Reading the counterparty-chain signal.
A guide to the five public-facts dimensions tracked on the quarterly /signal page — ETF concentration, authorized-participant structure, central-bank purchasing, IMF COFER reserve composition, LBMA cleared-metal structure. What each one says structurally, and how to read them together.
Sourced. Structural. Not predictive. No allocation advice.
The counterparty chain is a system. The five dimensions are how to read it.
The Office reviews five public-facts dimensions every quarter and posts the reading at /signal. Each dimension is sourced from a specific named primary — the World Gold Council, the IMF, the LBMA, or filings of the major precious-metals ETFs — and each is paired with a structural reading that does not forecast a price or recommend an action.
Why five dimensions and not one. The counterparty chain is not a single number; it is a system. Any one dimension, on its own, can move for reasons that are not structural — a quarterly outflow at one ETF is not, by itself, a signal about the system. What makes the readings useful is the convergence (or non-convergence) across five dimensions that are sourced from different institutional bodies, capture different parts of the chain, and do not move together for trivial reasons.
This note is a guide to reading them — on their own, and together. It is a companion to /signal, not a substitute. The current readings are on that page; the framework for interpreting them is here.
What share of physical-backed ETF gold sits in a handful of vehicles.
The first dimension tracks how concentrated the physical-backed gold-ETF market is — how much of the trust-held metal sits in the top one, three, and five vehicles. The data comes from the trusts’ own monthly bar-list disclosures and from the WGC’s aggregated ETF inflows/outflows series.
What it says structurally. A concentrated ETF market is a concentrated counterparty market. If a few large vehicles hold most of the institutional ETF-held gold, the operational health of those few sponsors, custodians, and authorized-participant syndicates determines the operational health of the segment. Concentration is not the same as risk — the major vehicles are well-documented and audited — but it is the structural feature the dimension is designed to track.
How it moves. Concentration shifts slowly under normal conditions; large vehicles compete for inflows on fees and on tracking quality. It can shift faster during episodes that concentrate or fragment investor allocation — for example, a custodian-specific operational issue at one vehicle, or a regulatory action affecting one sponsor. The reading is most informative when read against the next dimension (AP structure), because the AP syndicate is the operational mechanism that enforces concentration in practice.
Which institutions can create and redeem ETF baskets — and how concentrated that list is.
The authorized-participant (AP) syndicate is the small set of institutions — large investment banks, in nearly all cases — that can create and redeem ETF shares directly with the trust. The list is disclosed in ETF prospectus updates and varies by ETF. The dimension tracks how concentrated the AP rosters are, whether they overlap across the major vehicles, and any disclosed entries or exits in the period.
What it says structurally. AP concentration determines the operational reliability of the share-to-NAV arbitrage that keeps the share price aligned with the trust value. If the same handful of large institutions are APs across most of the major ETFs, balance-sheet or operational stress at those institutions affects the arbitrage at multiple ETFs simultaneously. The March 2020 liquidity episode is the documented precedent: dispersed APs, but enough shared operational stress that NAV discounts opened across vehicles in the same week.
How it moves. AP rosters shift slowly under normal conditions; entries and exits are rare and disclosed. The most informative episodes are when an AP exits or is excluded for cause (a regulatory or balance-sheet event). The dimension is most informative read alongside the ETF concentration figure: a concentrated AP roster acting on a concentrated ETF market amplifies any shared stress through the same small set of operational pipes.
The aggregated central-bank line. Quarterly. The cleanest single signal of the official-sector posture.
This dimension is read at length in Office Note 03. The quarterly reading tracks the aggregated central-bank line from the WGC Gold Demand Trends release: how many tonnes were bought net in the quarter, which institutions are inside the aggregate, and how the trailing twelve months compare to the long-run average.
What it says structurally. The aggregated line is the cleanest single signal of how reserve managers, in aggregate, are treating gold. Net-buyer fifteen years running is the baseline. Acceleration above the long-run average, persistence at elevated levels, and the named institutions inside the aggregate (PBoC, RBI, NBP, TCMB, MAS) all carry structural information about how the official sector reads its own counterparty exposure to the foreign-currency portion of its reserves.
How it moves. The line moves quarterly and decisively. Inflection points (the 2010 net- buyer flip, the 2022 acceleration above 1,000 tonnes/year) are visible at quarterly resolution. Read alongside the IMF COFER reading: the two dimensions are separate signals of the same structural posture — reserve diversification away from a single counterparty currency.
The currency mix of allocated FX reserves. Quarterly. From the IMF directly.
The IMF’s Currency Composition of Official Foreign Exchange Reserves (COFER) is the quarterly dataset on the currency mix of the allocated portion of global FX reserves. The dimension tracks the dollar share over time, against the euro, yuan, yen, sterling, Australian and Canadian dollars, and the residual “other” bucket.
What it says structurally. The dollar share has declined from approximately 73% in 2001 to approximately 58% in the most recent quarterly readings of 2024. The decline is not concentrated in a single rival currency; it is distributed across the others and the residual “other”, which tells the same single coherent story across two decades: reserve diversification away from a single counterparty currency, not reserve concentration into a different one. This dimension is the FX-side reading of the same structural posture Dimension 03 reads on the gold side.
How it moves. Slow but persistent. The 1.5%–2% per decade decline in the dollar share is not a forecast of a near-term break; it is a multi-decade structural drift. Quarterly noise is real (FX revaluation effects can move the shares without any actual reallocation), so the reading is most useful at trailing-four-quarters resolution.
LBMA cleared turnover, vault stocks, and the unallocated/allocated balance.
The London Bullion Market Association publishes monthly statistics on cleared turnover, vault stocks held in London (broken out by Bank of England and member-bank vaults), and other operational measures of the cleared metal market. The dimension reads these alongside the unallocated-vs-allocated balance the Bank for International Settlements observes through its institutional surveys.
What it says structurally. The cleared-metal market is the operational plumbing through which most institutional gold transactions move. Concentration of vault stocks at a few member banks, shifts in the ratio of unallocated to allocated balances, and step-changes in cleared turnover all carry structural information about where the chain’s operational pressure is sitting. This dimension is also the connection to Office Note 06 — the unallocated-vs-allocated balance at the institutional level is the systemic counterpart of the custody-mode question at household level.
How it moves. Slowly under normal conditions; visibly during stress. The 2020 unallocated/ allocated rebalancing episode (when several large institutional clients shifted from unallocated to allocated positions during the COVID liquidity episode) is the documented precedent for how this dimension behaves under pressure.
Convergence, not any single reading, is what carries structural weight.
No single dimension is meant to carry the structural reading on its own. An ETF outflow quarter, on its own, says nothing about the counterparty chain. A central-bank quarter near the long-run average, on its own, says nothing about the official-sector posture. A flat IMF COFER quarter, on its own, says nothing about reserve diversification. The reading is in the convergence.
What the Office watches for is the configuration in which several dimensions move in the same direction at once, sourced from independent institutions, with structural rather than narrative explanations. The 2022–2024 stretch is the most recent example: central-bank purchasing accelerating above 1,000 tonnes/year, IMF COFER continuing its multi-decade dollar-share decline, LBMA unallocated/allocated rebalancing toward allocated, ETF and AP concentration broadly stable. Four dimensions carrying the same structural signal — reserve diversification away from a single counterparty currency — for related but distinct reasons.
No single quarter’s reading should be acted on. The discipline is to read the dimensions together, over multi-quarter windows, and to document the structural reading rather than the narrative one. The Office’s job, in publishing the readings on /signal, is to do that work and show the work; the household’s job is to read the work alongside its own portfolio context.
The signal is in the convergence. The reserve is the response to it, not a forecast about it.
The reading the Office takes from the convergence, and what it does not.
What the Office takes from a convergent reading across the five dimensions is the structural posture: the official sector, the cleared-metal market, and the long-run reserve composition data are saying the same thing about the same risk — that a balance sheet entirely composed of claims on counterparties is structurally incomplete, and that the institutions whose job it is to hold reserves at scale have concluded so explicitly enough to act on it.
What the Office does not take is a price forecast. None of the five dimensions, alone or together, are reliable signals about the next quarter’s gold price. The structural argument is robust to that not being known.
“If five different bodies, looking at five different parts of the same system, with their own institutional reasons for caring about it, all show the same posture — that’s not noise. That’s the system telling you what it actually looks like. Most readers are watching the price. The signal is in what the institutions whose job is to be early are doing on their own balance sheets.”
Eric Roach · Co-founder
The reading is one input. The brief is the document that acts on it.
The Strategy Brief takes the structural reading from /signal as one of its inputs — alongside the portfolio-efficiency case in Office Note 02 and the chain-of-claims structural reading in Office Note 04 — and translates them into a directly-titled position the household can hold.
None of the inputs are decisive on their own. The brief’s job is to take the structural reading, the household’s objectives, the existing portfolio, and the household’s counsel and wealth advisor into the same document and to write down the position that is consistent with all of them.
“A signal is not a strategy. The five dimensions tell you what the system looks like; the brief tells you what your reserve looks like inside that system. The discipline is keeping those two questions separate. Misreading the signal as a directive to act on the price is the most common failure mode of people who otherwise agree with the structural argument.”
Jose Gomez · Co-founder
Primary sources for an advisor or counsel reviewing the five dimensions.
- World Gold Council, Gold Demand Trends and ETF Flows. Quarterly. Source for Dimensions 01 and 03 — ETF concentration and central-bank purchasing.
- SPDR Gold Shares (GLD) and iShares Gold Trust (IAU) prospectus updates. Source for Dimension 02 — AP roster composition. The full registration documents are available through SEC EDGAR.
- IMF Currency Composition of Official Foreign Exchange Reserves (COFER). Quarterly, IMF Statistics Department. Source for Dimension 04.
- LBMA monthly statistics; BIS quarterly review. Source for Dimension 05 — cleared-metal structure, vault stocks, and unallocated/allocated balance.
- HAR /signal page, current quarterly reading. The Office’s own publication of the five readings, dated and reviewed each quarter. Available at /signal.
- HAR Office Notes 02–06. The four bodies of long-form analysis the readings sit alongside. Available at /research/notes.
The six notes are the synthesis. The brief is the deliverable.
The six Office Notes are the long-form synthesis behind the four bodies of research the firm reads against. They are written as one continuous argument; they are also fine to read in any order. The Strategy Brief is what takes the synthesis and applies it to a household’s specific situation in writing.
The reviewed Physical Reserve Strategy Brief is delivered within five business days of intake. The brief is the document that takes the structural reading above and applies it to your situation specifically.
Begin a Strategy Intake →