Reading tokenized-card economics: sales, buybacks and valuation

Editorial decisions and publication status are recorded in the article metadata. Evidence read: 2026-10-19. Prepared with automated editorial assistance; automated work does not constitute human approval, and no product was directly tested. Collector Crypt's applicable legal jurisdiction and individual eligibility are not established here. Courtyard's company location and contractual choice of law do not establish worldwide availability. Reported estimates, official-document statements and fictional calculations are identified separately.

1. Separate sales, cash, inventory and accounting

An onchain card-pack sale can be followed by an immediate buyback, a retained token, a secondary transfer or physical withdrawal. Each path answers a different economic question. Gross pack sales measure purchasing activity. Buybacks measure one outward payment path. A token's stated reference value describes a valuation input. An accounting profit measure requires a defined treatment of inventory, acquisition costs, recognition dates and other expenses. These quantities cannot replace one another merely because they are all expressed in dollars.

This research uses Collector Crypt and Courtyard as document-based examples of those distinctions. It does not rank them, establish a current executable offer or independently verify their financial statements. The central conclusion is a measurement rule: first identify the period, product mix, valuation basis and included cash flows; then decide whether a numerical comparison is meaningful. A large volume does not by itself show a large profit, many distinct people or a large amount of new outside capital.

The two original research reports considered below have different samples. Pine's quarterly report concerns Q3 2026, with a stated data cutoff of October 10. Bitquery's realized-cash investigation covers December 7, 2025 through July 13, 2026. Neither the dates nor their measurement methods are interchangeable. These are original third-party investigations, rather than audited issuer accounts. Their reported observations remain attributed to their authors.

2. Map the item and the exit before calculating a return

The following is an original investigation diagram, not a claim that every platform implements the same custody or legal structure:

physical card + grading record
        -> recorded item identity + reference valuation
        -> token/control record + applicable contract
        -> custody obligations and verification questions
        -> keep | accepted buyback | secondary sale | physical withdrawal
        -> cash, a retained claim, or delivered property: different outcomes

Each arrow needs separate evidence. A grading identifier does not prove that the current vault contains the identified card. Token control does not establish an intellectual-property license. A described withdrawal process does not prove a completed shipment. An offered buyback is not an unconditional right to exit at the last displayed valuation.

Collector Crypt's metadata documentation describes fields including insured value. Its withdrawal documentation describes fees, token burning and shipping steps. Its VRF documentation describes a two-step verification mechanism. These document statements establish what the publisher describes; they do not independently establish inventory, policy coverage, physical performance, the actual deployment or the fairness of every draw. Metadata, withdrawal, VRF.

3. Read Pine's quarterly numbers with their method attached

Pine's report, published October 12, 2026, lists approximately USD 433.5 million in total revenue, USD 433.1 million in pack revenue and USD 408.1 million in costs for Q3 2026. Their reported difference is USD 25.4 million, labelled gross profit, or approximately 5.9% of total revenue. Those are Pine's estimates under its stated method, not a financial audit performed by this project. Original report.

The denominator matters: USD 433.5 million is total revenue, while USD 433.1 million is the pack component. A pack-only model cannot silently use total revenue as pack sales. The cost figure also needs its components: Pine includes buyback cash, a cost assigned to physically redeemed cards at 85% of insured value, and secondary card purchases. The assigned redemption cost is a valuation-based estimate, so the complete cost total is not simply observed cash leaving the platform during that quarter.

Inventory reference values and historical restatements further limit comparisons between report versions. A change in a displayed inventory valuation is different from a completed sale. Without a consistent acquisition-cost and inventory-recognition policy, the quarterly difference cannot establish an independently verified accounting margin, operating profit or distributable cash. Solana activity, fiat flows or a chain-specific sample identify transaction coverage; they do not identify the legal jurisdiction governing a holder's rights.

4. Read Bitquery's wallet sample as realized cash

Bitquery reports that 22% of 17,544 wallets in its December 7, 2025–July 13, 2026 sample had nonnegative realized-cash results. This is a statement about that investigation's covered wallets and flows. Retained cards, fiat activity and secondary sales are excluded from that result. Wallets are not people: one person can control several wallets, and a wallet can have operational purposes that the sample cannot resolve. Original investigation.

It would therefore be inaccurate to restate the result as the percentage of all collectors who earned a total economic profit. A wallet that paid cash and retained a card can have a negative realized-cash balance while still holding an item; the item's eventual sale value, costs and legal availability remain separate questions. Conversely, a reference value does not turn that retained item into realized cash.

This sample ends before the end of Pine's Q3 period and uses a different coverage boundary. Combining a buyback ratio from one study with a return ratio from another does not create a newly observed same-period retention rate. The publication date of the Bitquery page was not established in the supplied retrieval evidence; its sample dates and the October 19 retrieval date are recorded separately.

5. A reference price is not an executable exit

Suppose a pack advertises an expected reference value above its sale price. That statement concerns a weighted valuation, rather than a guaranteed cash receipt. An immediate offer may apply a discount, exclusions, fees or an acceptance condition. A holder choosing physical withdrawal may instead face shipping, identity checks, time and third-party charges. These paths cannot be valued by applying one headline percentage to every outcome.

The term insured value is especially easy to overread. A metadata field does not supply the full insurance policy, beneficiary, deductible, exclusions or claim procedure. It also does not independently prove that a sale can be completed at that value. Complete policy coverage, current acquisition cost, lien or insolvency treatment, inventory sufficiency and actual physical fulfillment remain unknown in this research.

Current pack-specific expected values and buyback rates were not verified from the live gacha interface. The USD 55.13 value in the supplied material is not used as a confirmed current quote. Four Pillars and Blockworks originals were not successfully reread in the source investigation, so their proposed explanations are not presented as newly verified evidence. This report does not infer a historical cause for margin compression from a scenario identity.

6. Use a conditional model with one consistent base

For a specified period and matched product mix, let S be pack sales, P pack price, EV expected reference value, v = EV/P and b the buyback fraction of that reference value. Let R = b × v. Define q = E[V × I(sell back)]/E[V]: the reference-value-weighted sellback share. It is not the fraction of cards, wallets or people. With differing packs or discounts, q and R must use compatible weights; separate averages cannot be multiplied without checking that condition.

R = b × v
B / S = q × R
g = 1 − q × R − K

B is the model's matching buyback amount. K is other modeled cost or adjustment divided by the same S, excluding amounts already included in B. Unknown K remains unknown. R already contains v; multiplying qR by v again counts the expected-value multiple twice. The result g is a conditional modeled balance rate, rather than a conclusion about audited gross margin.

Consider a wholly fictional teaching scenario, unrelated to any current platform offer: S = USD 100, v = 1.05, b = 0.90, q = 0.80 and K = 0.02. Then R = 0.945, B/S = 0.756 and g = 0.224. Modeled buybacks are USD 75.60, other modeled costs are USD 2 and the balance is USD 22.40, or 22.4%. This is arithmetic under stated assumptions, not a forecast.

Holding q, K and S fixed, increasing R from 94.5% to 95.5% reduces g by 0.8 percentage points and the fictional balance by USD 0.80. Formally, ∂g/∂R = −q under those fixed conditions. A real pricing change can also change sales, sellback choices and product mix, so this sensitivity does not identify a causal business response. Explore the same percentage units in the manual scenario tool; its inputs are user assumptions, not imported platform terms.

One cross-source calibration illustrates why numerical proximity is insufficient: assuming 0.920/0.94 gives q ≈ 0.9787234, not an observation that 98% of cards were sold back. Assuming S = USD 433.5 million and g = 0.059 gives USD 25.5765 million, rather than Pine's USD 25.4 million difference; the approximate rate and total-versus-pack denominator need reconciliation. Neither calculation converts the mismatched studies into an audited financial reconstruction.

7. Keep contract statements within their scope

Courtyard's NFT Terms and Asset Management Services Agreement distinguishes physical-asset terms, intellectual-property licensing and contractual obligations on transfer or succession. Its 6% provision applies to accepted vending-machine buyback offers; it is not a fee on standard marketplace transfers, a complete statement of all exit costs or a quote verified for a particular holder today. The provision is evidence of the captured contract wording, rather than independent proof of asset title, insurance or performance. Asset-management agreement.

The separately retrieved Terms of Service page has an outer help-center date of February 24, 2026 and a body marked last updated December 23, 2024. Both dates must remain visible as a version ambiguity; the newer outer date cannot silently update the body. The text identifies a Delaware company and California choice of law. Those facts do not show that residents of every jurisdiction qualify or that every dispute will have the same result. Identity and holder checks and additional charges must be considered separately. Terms of Service.

No aggregate issuer, custodian or holder-right classification is supplied for this multi-platform study. Unknown entries in its structured metadata are deliberate. A contract description, official documentation, original third-party research and direct transaction testing are different evidence levels. No account, draw, buyback, shipment or insurance claim was performed by this project.

8. Exercise, limits and source map

Exercise: in the fictional scenario above, keep S = USD 100, q = 0.80 and R = 0.945, but let K be unknown. Can the balance still be reported as USD 22.40? Answer: no. Only modeled buybacks of USD 75.60 are established by those inputs. The balance is USD 24.40 minus unknown other modeled costs; it cannot be labelled a known margin. If R already equals b × v, multiplying the buyback amount by v again is also incorrect.

For an update, first record the exact report version, period, product mix and cash/inventory definitions. Then check the contract version, offer conditions, reference methodology and independent operational evidence. Any new financial assertion or changed translation requires renewed review of that actual revision. This article does not recommend a product and has no sponsorship or affiliate arrangement; actual editorial decisions are recorded in its metadata. It does not establish individual eligibility, legal enforceability, fair-value accounting or insurance recovery.

Related foundations: assets and claims, valuation and liquidity, token versus legal ownership, legal structure and evidence-led product evaluation. Continue with the fictional verification lab and TCG overview.

Source IDs and actual documents read by the preceding source analysis on October 19, 2026:

• tcg-pine-q3-2026: Pine original quarterly report. Published October 12, 2026; Q3 estimates with October 10 data cutoff. Third-party research, not issuer-audited accounts.
• tcg-bitquery-gacha-study: Bitquery original investigation. Sample December 7, 2025–July 13, 2026; publication date unknown. Realized covered cash, excluding retained items, fiat and secondary sales.
• collectorcrypt-nft-metadata: Collector Crypt metadata. Official-document description only; publication and effective dates unknown.
• collectorcrypt-card-withdrawal: Collector Crypt withdrawal. Official-document process description only; publication and effective dates unknown.
• collectorcrypt-vrf-documentation: Collector Crypt VRF. Official-document mechanism description only; publication and effective dates unknown.
• courtyard-nft-asset-agreement: Courtyard asset-management agreement. Reused source identity; publication/effective dates and independent legal effect unknown.
• courtyard-terms-help: Courtyard Terms of Service. Outer February 24, 2026 date and body December 23, 2024 date conflict; applicable current version remains unresolved.