Collector Crypt CEO: The real barrier to a digital card platform lies in these four stages.

Written by: Thomas Holmberg, CEO of Collector Crypt

Translated by: AididiaoJP, Foresight News

This article is highly technical. Its goal is to clarify two things: what flexibility digital card pack platforms have when building packs, and what true transparency actually means. The discussion assumes the common multi-tier model: probability rates are published per tier, and within each tier, every card has an equal chance of being drawn.

Thomas Holmberg, CEO of Collector Crypt, breaks this down into four pillars. These pillars cannot substitute for one another; if any single one fails, the entire business collapses.

The First Element: Complete Inventory List

In many countries that allow card-pack products, regulations require platforms to disclose all possible cards inside the packs. Only showcasing a "Top 10" without publishing a complete list of common cards allows platforms to hide their commission margins within the value of those commons. Without a complete card database, there can be no talk of transparency; merely posting images of the Top 10 does not count as transparent.

Collector Crypt provides a public API from which anyone can fetch the complete card list. Thomas believes this should become an industry standard and a mandatory requirement.

The Second Element: Verifiable Randomness

This is where most platforms tend to fail. Some intentionally abuse it, while others simply do not realize how critical it is. To achieve full transparency, a Verifiable Random Function (VRF) is a hard requirement; operationally, proof must be submitted on-chain so anyone can verify it.

This involves several layers of requirements.

First, all possible assets must be publicly disclosed before a pack is opened.

Second, the seed input for the random number must come from outside the platform—for example, transactions constructed by user wallets, RPCs, and Solana—rather than from the platform's own servers.

Third, the published result must be mathematically traceable all the way back to that input and the key that generated it. This tracing process is the proof.

Some platforms claim, "We use VRF to pick packs on our server," or say, "Our server-side random function has been audited." Neither captures the point. That is not proof; it is merely a verbal promise. While server-side systems can certainly produce "provable randomness," the public cannot verify whether you skipped a top prize. Server-side randomness can always make a result appear random, but it cannot prove that the random number was generated on the first try rather than a second or third. A server could easily run a few lines of code like: if this card is drawn, jump to the next random number for display. No one audits these few lines, and they can be added after the audit.

Collector Crypt’s approach is different: the result is determined before you even see the pack opening. Their VRF public key is registered and frozen on Solana, meaning it cannot be rotated, replaced, or retroactively "updated," and it is publicly visible. When users sign a transaction, that transaction itself serves as the input, built by wallets, RPCs, and Solana, not by the platform. Combined with a frozen public key, this non-negotiable input yields only one valid output, leaving no room to "jump to the second number."

The platform runs RFC 9381 ECVRF, the standard widely adopted in the crypto industry. When a pack is opened, the proof and the hash commitment of the user input are submitted on-chain and published. Anyone in the world can take this published proof, the on-chain commitment, and the frozen public key to independently re-verify without ever logging into Collector Crypt.

If the card a user receives does not match the mathematically calculated result, platform cheating can be caught. Users do not need to calculate the random numbers themselves; they only need to verify whether the platform cheated. This is the core point. The platform’s role is strictly limited to reading results and issuing cards. Validators, program addresses, and open-source code are all hosted at vrf.collectorcrypt.com, so this is not an empty claim.

Conversely, if your VRF scheme relies on data stored on the platform's server, and that data was not submitted on-chain before pack opening, then you do not actually have VRF—you just have a phrase: "trust me." "Trust me" works for pub bets, but it is entirely unsuitable for card packs paid for with real money.

Collector Crypt opens this entire system to anyone on Solana as public infrastructure. The code can be forked and deployed independently. Any project needing randomness can use it for free, paying only minimal additional gas fees—a cost two orders of magnitude lower than existing alternatives. If a card pack platform does not use such an open, verifiable VRF, why wouldn't it? There is no reason not to.

The Third Element: Expected Value

This is the most difficult and ambiguous pillar. Collector Crypt also occasionally makes mistakes here and is continuously improving. Perfection is impossible; the key lies in establishing mechanisms for continuous improvement and defining how errors will be handled when they occur.

How much is a card actually worth? When platforms calculate and display the Expected Value (EV) of a pack, they must make assumptions about the value of every card inside. Collector Crypt has been collecting data from eBay, alt.xyz, and Fanatics for 14 months and has purchased a database traceable back years. Even for high-circulation cards traded multiple times a week, the highest and lowest closing prices for the same month, same card, and same grade can differ by 20%.

Some price differences are easy to explain: a seller based in Croatia with only one review auctions in USD while listing a $45 shipping fee. Others happen because buyers sitting in bars drinking beer during double overtime forget to bid. Still others stem from niche Instagram circles: someone holding 100 PSA 10 Bubble Mews raises bids on eBay for three copies, hoping to offload the five they listed at a premium on Fanatics. All sorts of edge cases exist. Ultimately, "market price" is less a precise figure and more a sentiment.

This topic deserves its own article because "subtle violence" hides here. How a platform prices the cards going into packs determines the pack’s EV, dictates how many cards users will keep versus sell back, and ultimately decides the platform's commission margin.

Imagine a platform called HotRips. HotRips claims to have a buyer network, but in reality, it’s just a guy in a garage setting prices based on alt.xyz. He goes to trade shows and tells dealers: "I’ll buy $100,000-worth of cards at 90% of the alt.xyz price." Dealers now know their baseline valuation and exactly where the arbitrage opportunity lies.

For every card brought to the booth, dealers pull up alt.xyz to compare recent closing prices against the platform’s quote. Say someone brings a card: alt.xyz lists it at $4,000, with a recent closing price of $3,500. The dealer knows they can sell it to HotRips for $3,600. But to the seller, they say, "Our best fair-price offer right now is 90%." They show the seller the $3,500 recent close, buy it for $3,150, flip it to HotRips for $3,600, and pocket a $450 profit. Both sides hear "90% off," but they are referencing different price tags, allowing the dealer to pocket the difference effortlessly. HotRips thinks they’ve built a buyer network, but they’ve actually just erected a vending machine that exclusively benefits dealers.

This vending machine only accepts one currency: overvalued cards. If alt.xyz quotes too high, dealers sell to HotRips; if it quotes too low, dealers won’t sell, instead listing on eBay, keeping them, or selling to someone nearby who buys at actual market prices. HotRips never sees those undervalued cards. Consequently, the $100,000 inventory delivered to HotRips’ door is not a market sample, but a filtered sample of cards priced inaccurately in favor of dealers. The data is already biased before HotRips even opens a spreadsheet. Averaged across the entire pool, this machine structurally overvalues its intake. No one is cheating; it’s simply that the filter at the door only lets overvalued cards through.

HotRips takes a card bought for $3,600 and places it into digital packs valued at $4,000 per alt.xyz listings. Not every card deviates this much, but mixed across the whole pool, the advertised $55 EV of a pack might actually be worth only $50 at true market prices. HotRips isn’t even trying to cheat; it’s just blindly following external price tags.

Add user choice into the mix. When a user pulls that "$4,000" card and the platform offers a $3,600 buyback, they will absolutely choose the buyback since the true market price is $3,500. The user might feel quite pleased: "My $50 pack turned into $3,600!" What they don’t realize is that the inflated EV acts as a hidden tax levied on every pack they pull. This is subtle violence: no one gets robbed at gunpoint, but everyone pays regardless.

Conversely, if a card is undervalued, users will keep it. For instance, a card valued at $4,000 is priced at $3,500 by the machine with a buyback at $3,150. The user perceives the undervaluation, chooses to keep it or redeem the physical item, and later sells it on eBay for far more than $3,150. Over time, undervalued cards are removed by users, while overvalued ones get sold back and recycled into the pool, artificially forcing the machine’s EV upward.

The only solution is to accurately price every single card in the machine. Collector Crypt spent 14 months building a pricing oracle for Pokémon and One Piece cards, with sports cards next on the roadmap. They benchmark this system against alt.xyz and cardladder over the long term, measuring the average percentage deviation between each card’s latest predicted price and its next actual sale price. Their error rate is roughly 30% lower than the best alternative. Even with ~30% higher accuracy, every card undergoes manual review after minting and pricing. Not out of distrust in the model, but because human-in-the-loop significantly improves upon standalone AI outputs; manual corrections also feed back as training data, refining future predictions.

This also means third-party websites tracking EV and ranking platforms by "best EV per pack" miss a crucial detail: they rely solely on self-reported EV and card prices from the platforms themselves. The author notes that running their pricing oracle against other platforms confirms that most competitors (though not all) overvalue their cards by 10% to 20%. This isn’t necessarily about top-tier cards featured on the front page, but rather the mundane cards appearing from page two onward, which account for 99% of the cards users actually draw.

The Fourth Element: Shipping and Customer Service

The fourth element has nothing to do with the chain; it’s about people and service, making it the pillar nobody writes a whitepaper about.

In the United States and most developed nations, Amazon has raised the baseline for e-commerce fulfillment: order in the morning, receive it by afternoon; broken items or misdescriptions warrant hassle-free returns without questioning reasons. Collectors shop on Amazon too, and naturally hold card platforms to that same standard.

The current state of the card collecting scene is slow shipping, fraud, unreachable customer service, and transit damage. People have grown somewhat accustomed to it—the worst part is that they really are. Users crack open a flattened booster box and just shrug; this is conditioned expectation. Collector Crypt partners with OmniVault to deliver the best possible redemption experience. Their customer support covers Discord, email, and AI chatbots, staffed by personnel who actually have the authority to resolve issues.

Customer service and fulfillment experiences are also forms of friction. When users spend real money on collecting, any issue that arises must be taken seriously. If a platform ships a box of supposedly "factory sealed" boosters that clearly look resealed, the platform must accept responsibility rather than blaming customs or similar factors. If users receive crushed card boxes or cracked grading slabs, they deserve immediate returns or exchanges without having to justify it, since they did not pack the shipment themselves.

Therefore, it is best for platforms to own their own vaults; at minimum, they must have direct contact numbers for vault managers. When issues arise, a quick call to the vault manager allows the platform to know within minutes how to respond to frustrated users. When a user submits a ticket, gets an automated reply three days later, and then hears nothing for weeks while waiting for an alleged "investigation"—that is not customer service; it is pure delay. "We will follow up" is not a status update; it marks the beginning of a ticket vanishing into the void.

Thomas acknowledges that customer service will never be perfect, and perhaps never should be aimed at absolute perfection. What matters is fostering a culture that builds continuous improvement processes and accepts that this work is inherently ongoing. Collector Crypt’s vault manager is a former Amazon warehouse supervisor; he knows exactly where the standards should be set because he used to hold others accountable to those exact metrics.

The Four Elements Cannot Substitute for Each Other

Thomas mentions seeing competitors struggle across these exact areas. The four elements cannot offset one another: if any single component fails, the entire business falls apart.

Failing to publish a complete card database renders VRF meaningless: if you do not even disclose what is in the pool, no amount of elegant randomness proofs can verify whether the drawn card actually came from that public pool. Having both VRF and a complete card table still fails if you rely on lazy third-party pricing: randomness and inventory become transparent, yet the EV remains fictional, causing the money users pay to mismatch the value they receive.

Having a complete card table without VRF is even more straightforward: the database is authentic, but the pack-opening process remains unverifiable, meaning whatever the platform says was drawn is taken on faith. Even if the first three elements are perfected, delivering damaged cards six weeks later stuffed into crumpled bubble mailers will render your cryptographic proofs irrelevant. Users will not care about your mathematics when the physical experience is poor. No matter how flawless the verification is, a terrible experience will still drive customers away.