TS-GMRI-AUT — worked example

B.1 Raw inputs (primary sources)

B.1.1 Physical gold reference — LBMA Gold PM Fix

• Source: LBMA (prices.lbma.org.uk, JSON feed)
• Instrument: Gold PM Fix, 15:00 London
• Value today: — / troy oz
• Role: primary physical-gold reference for the fix-view premium calculation

B.1.2 Regulated tokenized-gold reference — CFB PAXGUSD_RR

• Source: CF Benchmarks Ltd (cfbenchmarks.com/data/indices/PAXGUSD_RR)
• Administrator: CF Benchmarks Ltd, FCA-authorised Benchmark Administrator, FRN 847100 (Kraken subsidiary)
• Regulatory status: UK BMR–registered benchmark — the only BMR-registered PAXG fix in existence
• Publication time: 15:00 London (aligned with LBMA Gold PM Fix by design)
• Methodology: volume-weighted median of PAXG-USD trades on regulated CEX venues over the fixing window
• Value today: —
• Role: regulated on-chain reference — anchors the "fix view" premium for PAXG

B.1.3 Token daily closes and on-chain state — CoinGecko

Source: CoinGecko public API, /coins/{id}/history and /coins/{id}/market_chart. Snapshot time: 00:00 UTC daily close.

All eligible tokens are represented in the TS-GMRI-AUT constituent universe per the methodology's One True Source for All (OTSFA) principle. Any new token that meets the eligibility gate published in the methodology is added on the next quarterly reconstitution.

B.1.4 Live spot proxy — CME/NYMEX front-month futures

Used only for the live view (intraday premium tracker), not for the daily fix view. Source: Yahoo Finance, GC=F (CME gold front-month). Role: near-spot proxy for live-vs-live arbitrage signal when LBMA fix is not yet available (intraday).

B.2 Fallback cascade

BMR Art. 12 / IOSCO Principle 6–7 alignment. The methodology requires an explicit, deterministic input-hierarchy. The pipeline evaluates each layer in order and drops to the next only when the higher-priority source is unavailable or fails freshness checks.

B.2.1 Fix view — tokenized-gold reference

1. CF Benchmarks PAXGUSD_RR (UK BMR–registered, primary)
2. LBMA Gold PM Fix + issuer redemption ratio (direct physical reference)
3. CoinGecko token close (aggregated CEX)
4. Previous-session value, flagged stale (no fresh input available)

Status today: —.

B.2.2 Live view — intraday tokenized-gold reference

1. CoinGecko live (default)
2. CoinMarketCap live (fallback)
3. Chainlink PAXG/USD on-chain feed
4. Uniswap v3 30-minute TWAP
5. Freeze last known value, flag stale

B.2.3 Physical spot reference

1. prices.lbma.org.uk (JSON feed)
2. Yahoo Finance GC=F / SI=F / PL=F / PA=F (front-month futures proxy)
3. Previous business day close, flagged stale

B.3 Derived values — premium decomposition

Two independent premium calculations are published for every constituent, in line with the methodology's requirement to separate wrapper-attributable pricing from reference-choice noise.

B.3.1 Fix view — token vs regulated benchmark

Formula: premium_fix = (CoinGecko close / CFB PAXGUSD_RR) − 1

premium_fix(PAXG) = —

B.3.2 LBMA view — token vs physical

Formula: premium_lbma = (token_price / oz_per_token / LBMA_Gold_PM_Fix) − 1

premium_lbma(PAXG) = —

B.3.3 Reconciliation — the fix-vs-LBMA gap

gap = —

This gap is fully attributable to the wrapper-side pricing spread between the regulated on-chain benchmark (CFB PAXGUSD_RR) and the LBMA physical fix. It is not noise: the CFB benchmark is volume-weighted from regulated PAXG-USD venues at 15:00 London; the LBMA fix is a physical-metal auction at the same fix time. The gap represents the market's assessment of tokenization frictions (issuance, redemption, on-chain settlement, custody) on that day.

Publishing both views is required by the methodology so that users can attribute observed premium/discount to (a) the wrapper itself versus (b) the choice of physical-vs-tokenized reference. Reporting only one view would collapse two distinct signals into one number.

B.4 Peer table (fix view)

KAU typically trades at the tightest discount to the CFB reference. Kinesis operates its own layer-1 chain and offers direct physical redemption at allocated LBMA-vault gold; its low oz/token denomination (0.03189) reflects a per-gram (not per-ounce) unit. When re-scaled to per-troy-ounce, KAU's implied spot typically sits within a few basis points of the LBMA PM Fix, consistent with its physical-redemption model. All tokens are traded on-chain and their prices are sourced through the same aggregation methodology (CoinGecko), ensuring apples-to-apples comparison.

B.5 Governance trail

Every value on this page is reproducible from public data using the pipeline described in Chapter 11 (Methodology Index) of the book and in the source code published at github.com/TSM-Hub-AK. A scheduled task (daily, 14:05 UTC) executes:

1. LBMA JSON → data/gmri/spot.json
2. Yahoo Finance front-month futures → data/gmri/live-spot.json
3. CFB PAXGUSD_RR embedded JSON → data/gmri/cfb-paxg.json
4. CoinGecko for each constituent → data/gmri/tokens-current.json + data/gmri/tokens-history.json (dual-section: fix and live)
5. scripts/build-comparison.py → comparison series
6. scripts/refresh-family-from-cadences.py + scripts/generate-gmri-pages.js → 19 family-member pages
7. Commit the day's snapshot to the public repository

No value in the index can be changed without a corresponding commit in the public repository. This satisfies the "committed audit trail" requirement of both UK BMR (Art. 15) and the IOSCO Principles for Financial Benchmarks (Principle 18 — Audit Trail).

B.6 Static reference version

The static, book-quoted version of this worked example — anchored to the 2 July 2026 snapshot — is published as Appendix B of the TS-GMRI Methodology Reference edition. Download it as part of the full book:

→ TS-GMRI Book · Reference Edition (Vol II, Appendix B)