case.
That does not make every dealer bid correct or every post-boom decline a bargain. Sometimes the lower bid is simply the first honest price the card has received in years.
A slab is only semi-fungible</h2>
Grading performs a useful conversion. An idiosyncratic piece of cardboard receives a numeric condition grade, a tamper-evident holder, and a place in a population report. Buyers can search it, compare it, and transact without beginning condition analysis from zero.
That is real infrastructure. It is also easy to ask too much of it.
A grade remains a subjective human judgment. Two cards carrying the same number can differ in centering, print quality, eye appeal, and documented history. During a boom, buyers tend to ignore those distinctions because they want exposure to the label. When demand thins, the distinctions return to the table and introduce themselves to the bid.
Illustratively, modern cards can trade at PSA 10 prices somewhere around two to five times their PSA 9 equivalents. For scarce vintage, where top-grade populations may be tiny, the spread can reach ten to twenty times. Those are rough card-specific ranges rather than laws, and any given pair should be checked against actual recent sales, but they show how much money can rest on the distance between near-perfect and almost near-perfect.
What the slab standardizes is the label, not the buyer. Put five copies with the same grade on a dealer's table and the offers can diverge as soon as someone examines the actual cards.
The grading factory keeps running</h2>
Falling prices did not stop the industrialization of condition. Major grading companies processed roughly 26.8 million cards in 2025, about 32% more than in 2024. PSA alone handled around 19.26 million, close to 72% of the graded market.
That matters because graded scarcity can change after purchase. When another PSA 10 appears in the population report, the card in the box is physically unchanged, but its claim on rarity weakens. "Rare in grade" can describe a durable condition bottleneck. It can also describe a submission queue that has not finished moving.
The submission decision is an expected-value calculation:
p10 × V10 + p9 × V9 + pother × Vother − C > raw value</code>
The arithmetic is simple enough to do on the back of a toploader; the inputs are the part nobody can pin down.
Following PSA's September 2025 update, lower-tier prices ran from roughly $21.99 to $44.99 per card. As of PSA's May 2026 schedule, published turnaround estimates ran from 5–7 business days for Walk-Through to 140–160 business days for Value Bulk. Those figures change frequently and need checking before submission rather than after.
At bulk economics, grading, shipping, time, and tied-up capital can put the effective cost somewhere around $30 to $50 per card. Cheap raw cards often cannot clear that hurdle, which is why the low end of the market stays raw, thin, and information-poor. More valuable cards clear it only if the submitter estimates condition better than the seller did and the eventual grader agrees. Meanwhile, new supply can keep arriving from the queue long after the buyers who paid boom prices have gone quiet.
The dealer bid carries hidden data</h2> A dealer offer contains information a price guide cannot print: inventory already on the shelves, collectors on the call list, realistic holding periods, and how often supposed buyers disappear when asked to pay. Dealer networks function as informal market makers, except nobody obligates them to keep making a market. Their knowledge lives in conversations, prior trades, and memory rather than a centralized order book. On a convention floor, cash and reputation can settle a transaction quickly because the participants have already evaluated each other. A stranger may bring the same money and still lack the same access. That relationship gate frustrates scalable capital, which is precisely why it deserves respect. Authentication, counterparty trust, and specialist knowledge do not expand neatly with assets under management. A large buyer cannot automate eye appeal, manufacture repeat collectors, or diversify away a market that gets thinner as its inventory grows. And the failure mode here is not a markdown. A missed alteration or a counterfeit is a near-total loss, which is a different kind of arithmetic than being early. The remaining edge is correspondingly small. One card from an impatient seller may fit inside it; a full collection may overwhelm it. Buying cheaply from a departing tourist means inheriting the thin exit that helped make them a seller. The discount is payment for waiting, assuming it is large enough to cover the wait. Size to the second bid</h2> Better grading, cleaner databases, and deeper auction histories improve the map. They cannot populate the market with buyers. The method has an obvious blind spot, and it is worth naming. A buyer can be right about the player, the card, the grade, and the long-term demand while still being wrong about how much inventory the market can absorb — and the only honest test of absorption is the one you run by selling, which is the test nobody wants to run first. Buying one attractive card below a stale comp may work. Buying enough copies that the thesis requires the tourists to return is inventory financing with a hobby attached. So leave the price guide open, but ask a dealer for an executable cash bid. Then ask whether they would take a second copy at the same price. That answer determines the size.
Authentication as the Actual Collectible-Market Edge 2026-06-10T00:00:00+00:00 The reference number matches. The serial scan comes back clean. The watch has sapphire, convincing steel, a sweeping seconds hand and an engraved rehaut. Everything visible at arm's length agrees with the catalog. Open the case and the problem changes. A franken-watch can combine genuine components with counterfeit or replacement parts, validating one test while failing the next. The database has priced a category. Somebody still has to decide whether the object belongs in it. That distinction carries most of the friction in collectible markets. Rejecting a counterfeit produces no visible return. There is no gain to record and no sale to celebrate. Accepting one can produce something close to a total loss, because an altered piece cannot be passed to the next buyer without handing them the same problem you failed to catch. Capital helps you buy more inventory. It does not give you more competent inspections. What, exactly, have you agreed to own? What the database already believes</h2> Swiss Customs and the Federation of the Swiss Watch Industry have long put annual fake-watch production between 30 and 40 million units — figures repeated without revision for over a decade. Rolex publishes nothing, but Morgan Stanley and LuxeConsult estimate genuine production at roughly 1.2 million watches a year. Those estimates are not an audited census, but the order of magnitude explains why casual inspection is no longer a serious method. High-grade super clones sell to buyers for roughly $450 to more than $1,000 — a fraction of the genuine article, and by most accounts a large multiple of what they cost to produce. They are designed to defeat the checks an ordinary buyer knows to perform: sapphire, 904L-like steel, smooth-looking seconds, engraved details and convincing paperwork. Frankens are more difficult because some of the evidence is genuine. A serial may validate while the movement or another component tells a different story under a loupe. Verification therefore happens in layers. Construction, movement, originality, service history and provenance each narrow the range of plausible stories. Box, papers, receipts and prior ownership help, but documents do not become independent evidence merely because the seller arranged them neatly. Counterfeiters understand that buyers find barcodes soothing. Under magnification, the sales pitch depreciates one component at a time. The slab industrializes judgment</h2> Sports cards handle a related problem by packaging authentication and condition into infrastructure. In 2025, the major graders processed roughly 26.8 million cards, about 32% more than in 2024. PSA alone graded approximately 19.26 million, representing about 72% of the market. A slab converts an idiosyncratic physical object into something closer to a standardized unit. The market can search it, compare it and quote it. What trades is no longer merely a card; it is a card attached to an authentication decision, a condition opinion and a population count. The opinion remains subjective. A PSA 10 requires near-perfect centering, corners, edges and surface, while a 9 permits a minor flaw. The boundary is still drawn by a human grader. Crack-outs and resubmissions exist because the same card can receive a different decision on another attempt. That uncertainty has a price. On modern cards, a PSA 10 can trade around two to five times the value of a PSA 9. For scarce vintage cards, reported spreads can reach ten to twenty times because top-grade examples barely exist. Those are illustrative ranges, not a universal schedule. The exact multiple depends on the card, its population and the buyers present when you need to sell. The submission decision reduces to an inequality: p10·V10 + p9·V9 + pother·Vother − C > R</code> The probability-weighted graded value, less grading costs, must exceed the raw value. Lower PSA service tiers ran roughly $25 to $50 per card after the February 2026 pricing changes, before shipping and the value of time. Estimated turnaround runs from roughly 25 business days on faster tiers to 150–170 business days for bulk — PSA extended it twice during 2026 while raising the bulk minimum from 20 cards to 50 — and the published tier schedule changes often enough that any figure should be checked at submission rather than remembered. That keeps much of the low end raw. A card worth only a few dollars cannot absorb $30 or $50 of all-in cost and months of illiquidity. Higher-value cards may clear the threshold, but submitting one remains a wager on somebody else's condition call. Even after receiving a 10, the population report can add more 10s and compress the premium without anything changing about the card in your hand. The counterparty is part of the object</h2> Every transaction asks two questions. Is the object genuine, original and accurately described? If the answer later changes, who bears the loss? The second question makes reputation economically useful. What matters is whether a seller discloses a flaw before you find it, whether prior deals remain defensible after inspection, and what happens when uncertainty appears. A sincere seller can unknowingly pass along a fake. A dishonest one can mix authentic and counterfeit components while leaving the serial untouched. A trusted counterparty lowers search and verification costs. Dependence on one trusted counterparty creates a different problem, which is concentration risk with pleasant manners. This is why rare vintage watches often circulate through dealer and collector networks before reaching a public listing. First-right-of-refusal arrangements and private relationships can move a piece several times while databases see nothing. By the time an auction result appears, the public record may be describing the end of a chain rather than the market's first encounter with the watch. Convention floors operate on similar mechanics. Deals can settle quickly with cash and reputation, while the useful record of who stands behind an object remains off-platform. Access is earned through repeated transactions. The gate is not a subscription fee; it is the willingness of knowledgeable people to trade with you again. Capacity counted in inspections</h2> Collectibles look friendly to capital because individual unit sizes can be small. Capacity here is better counted a different way: how many objects can be inspected competently, how many counterparties have earned trust, and how much inventory the eventual buyer pool can absorb. Institutions struggle with that structure. Supply arrives irregularly, each piece requires separate diligence, and much of the best inventory never lists publicly. Delegating inspection introduces agency risk. Building specialist knowledge in-house requires enough transaction volume to justify the overhead, but forcing volume is exactly how judgment deteriorates. The attempts to financialize the problem have not solved it either; fractional card platforms did worse than the underlying market — they bought at the 2021 peak, sit well underwater, and the promised secondary market for shares never produced buyers. Several sponsors went bankrupt or pivoted, leaving holders with neither liquidity nor the object. A small operator can reject almost everything without needing the next object to satisfy a deployment target. Specialization can be narrow enough to become useful: one maker, one period or one card set. The position can also remain small enough for the same network that helps validate the purchase to absorb the eventual exit. That network is protective until the holding outgrows it. You may end up selling back to the dealer who helped authenticate the purchase, creating a closed loop where the apparent market consists of a few familiar people passing inventory among themselves. The edge disappears when holdings exceed personal re-underwriting capacity, when the discount falls below the real cost of verification, or when inspection gets outsourced to maintain pace. Opacity is not automatically mispricing. Sometimes the market is difficult because the object deserves to be difficult, and a framework built on loss-avoidance has an obvious flattering property: it never has to produce a trade to feel vindicated. Refusing everything is indistinguishable from good judgment right up until the moment you notice you own nothing. Size accordingly. An independent buyer wants records that survive the next buyer's skepticism, a larger discount when provenance is incomplete, and enough time to stop when a seller introduces urgency. The question worth answering before buying is where the next competent operator will stand when it is time to sell, and what evidence that person will refuse to take on faith. Authentication does not remove market risk. It identifies which market risk you actually own. The best mistake in collectibles is the object that never enters the portfolio.