What you can check yourself
Writing the word secure on a page costs nothing, so this page does not spend your time on it. Instead it states what OLTA is structurally unable to do and shows you how to confirm each one, then where the assets are, how every published number is computed, what can go wrong for you, and which controls are actually running rather than written down. OLTA is in public preview: the funds are simulated, nothing settles on any chain, and mainnet settlement is targeted for H1 2027.
- Client capital held
- None
- Keys OLTA can use
- 0
- Baskets with a grade
- 71 of 86
- Independent attestation
- None
Straight answers
No, and this is structural rather than a policy we could revise. The signing key is reconstructed in your browser, OLTA never receives it, and you can export it and walk away.
What OLTA cannot doNowhere. This is a public preview, every balance and position in the application is simulated, and nothing settles on any chain. Mainnet settlement is targeted for H1 2027.
Where the assets areIndex levels are computed by the divisor method from public constituent prices. Grades come from historical simulation, with the measurement window printed beside the figure.
Where the numbers come fromBelow 90 days of history. The withholding happens in the data layer, so no screen in the product can print a grade the window does not support, and the slot reads Not graded instead.
The 90 day ruleYou lose the key and nobody can restore it, or a basket falls and the loss is yours. Both are listed below with what reduces them, and where nothing does, the page says so.
What can go wrongYou can, for every row marked in force. No independent party has attested to these controls, OLTA holds no licence and no insurance covers a user of this preview.
The registerWhat OLTA cannot do
The list of things a platform promises is worth very little, because you would have to take all of it on faith. The list of things it is unable to do is different: each line rests on a mechanism you can inspect, and it holds whether or not anyone here behaves well. That is the only part of this relationship you do not have to trust us on, so it goes first.
Every row below names the capability OLTA does not have, the mechanism that removes it, and the step you take to confirm it without asking us for anything. A row that could not name that step would be a promise dressed as a boundary, and it would not be here.
OLTA cannot sign anything on your behalf
A transaction on your address is only valid once it is signed by your key, and that key is reconstructed inside your browser when the session opens. It is handed to the code running on your screen and to nothing else. No OLTA endpoint receives it, so no OLTA endpoint can use it.
Check it yourselfConnect with a social login, open Settings, then Security, and reveal the key. It appears because your browser already holds it. If you connected an external wallet instead, that wallet keeps the key and never exposes it to a web page, which reaches the same place by another route.
Settings, SecurityOLTA cannot see or reconstruct your private key
The reveal is a local operation. The value lives in component state for exactly as long as it is on your screen: it is never written to storage, never logged, never sent, cleared when you leave the card, and hidden again after sixty seconds if you walk away.
Check it yourselfOpen the network panel of your browser before you press reveal. Nothing leaves the page carrying the key. Then close the card and reopen it: you get the same key from your own browser, not from a record of ours.
Settings, SecurityOLTA cannot move, freeze or claw back what you hold
The activity ledger is append only by construction. Its endpoint answers reads and accepts new entries, and it exposes no update method and no delete method, so there is no route by which an operator could rewrite a position or reverse one. What you see is the sum of events you caused.
Check it yourselfExport your full activity from Settings as CSV or JSON and read every row. There is no administrative adjustment line, because there is no administrative adjustment. The file is built inside your tab and uploaded nowhere.
Settings, ExportOLTA cannot stop you from leaving
Self custody is only real if the exit works without our cooperation. Export the key, import it into any wallet that speaks the EVM rail, and the address is yours with no reference to OLTA. Nothing at that address is worth anything during the preview, which is precisely why now is the moment to test the exit rather than later.
Check it yourselfExport the key, then erase the session from Settings, Data and privacy. The page drops you back to the sign in gate and the local footprint goes to zero. Import that key elsewhere and the address still resolves, without us.
Settings, Data and privacyOLTA cannot ask you for an identity document
There is no upload field anywhere in this application, no proof of address step and no accreditation form, because no client capital is accepted during the preview. Identity checks belong to the mainnet design and they are not running on this surface.
Check it yourselfWalk the whole product. Nothing asks you to attach a file. The only personal detail any surface accepts is an email address, and only on a form you filled in yourself.
One honest caveat on the reach of all this. These boundaries describe the key and the record, which is what a self custodial design puts beyond our reach. They say nothing about the risk of the assets themselves, and nothing about what could go wrong at mainnet with counterparties that do not exist yet. Those live further down the page.
Where the assets are, and who holds them
Nobody holds anything, and that is the complete answer today. OLTA is in public preview, accepts no client capital, and every position, balance and profit or loss figure in the application is simulated: computed from public market data against a published rulebook rather than from trades that settled somewhere.
This matters more than it might look. A trust page that reads like an operating custody service, on a product that safekeeps nothing, is exactly the failure a careful reader is scanning for. So the table states what is true now, and marks the rest as design.
Verifiable marks a statement you can confirm in the product right now. Nothing in this table asserts a licence, an audit or an insurance policy, because OLTA has none of the three.
No custodian, broker or settlement venue is named anywhere on this site. The selection is in diligence and nothing is signed, so publishing a name would tell you something we do not yet know. The framework a counterparty has to satisfy before any capital moves is published in full, including segregation, signing authority and what happens if OLTA stops operating.
Where every figure on this site comes from
A number on a finance surface is worth exactly as much as your ability to find out how it was produced. Six kinds of figure carry most of the weight across the product. Each one is computed the same way on every screen that prints it, and each one has a thing it is not, which is usually the part that gets left out.
- Index levelSource: Divisor NAV
The divisor method, the same convention S&P, MSCI and FTSE use. The weighted sum of constituent prices is divided by a divisor that stays fixed between rebalances and was calibrated at launch so the basket opens at a designated level. Change the constituent prices and the level moves; change nothing and it does not.
What it is notNot a quote and not a traded price. Nobody is standing behind it with a bid, and no fill happened at it.
- 24 hour changeSource: Market datameasured over 24 h
The weighted average of the 24 hour returns of the constituents, which is arithmetically the same thing as the index level today over the index level yesterday, minus one.
What it is notNot a realised return. No position of yours was opened or closed to produce it.
- Sharpe and the gradeSource: Backtest
Computed from a historical simulation of the construction rules over public daily prices, with the window printed next to the figure everywhere it appears. Two references are published beside it, bitcoin and a broad equal weight basket, because a ratio with no benchmark says very little.
What it is notNot a track record. It is what the rules would have produced, measured after the fact, with no fees, no slippage and no capital at risk.
- Maximum drawdownSource: Max drawdown
The deepest peak to trough fall inside the measured window, carried with its minus sign on every surface that prints it, so a bad number never gets to look like a good one.
What it is notNot a limit. A future fall can be deeper than the worst one in the window.
- Value at RiskSource: Historical VaRmeasured over 7 d and 30 d
Historical simulation at the 95% level: the fifth percentile of the basket's overlapping 7 day and 30 day return distributions, over the longest window available for that basket.
What it is notNot a floor and not a worst case. It is the level that one horizon in twenty was worse than, and it says nothing at all about how much worse.
- Your portfolio valueSource: Simulated session
Your simulated units multiplied by the current index level of each basket you hold, summed. The units come from your own preview session.
What it is notNot a statement of account. No capital was committed and nothing settled.
Under 90 days, the basket has a track record and not a grade
A Sharpe ratio measured over a short window is not a weak signal, it is noise. The awkward part is that noise flatters: the shortest windows in this catalogue produce the most spectacular ratios, several times higher than any genuine multi year figure in the library. Printing one of those beside a real one invites a comparison between two numbers that are not comparable, and the flattering one always wins that comparison.
So the rule is absolute. Below 90 days of measured window, the grade is withheld and the slot reads Not graded or Track too short. It is not estimated, not annotated with a caveat, not shown in grey. It is not shown. The gate lives in the data layer rather than in each component, which is what makes it a rule instead of a convention some screen could forget to apply.
Two consequences worth stating plainly. First, this page will not print those flattering short window figures either, not even as an illustration of the problem. Second, the gate covers the grade and not the basket: a tradeable basket with a short window keeps its page and its index level, and simply carries no comparable quality figure. Of 86 baskets in the catalogue, 71 currently clear the window.
What the catalogue deliberately does not show is as much a part of this section as what it does. There is no live trading and no order book, because nothing settles. There is no realised return on real capital, because no real capital is deployed. There is no fee schedule, because commercial terms are not set. There is no dated per basket change log, which is a genuine gap in the transparency stack rather than an oversight, and it is on the register below. And there is no third party attestation of any figure here.
What can go wrong for you
Written for the person carrying the consequence rather than for a risk annex. Each entry names something that happens to you, states the concrete result, and then says what reduces it. Where the honest answer is that nothing does, the entry says that instead of filling the column.
- Risk 01
You lose your key
The position is gone, permanently, for everyone. There is no reset link, no support override and no recovery desk, because the same design that stops OLTA from touching your funds stops OLTA from restoring them. This is the cost of self custody and it is not negotiable.
What reduces itExport the key from Settings and store it offline, on paper or on a device that is not the one you browse from, before you ever hold something of value. Test the exit while the stakes are zero.
- Risk 02
The basket falls
A diversified basket spreads risk across constituents that mostly move together in a crypto drawdown. It is not a hedge and it is not capital protection. Every drawdown published in the catalogue is a real historical fall, several of them severe, and a future one can be deeper. A total loss is possible.
What reduces itRead the drawdown and the Value at Risk before the Sharpe, size the position against the drawdown rather than the return, and treat the backtest window as the limited evidence it is.
- Risk 03
A basket is graded on too little history
A ratio measured over a couple of months is noise, and noise flatters. The product refuses to print a grade below the window threshold, which means some baskets you can read carry no comparable quality figure at all.
What reduces itCheck the window next to any grade. Where it reads Not graded or Track too short, treat the basket as unmeasured rather than as average.
- Risk 04
A basket is retired from the catalogue
A strategy shelved after methodology review keeps its page and its retirement note, so the public record shows what was tried and why it stopped, but it loses trade access. A holder of a retired basket at mainnet would need an exit path, and that path is part of the mainnet design rather than something running today.
What reduces itRead the retirement note on any shelved basket before assuming a family is still supported. Retirement is a terminal state and the catalogue never promotes a basket back out of it automatically.
- Risk 05
Your jurisdiction is not served
The application resolves your country into an access tier before it renders anything, and an unrecognised country fails closed rather than open. Several country codes are blocked outright, and others reach the crypto baskets but not the tokenized equity ones. Access can also change after you have started.
What reduces itCheck the tier that applies to you before you build a plan around this product. The rule is applied by the application itself, not by a checkbox you can decline.
- Risk 06
A published number turns out to be wrong
Data pipelines break and figures get restated. OLTA has drafted a policy covering the materiality threshold, the duty to disclose and a ban on silent edits, but it is not published, so today you have no document to hold us to on this point.
Nothing reduces itThis is a real gap and it is listed on the register below rather than smoothed over. Until the policy is published, the only remedy is to tell us and to see the correction made in public.
Two of these have numbers attached that you can check right now. 11 baskets in the catalogue are marked retired and keep their page with the note explaining why the strategy was shelved. 16 country codes are not served at all, resolved before the application renders and failing closed on anything it does not recognise.
Nothing on this page is investment advice, a recommendation or a forecast. Every rate and every ratio in this product is either simulated or measured after the fact, so none of it is a statement about what happens next. Rule based construction spreads risk across constituents. It does not remove it.
What is in force, and what is still on paper
The split this register draws is the one that decides whether the rest of the page is worth anything. The first table holds controls that run today and that you can confirm without asking us. The second holds the things that exist as intention, including the ones a careful reader would want most.
Every row here is checkable in the product, in the published rulebook, or by reading what the application does. Nothing enters this table on the strength of an intention.
The table below is the part that usually gets left off a page like this one. There is no independent attestation, no published incident policy, no per basket change log and no signed custody counterparty. Each is a precondition for accepting client capital rather than an improvement to be added afterwards.
These are stated as gaps rather than as features in progress. When one of them ships it moves to the table above, and not before.
This register covers the trust surface only. The full product record, meaning what has shipped, what is in build and what is still in design across the whole application, is maintained per item on the roadmap rather than duplicated here.
If something looks wrong, tell us
One desk answers everything, at business@olta.finance. There is no separate security address and no bug bounty programme yet, and inventing either on this page would be the kind of small overclaim the rest of it exists to avoid. Write to the desk and say what you found.
A report is most useful with four things in it: the page or the ticker, what you saw, what you expected instead, and roughly when. If you believe a published figure is wrong, include the number you read and the number you computed. What we intend on a confirmed error is a correction made in public rather than a quiet edit, and until the restatement policy is published that is an intention rather than something you can hold us to.
Never send us a private key, a seed phrase or a recovery share, in a report or anywhere else. We cannot use one, we do not want one, and a message asking you for one did not come from OLTA.
What OLTA will never ask you for
- Your private key, your seed phrase or your recovery share, in any channel, for any reason
- A signature to unlock, verify, migrate, validate or restore a wallet
- A transfer to a support address, a recovery address or a validation address
- Remote control of your screen or your browser
- A payment to release, upgrade or reactivate an account
Anyone asking for one of these is not OLTA, whatever the message looks like. Nobody at OLTA can act on your key even if you send it, and sending it hands your position to whoever asked.