PSR100-MY1,295.3+0.8%Pokémon (30d)RM938.2k+6.4%Magic: The Gathering (30d)RM407.1k+2.5%One Piece (30d)RM63.0k+3.8%Event VolumeRM13.4kFlagged Sales6Fresh Constituents10/12Monkey D. Luffy (Leader)RM253.5-2.6%Shanks (Manga Rare)RM833.55-2.2%Jinx, Loose CannonRM571.73-1.5%RX-78-2 GundamRM188+1.1%
SimulatedSimulated market — the index is moving on demo trades from the market simulator, not yet a live transaction feed. Every value is labeled accordingly.
Home / Methodology

PSR100-MY Index Methodologyv2.0

The complete, versioned rulebook for how Metaru Research computes the PSR100-MY TCG Index: data sourcing, eligibility, pricing, weighting, calculation, governance, and error correction. Changes to this document require committee approval and are announced in the Metaru Monthly Index Report.

1. Objective

PSR100-MY measures the aggregate price performance of the most actively traded trading-card assets in the Malaysian market, denominated in ringgit and derived from verified transactions. It is designed to be replicable from published rules, resistant to manipulation, and honest about data quality — every published value carries a data status and a methodology version.

2. What PSR100-MY is — and is not

  • It IS a transaction-based market benchmark: a measurement of realized prices.
  • It IS ringgit-first: computed from MYR sales, not converted overseas prices.
  • It IS rules-based: eligibility, weighting, and pricing follow this public document.
  • It is NOT investment advice, a recommendation, or a rating of any card or vendor.
  • It is NOT a security, fund, ETF, or any investable product.
  • It is NOT a guarantee of liquidity — cards may be unsellable at the reference price.
  • It is NOT predictive: historical index values do not forecast future results.

3. Data source hierarchy

Every observation is tagged with a source tier. Higher tiers carry more weight in pricing; Tier 4 never prices anything.

TierSourceBase weightDescription
Tier 1Event data network1.00First-party POS transactions captured at partner events: negotiated final prices, barcode-verified card identity, cert numbers, timestamps.
Tier 2Licensed feeds0.90Contracted data from Malaysian shops and marketplaces with verified settlement — invoice-level sales, not listings. Auctions weight 0.85.
Tier 3Public records0.60Completed public auction and marketplace sales that can be identity-matched to a tracked asset with reasonable confidence.
Tier 4Listings0.00Asking prices. Never used to compute a reference price — only as a sanity bound and staleness signal.

4. Card identity model

An indexed asset is a fully specified variant, not a card name. Identity is the tuple of game, set and set code, collector number, language, printing, finish, and condition class — raw condition grade or grading company, numeric grade, and cert number for slabs. Two printings of the same artwork are different assets; the same cert re-sold is the same physical asset, which is exactly what the circular-trading screens rely on.

5. Eligibility rules

An asset qualifies for the constituent universe when all of the following hold:

RuleThreshold
Accepted sales, trailing 12 months≥ 10
Accepted sales, trailing 90 days≥ 3
Accepted turnover, trailing 12 months≥ RM5,000
Data confidence score≥ 70
Manipulation risk score≤ 60

Assets that meet some but not all thresholds may be placed on the watchlist by the Metaru Index Committee; watchlist assets are published but carry zero index weight.

6. Pricing model

The reference price is a weighted median of log prices over the pricing window. Working in log space makes the estimator robust to multiplicative outliers; the median makes it robust to a minority of bad observations even after weighting.

RefPrice = exp( weighted_median( ln(P₁)…ln(Pₙ); w₁…wₙ ) )

Each sale’s weight is the product of four factors:

wᵢ = SourceWeightᵢ × CertWeightᵢ × OutlierAdjᵢ × RecencyDecayᵢ

Recency decay uses a 45-day half-life: an observation 45 days old carries half the weight of one from today.

RecencyDecayᵢ = 0.5 ^ (ageᵢ_days / 45)

Cert-confidence weights

Match levelWeightNotes
Cert matched1.00Cert number verified against the grader's registry.
Cert probable0.85Cert visible but not registry-verified at capture time.
Identity matched0.70Exact variant match without a cert (raw cards).
Aggregate only0.40Sale matched at card level but not variant level.

Outlier adjustment

The estimator does not band observations by standard deviations. Each sale carries an outlier adjustment set by its review disposition — accepted at full weight, accepted with an analyst-set haircut, or excluded — so a genuine but unusual price is only ever attenuated by a documented human decision, never by an automatic σ rule.

Review dispositionWeightNotes
Accepted1.00Passed automated screens or analyst review — full weight.
Accepted with haircut0.01–1.00Genuine but attenuated by an analyst-set weight (defaults 0.50).
Excluded0.00Rejected, quarantined, or pending review — contributes nothing to price.

7. Freshness labels

LabelRuleFactorEffect
Fresh≥ 3 accepted sales in the trailing 90 days1.00Full freshness factor.
Acceptable≥ 3 accepted sales in the trailing 180 days (but not 90)0.80Down-weighted; monitored.
StaleFewer than 3 accepted sales in 180 days0.30Heavily down-weighted; weight capped at the stale cap; review triggered.
SuspendedNo accepted sale in 365 days, or committee/integrity action0.00Removed from the calculation until reinstated.

8. Data confidence score

A 0–100 composite of seven components; the weights sum to 100:

ComponentWeightMeasures
Transaction count25Linear; full credit at 40+ accepted sales / 12 months.
Source quality20Weight-average source-tier quality of contributing sales.
Source diversity15Full credit at 3+ distinct accepted sources.
Cert confidence15Average cert-confidence factor of contributing sales.
Freshness10Fresh 1.0 / Acceptable 0.7 / Stale 0.3 / Suspended 0.0.
Outlier cleanliness10Share of sales that were not rejected or haircut.
Dedup confidence5Confidence that cross-source de-duplication is correct.

9. Manipulation risk score

A 0–100 score built from automated screens over the raw transaction log; the weights sum to 100. Assets above 60 are ineligible; assets above 40 receive enhanced review.

ComponentWeightMeasures
Low transaction count25Full risk at 0 sales; zero risk at 24+/12 months.
Abnormal price jump20Score from the consecutive-return jump detector.
Venue / vendor concentration20HHI of venue shares, or a max-share figure.
Repeated counterparty / device15Share from repeated buyer/device concentration.
Stale price10Freshness-derived staleness risk.
Refund / cancel / override rate10Share of reversed or overridden transactions.

The qualitative screens that feed those components include:

  • Counterparty concentration — repeated buyer/seller pairs or a single vendor dominating an asset's volume.
  • Circular patterns — the same physical cert re-trading in short windows at escalating prices.
  • Price laddering — sequences of small incremental sales that walk the reference upward without independent demand.
  • Deviation pressure — clusters of sales far from the prevailing reference (both directions).
  • Override frequency — POS price overrides and unusual discount patterns vs. asking prices.
  • Timing anomalies — off-hours or end-of-window clustering that correlates with rebalance dates.

10. Weighting — liquidity-adjusted capped turnover

Weights reflect where the market actually trades, damped so that a single expensive card cannot dominate. For each constituent, with T the trailing 12-month accepted turnover (RM), N the trailing 90-day accepted sale count, C the data confidence factor (score / 100), and F the freshness factor:

RawWeight = √T × min(1, ln(1 + N) / ln(26)) × C × F

Raw weights are normalized to sum to 100%, then the following caps are applied iteratively (excess weight is redistributed pro-rata) until all constraints hold:

ScopeCap
Single card (asset)5%
Single character / subject12.5%
Single set20%
Single game45%
Stale constituent (max weight while STALE)1%
Note: the current demo universe is intentionally small, so published weights use proportionally scaled caps. The production caps above apply once the full universe is live.

11. Index calculation

The index is chain-linked from constituent price relatives using the capped weights:

Index_t = Index_(t−1) × Σᵢ wᵢ × ( Pᵢ,t / Pᵢ,t−1 )

The index was set to its base level on the base date (published on the index page). Weights are rebalanced monthly at committee effective dates; between rebalances, weights drift with prices as in any chained index.

12. Index lifecycle and inaugural launch

An index does not exist — and accumulates no history — until it launches. Every instrument moves through a published lifecycle:

  • Incubation. A candidate universe accumulates verified transaction data. No levels, constituents, or charts are published — the index has no values to publish.
  • Inaugural report and slate selection. Analysts publish the inaugural Metaru Monthly Index Report and the Metaru Index Committee selects the top-N most investable eligible assets by PSR Eligibility Score — a 0–100 composite of liquidity (35%), data confidence (25%), sales breadth (20%), market continuity (10%), and manipulation resistance (10%) — subject to the hard eligibility rules in section 5. At least 8 eligible constituents are required to launch.
  • Launch. Launching fixes the base date and the index starts at 1,000.00 on that date. No history is backfilled, ever — a shorter honest history is better than a long fabricated one.

13. Reference Print vs. indicative values

The market runs continuously — 24/7, with no open or close. Once per day at 16:00 MYT the Metaru Research takes a Reference Print: a reference mark computed after that moment’s data cut-off and analyst review queue are cleared. The Reference Print is the value of record for reports and corrections; it is a bookmark on a continuously-trading index, not a settlement or a close. Indicative values update in real time as accepted transactions arrive and are marked as such — they may be revised by the next Reference Print and carry the live data status shown next to every number.

14. Monthly committee process

  • The Metaru Index Committee meets monthly on a published schedule.
  • Inputs: the constituent packet — eligibility screens, watchlist, stale and high-risk assets, and candidate additions.
  • Decisions (add / remove / keep / watchlist / suspend / reinstate) are taken by recorded vote.
  • Members declare conflicts before each vote and abstain where conflicted.
  • Decisions and effective dates are summarized in the next Metaru Monthly Index Report.

15. Error correction policy

TierThresholdAction
Tier A — minor≤ 0.10% index-level impactCorrected in the next daily Reference Print; footnoted in the monthly report.
Tier B — material0.10% – 0.50% impactRestatement within two business days with a public correction notice on the reports page.
Tier C — significant> 0.50% impactFormal correction notice, republication of affected Reference Prints, and a committee post-mortem published in the next report.

16. Conflicts of interest

Analysts and committee members must disclose personal holdings in tracked assets and may not review or vote on assets where they hold a position. Vendor identities in the event data network are pseudonymized in analyst tooling. Metaru Research does not deal in the cards it indexes.

17. Disclaimers

Metaru Research publishes market data and benchmarks. PSR indices are not investment advice, not securities, not ETFs, and not a guarantee of liquidity or future performance. Cards do not produce cash flow. Historical index values do not predict future results.