How the indices are governed

Every OLTA index is governed first by its rulebook, not by discretion. Composition and weighting follow a documented, rules-based methodology, and the divisor method holds the index level continuous across rebalances and corporate actions so a change in constituents never injects a phantom move into the published series. The rules are written before they are applied, and the same rules apply to every basket in the family.

An investment committee owns the methodology and reviews composition against it. It is the body that ratifies a rebalance, signs off on constituent additions and removals, and records the rationale for each change. Its remit is narrow on purpose: confirm that what the methodology produces still matches the thesis the basket was built to express, and document the reasoning wherever judgement is exercised inside the rules.

Listing and delisting follow the same discipline. A basket earns its place by tracking its construction thesis, and when the empirical record diverges materially from that thesis the divergence triggers a review rather than an automatic removal. The committee examines whether the methodology still describes the basket honestly and acts on the evidence. The bar to add is the bar to keep.

The useful description of a committee is not who sits on it, it is what it is not allowed to do. OLTA is an early-stage firm and the committee is small. Stating that plainly is more useful to an allocator than implying a board, so the model is described here by its constraints.

Four limits define it. The committee cannot restate a published historical series to improve how a basket reads, and a correction is marked as a correction. It cannot change the construction rules by a single decision, because a rules change is a methodology amendment that gets published rather than a discretionary call in a meeting. A member with a position in a constituent under discussion recuses from that item. And no change is approved without a written rationale attached to the record, which is what makes a decision reviewable a year later.

That fourth limit is worth stating precisely, because a committee limit that lives only in a paragraph is a promise rather than a control. For one class of change it is now structural. Whether a basket is open, and at which tier, is derived from published data by a rule this page describes in section 05, and the only way to depart from that derivation is a register entry carrying a reason of at least 120 characters that states a fact a reader can check, an author, a date and a term. Code refuses the entry when the reason is missing, so the badge cannot move ahead of the record. For constituent changes the same rationale is committee policy rather than an enforced control, and section 06 names that gap rather than borrowing credit for it.

Reviews are scheduled and quorate. Any member can call an unscheduled review, and an unscheduled action clears a higher bar than a scheduled one and carries a published reason. The triggers that justify one are named in advance rather than argued in the moment: a constituent delisted from the reference venue, a pegged asset breaking its peg beyond a stated threshold, a token migrating contract, or an issuer failure on the tokenized-equity side.

The point of naming triggers ahead of time is that the alternative is discretion wearing the costume of process. A committee that can decide, after the fact, that this particular week was exceptional is not applying rules. A committee that has to point at a listed trigger is.

Composition is reviewed on the cadence set in the rulebook and not on impulse. A predictable calendar is itself a governance control, because it removes the temptation to reshape a basket reactively when one name moves. The rulebook is the authority on that calendar, and it is public.

One distinction matters more than any single date, and it is the one most often blurred by index providers. The backtests published across the product are computed on a quarterly rebalance assumption. That is a modelling parameter of the simulation, disclosed per basket, and it is not a commitment about how often a basket will be reconstituted once real capital settles against it. A reader should treat the backtest cadence as an input to the simulation and the mainnet cadence as a matter for the rulebook, because they answer different questions.

The empirical record is refreshed independently of composition, so a basket does not have to change for its statistics to stay current. That separation is deliberate. It means a review is triggered by evidence rather than by the arrival of new data, and it keeps the published record honest in the stretches where the right decision is to do nothing.

An index provider publishes numbers, so sooner or later one of them is wrong. A bad price print, a corporate action applied a day late, a divisor not adjusted through a constituent change, a stale composition. The question an investment committee should ask is not whether it will happen. It is what the provider had written down before it did.

The restatement policy has four parts. A materiality threshold that separates an error corrected quietly in the next refresh from one that triggers a notice. A duty to disclose a material error together with the window it affected, so a reader can tell whether a decision they took was made on a wrong number. A ban on editing a published series without marking it as restated. And a record of every restatement, because a provider that corrects silently is indistinguishable from one that never erred.

Where this stands today: drafted, not published. It appears in the custody control table as being drafted rather than in force, and it is a precondition for accepting client capital rather than something to be added afterwards.

One preventive control does already run. A basket with less than 90 days of backtest window returns no Sharpe ratio anywhere in the product, and the refusal sits in the data layer rather than in the copy, so no surface can print a figure the window does not support. It is a small control, and it is the shape the rest should take: structural, not editorial. The register in the next section is the second control built to that shape.

Whether a basket is open, and at which tier, is derived rather than assigned. The derivation is a conjunction and both halves are published. A basket is open when its visibility is active, it has not been retired, no methodology defect is open against it and a measured series exists for it. It reaches the top tier when its mandate is broad rather than a tilt and its published grade clears the threshold. A weak grade does not withdraw a basket, it is printed beside it, because a rule that closes a product on a number can close one without anybody deciding to.

The register below is the only way to depart from that derivation, and it is deliberately a narrow instrument. Four properties make it one. An entry has to carry a reason of at least 120 characters stating a fact a reader can check rather than an opinion about merit, an author and a date, and code refuses the entry when the reason is missing, so a status cannot move ahead of its record. An entry lapses on its own after 180 days, so nothing survives by being forgotten. A revision does not edit the entry it replaces, it carries it whole, so the sequence of what was decided stays readable. And an entry can hold a basket at a lower tier or close it, and can never lift one, which is the property that matters most: nothing reaches the tradeable set except by clearing the derivation on the data.

Retirement is the one status that stays on the basket rather than in the register. A shelved basket keeps its page and its retirement note, naming the date, the reason and the successor where there is one, and it is terminal: nothing lapses it and no entry here can express it. Reading the catalogue with the retired filter is the fastest way to see what the same discipline looked like applied to a strategy that did not work.

Governance only counts if it can be inspected. The rules, the reasoning and the record should sit where an allocator can read them without asking, so the decisions that shape a basket are legible from outside rather than held privately.

The missing row is the important one, and it is now narrower than it was. Status changes are recorded, dated and published in section 05. Constituent changes are not: there is no dated, per-basket log of additions, removals and rebalances on this surface. They are written up in research desk notes, which is useful reading but is not an auditable log, and calling it one would be exactly the small overclaim this page exists to avoid. Building it is a mainnet requirement.

As governance moves onchain, that record stops being a document somebody maintains and becomes a ledger of proposals and votes. Same transparency, with the trust assumption removed. That is the argument for the roadmap below, and it is the only argument this page makes for it.

The committee model is the right governance for a preview surface. It is accountable, documented and fast enough to keep the rules honest. It is not the end state. The direction of travel is to move governance onchain, from a committee that publishes its reasoning to a community that proposes and ratifies changes in the open.

None of this exists today. There is no onchain DAO, no governance token and no vote on this surface. Treating it as a roadmap rather than a feature is itself a governance choice: decentralised control is credible only once the methodology, the tooling and the participant base are mature enough to carry it, and handing a vote to a thin electorate is worse governance than a committee that writes down its reasoning. The sequencing is deliberate, and the target is mainnet, H1 2027.

Governance is the part of an index that allocators scrutinise hardest, because it is where the long-run integrity of a basket is either protected or quietly eroded. The answer here is to keep the rules published, the cadence fixed and the record open today, to name the gaps rather than paper over them, and to move that discipline onchain as the platform matures.

Allocators evaluating the governance model for a mandate can ask for the detail directly, including the draft restatement policy and the unscheduled-review triggers in full. The contact for that is at the foot of this page.