This document states what we believe about Pokémon cards as an asset class, why we believe it, and what would change our mind. Every figure in it is drawn from a published Moonstone Research study and is traceable to its methodology section; where our evidence is thin or missing, we say so. Nothing here is financial advice.

I. Thesis statement

Pokémon cards are a supply-driven asset class wearing a demand-driven costume.

The demand side — thirty years of franchise attachment, a collector base that regenerates with every generation of children, and trading volume that now settles modern chase-card medians on hundreds to over a thousand recorded sales — is real, durable, and largely already priced. What the market persistently misprices is supply: when the presses stop, how populations grow inside grading slabs, and how long launch-window inventory keeps arriving after launch-window demand has peaked.

Our published research supports a specific, falsifiable version of this claim:

  • The highest-priced gem-grade cards today disproportionately come from the low-print window — consistent with, but not yet proof of, scarcity-driven returns. The 2002–2007 print trough (Aquapolis, Skyridge, the EX-era chases) out-prices Base Set icons in gem grades today — an Aquapolis Kingdra carries a PSA 10 median of $7,544 — because those sets were printed when the franchise was commercially cold, and graded populations can never catch up. This is a price-level observation: no entry-baseline return series exists for these singles (our daily history starts January 2025), so we state it as level evidence, not measured returns (The Best Pokémon Investments of the Past 30 Years).
  • The supply state of a set is observable from the outside. Our print-clock model — cheapest standard Elite Trainer Box ÷ $49.99 MSRP — classifies all 17 known out-of-print Sword & Shield sets correctly at ≥2.3× retail, while freely printed Mega-era sets sit at ≤1.4×. No out-of-print set trades below 2.2×; no freely printed set trades above 1.6× (The Print Clock).
  • The launch curve is measurable and repeats. Across all nine sets launched January 2025 – July 2026, chase cards fell to a median 0.85 of launch peak by day 30, 0.81 by day 90, 0.68 by day 180, and troughed at ~0.56 around day 224 — month seven, not launch week. Sets reading near or above launch (≥0.9) at day 90 went on to hold or beat launch; every set below 0.7 stayed depressed; the band between is indeterminate on this small sample (The Launch Curve).
  • Grade economics are population economics. The PSA 10 premium over a 9 runs 2.59× (SWSH) to 8.82× (DP–HGSS) by era, and a one-factor power law on population structure — premium ≈ 3.7 × (pop9 ÷ pop10)^0.47 — explains nearly half the variation (R² 0.46, n=278) (The Grade Ladder).

The investable conclusion: buy supply that is ending, avoid supply that is arriving, price the grade with the registry rather than the story, and treat liquidity as an exit tool rather than a return signal. The rest of this document is the evidence.

II. Core principles

1. Print-window scarcity — not fame — is the alpha

Claim. The market's top price tier is dominated by cards and sets printed when nobody was paying attention, because demand recovered and supply cannot. We read this as scarcity-driven appreciation — but the evidence is current price levels on thin samples, not return series with entry baselines, and we hold the claim at that strength.

Evidence. In current PSA 10 medians, the $5,000–$7,500 tier is dominated by the 2002–2007 scarcity window, not Base Set icons: Aquapolis Kingdra $7,544 (n=7), Metagross Gold Star $7,425 in PSA 9 (n=32), Salamence ex δ $7,225 (n=4). The July 2026 tape rhymed: 11 of the top 20 raw movers came from the neglected 2007–2014 BW/DP window, led by Riolu BW33 at +172% in 30 days (30 Years; The Month in Pokémon).

2. The print clock is readable, and the era prior turns one number into two signals

Claim. Retail-price ratio detects print status, and identical ratios mean opposite things on dead versus current eras.

Evidence. Calibration: 17/17 SWSH sets ≥2.3× retail (known OOP); freely stocked Mega sets ≤1.4×; the bands do not overlap. A >2× ratio on a dead era is permanent scarcity (Evolving Skies ETB at 10.46× is the ceiling); the same ratio on a current era — Phantasmal Flames at 3.07× while presses run at stated maximum capacity (~10B cards/yr per TPC fiscal disclosures) — is a supply failure that the company's documented reprint playbook has repeatedly crushed (The Print Clock).

3. Launch supply is a tax; the trough is month seven; day 90 tells you which curve you are on

Claim. Paying week-one prices for modern chases has been a reliably losing trade, and the accumulation window is months six through nine — with a classifier available at day 90.

Evidence. Cohort medians across nine sets: 0.85 / 0.81 / 0.68 of launch peak at days 30 / 90 / 180, trough ~0.56 at day ~224, then recovery (Prismatic 0.56→0.94, White Flare 0.72→1.06, Black Bolt 0.84→1.28). Every set reading ≥0.9 of launch at day 90 (including White Flare at 0.94) traded at or above launch a year later; every set below 0.7 stayed depressed; between those bands the sample is too small to call. Prismatic Evolutions' mania-baseline detour is the disclosed exception that makes this a prior, not a law. Separately, six of the fourteen worst full-window performers in our loss ledger were Prismatic chases bought at January 2025 release pricing, down 24–34% in six months (The Launch Curve; Where Money Went to Die).

4. The entire condition premium lives in the last grade point — and the registry prices it

Claim. A 10-over-9 spread is the hobby's largest return lever; an 8-to-9 upgrade buys almost nothing; and population structure sets fair value measurably.

Evidence. Era medians: 10÷9 runs 2.59× (SWSH) to 8.82× (DP–HGSS); 9÷8 runs only 1.18–2.25×. Pop quintiles are monotonic: pop-6–83 cards trade at 7.66× their 9s, pop-5.9k–49.5k at 2.46×. The fair-value power law (R² 0.46) yields a discrepancy screen — liquid PSA 9s trading 50–65% below population-implied fair (Gyarados EX BREAKpoint at −65% on 94 sales) — July's slab winners were vintage 9s under unbuyable 10s (Shining Celebi 9 +45% MoM) — a condition-substitution pattern; note these names traded ABOVE the screen's pop-implied fair value, so the discount screen and the substitution pattern are distinct signals, not one (Grade Ladder; The Month in Pokémon).

5. Grading is an option, and it is frequently negative-EV before the box ships

Claim. Expected value — registry grade distribution × per-grade sold medians, minus raw price and fee — must be run before every submission; intuition fails in both directions.

Evidence. 31% of WotC-era cards in our screen grade at negative expected value versus 2% for Scarlet & Violet. A raw Shining Tyranitar destroys $2,007 in expectation per submission (raw $4,250, pop-weighted slab EV $2,265, 5.6% gem odds). The reverse trap: Champion's Path Charizard VMAX gems at 69.7% and is still negative-EV, because 29,541 existing PSA 10s crushed the spread (Where Money Went to Die).

6. Liquidity is exit quality, not a return signal — it has been negatively correlated with returns

Claim. Use volume to size and exit positions; do not buy it as a signal.

Evidence. Across a 424-card join of the eBay volume tape to 18-month price histories, the most-traded quintile returned a median +51% while the middle quintiles did +113–114%; correlation of log-volume to return is −0.14. The mechanism is Principle 3: the highest-velocity names are new-set chases and icon reprints whose supply keeps arriving (The Velocity Report). The structural corollary from our 30-year study: modern medians settle on 100–1,500 sales while vintage rarities price on 3–7 — two asset classes in one sleeve; size vintage like art, trade modern like equities.

7. Sealed is the index fund — but only pack density and chase equity compound

Claim. Sealed product delivers EV-weighted set exposure without grading risk, and every surviving flagship era compounded at double-digit rates from retail; format selection inside sealed is decisive. Note the conditioning in that sentence: "surviving" means the exhibits below are the winners' bracket, and the CAGRs are gross, listing-derived, and survivor-selected — they have not yet been tested against realized sold prices (agenda #10).

Evidence. From documented retail baselines to July 2026 list prices: Base Set shadowless 1st Edition box 19.7%/yr over 27 years; Gym Challenge 20.2%; Legendary Treasures 46.6%; Evolving Skies case 80.9% over five years (listing prices, thin market — trajectory, not executable size; gross of all costs, see Section IV). Meanwhile no Elite Trainer Box among 167 tracked trades below its $49.99 retail — but pin and trinket formats compounded at only ~9–12%/yr against 25–34% for pack-dense and chase-promo formats of the very same sets (Hidden Fates Charizard tin ~34% vs Champion's Path pins ~11%) (30 Years; Where Money Went to Die).

8. Demand is attachment-driven and durable — but it only converts to returns through supply

Claim. Character equity (and artist equity) is a real, persistent bid, yet it pays only where the print window caps supply.

Evidence. The strong camp beat the cute camp on consistency (+110% vs +71% median 18-month return; 98% vs 76% of cards positive), but cute owns the right tail (75th percentile +185%) and the throne — Flareon Gold Star PSA 10 at $51,994, the most expensive card in either camp. The same lesson in artist markets: Mitsuhiro Arita's index leads all 16 tracked illustrators at $22,952, yet his median tracked card trades at $17.41 across 400 printings — the market prices the print, not the pen. Buy the window, not the byline (Cute vs. Strong; Illustrator Spotlight: Arita).

9. The chase-card halo is a real demand catalyst — but a short-lived one, and we treat it as a timing overlay, not a holding thesis (factor under validation)

Claim. When a new set's chase card revives a specific older Pokémon, that Pokémon's older, unrelated prints earn an excess return over the market — but the excess lives in the announcement-to-release window and dissolves at launch, so it is a catalyst to be timed and sold into, never a reason to hold. We label this a factor under validation: the direction is measured and consistent, but the sample is small and was observed inside a broad vintage bull market.

Evidence. Two clean natural experiments in our daily data. When Black Bolt / White Flare were announced (May 6, 2025) built around Zekrom and Reshiram, seven older Zekrom/Reshiram prints ran +19.2% in the six weeks after the announcement against a +1.7% market, and +26.1% by international release against +8.0% — roughly ten points of that excess attributable specifically to being the chase (a same-era non-featured legendary control rose ~16.5%). After launch, the featured basket underperformed that control — the halo dissolves rather than reverses. Live, since the Delta Reign / Storm Emeralda reveal of Mega Rayquaza ex (late June 2026), old Rayquaza prints are +15.4% while same-trio non-featured Groudon and Kyogre are −1.5% and −1.6% — a clean +17-point within-family spread. Two boundary conditions keep the factor honest: it is undetectable on saturated names (old Charizard earned ~0 excess from the September 2025 Mega set — a name the market never stops pricing has no slack), and it round-trips for late buyers (the pre-launch anticipation on vintage Umbreon into Prismatic Evolutions had exhausted itself before the set shipped; the Eeveelution complex then returned +74% against a +139% universe over eighteen months — left behind, halo already spent) (The Chase-Card Halo; Where Money Went to Die).

How it feeds ratings. The halo enters the Weekly Buys factor model as a timing overlay, not one of the two independent factors a pick must clear — precisely because it is transient and its cleanest cases sit inside a rising market. Concretely: (1) it can only upgrade a card that already clears the two-factor bar on durable grounds — a supply-frozen, dead-era print (print-clock TIGHT/OOP) of a newly-featured Pokémon that the halo has not yet reached (a laggard, not a name that has already run); it never manufactures a pick on its own. (2) It carries an explicit exit rule inherited from the launch-curve work: sell into the launch, do not hold through it — the excess is a ~3-month rental. (3) It is disqualifying in reverse: a featured Pokémon whose prints have already run hard into the reveal (Evolving Skies Rayquaza VMAX, +27% pre-launch) is flagged as buying the vertical move, the most expensive habit in our loss ledger, and screened out. This is the same discipline as every other factor here — the halo pays only where supply is fixed and only before the crowd agrees, so we underwrite the frozen supply and merely time with the halo.

III. What we avoid, and why

In-print sealed above retail. A current-era set trading at 2.4–3.1× MSRP (Mega Evolution, Phantasmal Flames today) is a supply failure, not scarcity. The publisher has a stated maximum-capacity commitment and a documented history of crushing exactly these premiums with reprint waves. We buy wind-down laggards on dying eras (Twilight Masquerade at 2.12×, Temporal Forces at 2.28× — the least scarcity priced before a reported early-2027 print stop); we do not pay scarcity prices while presses run (Print Clock).

Launch-window chase singles. The cohort's default path is −44% from launch peak to a month-seven trough. Week-one buyers of every recent set have funded this. We wait for the day-90 reading, and we accumulate sliders in months six through nine, not week one (Launch Curve).

Buying the vertical move. Every deep drawdown in our loss ledger was an entry problem, not a card problem: Neo Discovery Umbreon −91% from its 2025 peak, Base Set Charizard −70% peak-to-now for whoever bought the spike, the whole Umbreon complex unwinding together. Momentum without a supply story is the most expensive thing we track (Where Money Went to Die).

Top-of-tape liquidity as a thesis. The most-traded quintile was the worst compounder (+51% vs +114% mid-curve). We hold liquid names for deployability, priced with that expected-return haircut in mind — never because volume itself looks like validation (Velocity Report).

Negative-EV grading submissions. No raw card is submitted without the pop-weighted EV math. 31% of WotC-era candidates fail it; so does a 70%-gem modern card with a 29,541-deep PSA 10 population (Where Money Went to Die).

Accessory-format sealed. Pins, trinkets, and merchandise-padded boxes lagged pack-dense formats of the same sets by 2–3× in CAGR for a decade. Value scales with pack content and chase equity, never with accessories (Where Money Went to Die).

IV. The cost stack: every headline number above is gross

Every figure in Sections I and II — the CAGRs, the medians, the ratio floors, the EV screens — is gross and pre-tax. None of it nets out what a realized strategy actually pays. Stated plainly, the stack is:

  • Marketplace take on exit: ~13%. eBay and TCGplayer final-value fees plus payment processing run roughly 13% all-in on a sale. Every exit in this document silently pays it.
  • Grading all-in: roughly $20–40 per card, not the fee alone. Submission fee plus two-way shipping and insurance. Our grading-EV screen currently uses the ~$22 fee by itself, which means the 31% negative-EV share for WotC-era submissions is understated — at all-in costs it worsens, and every marginal-EV submission flips negative.
  • Storage and insurance carry. A physical portfolio of size pays ongoing custody costs that no listed price reflects.
  • US tax: collectibles are taxed at a 28% maximum long-term capital-gains rate — worse than the 20% cap on the equities every S&P comparison implicitly benchmarks against. Net-net, several percentage points of any headline gap versus equities evaporate.

A worked example from our own book. The wind-down trade in Section III buys Twilight Masquerade ETBs at 2.12× MSRP (~$106) betting on convergence to the ≥2.3× out-of-print floor (~$115). That is +8.5% gross — and negative net of a 13% selling take. The floor-convergence leg alone does not clear costs; the trade pays only if it reaches something like the SWSH median (>3×) outcome, which is a separate and so-far undemonstrated claim.

We have not yet re-run the flagship trades, the sealed CAGRs, or the grading-EV screen at all-in costs, and we will not invent the numbers here. As of July 2026 we commit to a costs appendix — round-trip cost stack by asset type (raw single, slab, sealed) and holding period, with the wind-down trade and the EV screen shown gross and net — building on agenda #12 (fee-regime sensitivity, already scoped as requiring no new data). Until it is published, every return figure in this document should be read as "gross, indicative."

V. Portfolio construction principles

We publish no numeric position caps, stop levels, or return targets here, because the studies that would justify specific numbers (agenda #3, #6, #8, #10) are not yet complete — and this document does not print numbers its research cannot support. What the evidence already forces are principles:

  • Themes are positions. Our loss ledger documents the entire Umbreon complex unwinding together — intra-theme correlation approaching 1 exactly when it hurts. A basket of correlated cards is one position wearing several sleeves; exposure must be capped per card, per theme, and per era, with the numeric caps to be set by the exit-haircut study (#8) (Where Money Went to Die).
  • Exits follow signals, not moods. The same clocks that gate entries gate exits: a set reading below 0.7 of launch at day 90 has stayed depressed at every later checkpoint in our cohort — that is a hold-nothing signal, not a dip; the best sell window for launch chases has been launch itself; and a current-era ratio above ~2× while presses run is a sell-into signal, not a scarcity confirmation (Launch Curve; Print Clock).
  • Capacity honesty. Vintage medians in our own data rest on 3–7 recorded sales, and exits take months and a discount. Sleeves priced on single-digit sale counts cannot absorb size: most of the measurable edge described in this document is capacity-constrained to hobbyist scale, and a vehicle of any institutional size would move these prints on entry. We say this because the alternative is discovering it on exit.
  • No unevidenced numbers. When the wind-down backtest (#3), direct graded returns (#6), and exit-haircut (#8) studies are published, this section gains numbers. Not before.

VI. Risk factors

We hold these views with stated uncertainty, and the following are the honest ways they break.

1. Reprint risk. Production capacity is ~10 billion cards/year — roughly ten times pre-2020 levels — and the publisher has publicly committed to reprinting hot product at maximum capacity. Any thesis premised on modern scarcity is one restock wave from repricing. Our print-status model is inferred from market data; TPCi publishes no schedules, and the ~3-year wind-down and early-2027 SV print stop are community/distributor-reported, not official.

2. Grading population growth. Slab populations only grow. SWSH-era product already gems at a 55% median rate and carries a compressed 2.59× gem premium; Champion's Path Charizard VMAX shows what 29,541 PSA 10s do to a spread. Every gem-premium figure in this document is a snapshot of a denominator that rises daily. We also found that certified scarcity has not been a momentum trade — the hardest-gem quartile appreciated least (+20.2% vs +39.1% for the easiest) over the 180 days we measured — so scarcity supports the level of premiums, not necessarily their near-term direction.

3. Era rotation. July 2026's median graded card fell 2% while select raw eras ran 50–170%. Rotation cuts both ways: the BW/DP repricing rewarded a decade of patience, but whoever holds the era rotation exits from is on the other side of those trades. We do not have evidence on how long rotations last or how to time their start — only that dispersion around flat medians is where selection matters.

4. Liquidity gaps. Vintage medians rest on 3–7 recorded sales; a single motivated buyer moves the print, and exits can take months and a discount. Several headline figures in our own research are explicitly thin (Salamence ex δ n=4; Butterfree Jungle PSA 10 n=4; Flareon ☆ n=8). Sealed CAGRs are computed from listing prices, not confirmed sales, in thin markets — indicative of trajectory, not executable at size.

5. Data limitations. Our daily price histories reach back only to January 2025 (~18 months); decade-scale claims route through documented MSRP baselines. eBay comp coverage skews toward graded sales by construction. Our own accuracy audits found that raw vintage near-mint cards sell 3–6× above mixed-condition market prices — a known skew we flag rather than hide. The day-90 launch classifier rests on seven sets with outcomes. The fair-value power law explains R² 0.46 — meaning the majority of a gem premium is still demand factors we do not model. The grade-ladder momentum finding measures raw prices because per-grade graded return history is not yet deep enough; we are accumulating daily graded snapshots to close that gap.

6. Regime risk. Our measured window covers a strong collectibles market. We have no measured evidence on how this asset class behaves through a sustained demand contraction, and we do not pretend otherwise; the 2002–2003 trough that created the best supply of the modern era was, for contemporaneous holders, a bear market.

VII. Methodology note

All Moonstone Research figures are:

  • USD only. No EUR or mixed-currency figures appear anywhere on the platform or in research.
  • Real transactions. Graded figures are median recorded eBay sold prices per PSA grade with stated sample sizes; raw prices are TCGplayer market data with daily history; population figures are true PSA registry counts. Sealed current values are tracked market prices; where they are listings rather than confirmed sales (thin vintage boxes), the article says so.
  • Baseline-honest. Decade-and-longer return claims use documented retail baselines (e.g., $3.24/pack WotC MSRP; $49.99 ETB MSRP) against current prices, with CAGR windows labeled.
  • Anomaly-excluded, never smoothed. Grade-inverted medians on thin samples, comp-mixed windows (mislabeled 1st Edition listings polluting unlimited medians), and >40× outliers are excluded and disclosed — 17 exclusions in the grade-ladder study alone, two in the July monthly.
  • As-of dated. Every article states its pull date; linked product pages price continuously and will drift from captured figures.
  • No synthetic numbers. Where the data does not exist (per-grade graded return series, official print schedules, pre-2025 daily history), we say the evidence is missing rather than model it into existence.
  • Selection-biased return universe — disclosed, rebuild in progress. Our 520-card graded universe was sampled on July 2026 prices (top-of-market plus stride-sampled mid-tier per era) — that is, selected at the end of the January 2025 – July 2026 measurement window, with a $10 floor that additionally drops anything that decayed out the bottom. Measuring backward-looking returns on a universe selected by end-of-window price inflates them: today's top-of-market disproportionately contains the window's winners. The direction of this bias is unambiguously upward. The "median 91%" universe return, the velocity quintiles, cute-vs-strong, and the gem-rate momentum quartiles all inherit it; until the rebuild ships, cross-sectional return magnitudes should be read as upper bounds and relative rankings as ordinal hints. The fix — a point-in-time rebuild that freezes a January 2025 basket on entry-date criteria only and tracks it forward including dropouts and sub-$10 decays — is agenda #13, in progress.

VIII. Research agenda — proposed future articles

The thesis above is strongest on grade economics, launch dynamics, and print-status inference; it is weakest on event-driven causality (reprints), longitudinal population effects, non-English markets, and realized (vs listed) sealed returns. The following studies would close those gaps, ordered roughly by how much of the thesis each would underwrite. Three of them — #3, #6, and #10 — are no longer merely proposed: as of July 2026 they are committed and gate deployment; no strategy premised on the wind-down trade, graded momentum, or sealed CAGRs deploys capital before its gating study is published. Available data: PPT API per-card price histories up to 10 years (weekly; depth varies), eBay per-grade sold-comp CSVs (daily export), PSA population data (point-in-time, batchable), set-value histories (~2.5 months accumulated), sealed price histories, and accumulating daily graded snapshots.

1. The Reprint Event Study: What Actually Happens When the Restock Hits

Question: Do reprint/restock announcements cause price declines, or merely confirm the launch curve already underway — and how large is the sealed vs singles split?

Data needed: A hand-built dated event list of documented reprint waves (Hidden Fates, 151, Prismatic restocks — community/retailer archives), joined to our daily sealed and singles histories for ±90-day windows around each event. Directly tests the "announcements are lagging confirmation" claim we currently assert from one cohort.

2. Population Growth vs. Price Decay: The Slab Supply Curve

Question: How fast do PSA 10 populations grow per era, and what is the measured elasticity of gem premiums to population growth?

Data needed: Recurring (monthly) PSA population snapshots over the 520-card universe — we hold point-in-time pops today; the longitudinal series must be accumulated. Joined to per-grade sold medians, this converts the R²-0.46 cross-sectional power law into a forward decay model, and quantifies risk factor 2.

3. The Wind-Down Trade, Backtested: SWSH from Era Death to Today

Question: When an era ends, how fast do laggard sets converge to the ≥2.3× OOP floor, and what did buying the cheapest sets at era-end actually return?

Data needed: Sealed price histories for all 17 SWSH sets from January 2023 forward (PPT sealed histories), ETB-ratio time series reconstructed backward. This backtests the print-clock buy signal we currently state as inference — directly underwrites the Twilight Masquerade / Temporal Forces positioning.

Status: Committed — required before any strategy deployment (July 2026).

4. Japanese vs. English: One Franchise, Two Markets

Question: Do JP prints of the same chases lead, lag, or diverge from English — and does the JP market's different print/reprint culture change the launch curve?

Data needed: PPT `language=japanese` coverage (verified working; USD-priced), matched JP/EN card pairs by set + collector number, plus the open JP tcgId backfill for our 3,297 ja- cards. The entire current thesis is English-only; this is its largest blind spot.

5. The Stamp Premium: Pokémon Center Exclusives vs. the 1st Edition Precedent

Question: Is the PC-stamp premium structurally similar to the 1st Edition stamp premium — a print-run marker that compounds — or a cosmetic tax that decays?

Data needed: Per-variant price and pop series for stamped vs unstamped modern pairs; the historical 1st Ed vs unlimited spread curve across WotC sets (per-grade eBay CSVs + pops) as the precedent benchmark. Our current 1st-Ed evidence is fragmentary (box CAGRs, gem-rate gaps); this would make it a principle.

6. Graded Returns, Measured Directly

Question: What do PSA 9s and 10s actually return by era and price tier — as opposed to the raw-market proxy our momentum findings currently use?

Data needed: The accumulating daily graded-snapshot series (started 2026), per-grade eBay sold CSVs over time. Closes the explicitly disclosed gap in the grade-ladder study and re-tests "certified scarcity is not momentum" on the correct asset.

Status: Committed — required before any strategy deployment (July 2026).

7. Rotation Clocks: How Long Eras Sleep

Question: The BW/DP repricing took roughly a decade of dormancy — is there a measurable dormancy-to-rotation pattern across eras (EX 2016–18, WotC 2019–20, BW/DP 2026) that puts current sleepers (XY? early SM?) on a clock?

Data needed: 10-year weekly PPT histories on era-representative baskets (depth varies per card — feasibility audit first), MSRP baselines where history is shallow. Turns risk factor 3 from a caveat into a signal.

8. The Exit Haircut: Measuring What Illiquidity Actually Costs

Question: What is the realized spread between market price, lowest active listing, and next sold comp across liquidity tiers — i.e., the true cost of exiting vintage vs modern?

Data needed: PPT market/low/listings/sellers fields snapshotted over time, plus dated sold comps per card. Quantifies the "size vintage like art" rule with a number.

9. The Near-Mint Ladder: Condition-Aware Vintage Pricing

Question: How large and how systematic is the NM premium over mixed-condition market for vintage raw — our audits found individual sales at 3–6× displayed market?

Data needed: Cleaned raw sold comps (graded-title screening already built) bucketed by stated condition vs TCGplayer condition ladders across a vintage universe. Both a research article and a product fix — the current skew is a trust risk.

10. Listing vs. Reality: What Sealed Actually Sells For

Question: How large is the gap between the listing prices behind our sealed CAGRs and realized sold prices — does the 80.9%/yr Evolving Skies figure survive contact with sold comps?

Data needed: eBay sold data for a sealed panel (the per-grade CSV export covers cards; sealed solds may need targeted pulls), against tracked listing prices. Hardens Principle 7's most-caveated numbers.

Status: Committed — required before any strategy deployment (July 2026).

11. The Artist Premium, Controlled

Question: Does an illustrator's name carry measurable price premium after controlling for set, rarity, era, and character — or is the artist index just print-window exposure in disguise?

Data needed: Full artist-attribution sweep (PPT artist field, known-inconsistent — needs cleaning), hedonic regression over the 20k-card catalog with per-card prices. Tests Principle 8's artist half; also fixes the disclosed Arita undercount.

12. Grading-Fee Regimes and the EV Frontier

Question: How sensitive is the negative-EV share to grading price and turnaround tiers, and where is the raw-price frontier below which grading never pays?

Data needed: Existing pop-weighted EV pipeline re-run across fee scenarios ($15–$75) and per-grade medians from the eBay CSVs; no new data required. Cheap to produce, and makes Principle 5 actionable as a calculator. Also feeds the costs appendix committed in Section IV.

13. The Point-in-Time Universe Rebuild

Question: What did an honestly frozen January 2025 basket — selected on entry-date price, population, and liquidity only, tracked forward with dropouts and sub-$10 decays included — actually return, and how much of the current "median +91%" survives?

Data needed: Our own January 2025 daily-history starting points and catalog snapshots; no new sources. Every cross-sectional figure in this document (universe median, velocity quintiles, cute-vs-strong, gem-rate quartiles) gets republished on the frozen basket. This is the selection-bias fix disclosed in the methodology note.

Status: In progress (July 2026).

14. The Chase-Card Halo, Backtested Across Every In-Window Launch

Question: Does the announce-to-release excess return on a new set's featured Pokémon hold across all launches in our history — not just the three hand-picked cases — and what is its mean magnitude, decay half-life, and hit rate once we control for the broad vintage rally?

Data needed: A dated event list of every set launch January 2025 – present with its unambiguous chase Pokémon, joined to the daily universe for standardized ±90-day event windows, with a matched non-featured control basket per event and a market-beta adjustment. This upgrades Principle 9 from "factor under validation" (three clean cases inside a bull market) to a measured overlay with a stated hit rate and exit rule — or retires it if the excess is just vintage beta in disguise. Directly gates whether the halo may ever upgrade a Weekly Buys pick.

Status: Committed — required before the halo influences any published rating (July 2026).

Moonstone Research, July 2026. All cited figures appear in the linked articles with their full methodologies, sample sizes, and exclusion rules. Collectibles are illiquid, volatile, and can lose value; nothing in this document is financial advice.