Manifesto·v2.0·Kraków, September 2026
Why Viroan exists, what we refuse to build, and what we want this market to look like in five years. The short version is on every other page. This is the long one.
It is versioned rather than quietly edited. Version one described a product we have since narrowed, and saying so on the page is part of the point — what changed is at the bottom.
I · Why we exist
Algorithmic trading has been an insiders’ game for a long time, and the reason is smaller and more boring than a conspiracy: there is no way to tell a working system from a well-chosen screenshot. Not from the outside, and not from the inside either.
The period, the settings, the costs left out of the arithmetic, the version of the code that produced the chart. Each of those is a decision made by someone with an interest in the answer. A second screenshot does not fix a first one.
Publishing the source is not an answer most creators can afford to give. Posting more numbers is not an answer anybody believes. The honest work gets priced next to the dishonest work, which is a tax on exactly the people who should win.
We are not the strategies. We are the thing that makes it possible to find out what they are.
Viroan · founding principleSo the job is narrow on purpose: read the source against what the creator declared, run the build on periods the creator never chose, count the costs the creator left out, seal what survived, and publish the record with its own limits printed on it. We sell licences to software. We do not take custody and we do not place orders.
II · What we value
When two of these conflict, the one higher on the list wins. That is the only thing that makes a values list useful rather than decorative.
The check is worth exactly as much as our freedom to fail it. We are paid when a licence sells, never for a verdict, and the fee does not move with the score. A creator can pay to be seen — featured placement and search priority are products — and cannot pay to be passed. Any arrangement that makes an approval worth more to us than a rejection is refused, whatever it pays.
this one is not tradeableEvery certificate prints what the check did not cover, on the same page as the score. A check that claimed to cover everything would be a marketing document. Naming the gap is the difference between a measurement and a badge.
the gap is published, not buriedThe rules do not become negotiable when there is revenue on the other side of them. We would rather lose a listing, a partner or a round than approve something we could not defend to the person who bought it.
including when it is expensiveThe buyer keeps their keys, their funds and the ability to stop without asking us. The creator keeps every right to their code — we read it, we never own it, and we never publish it unless they choose to. Neither side is locked in by anything except whether the product is good.
no custody, no code ownershipWe do not complicate things to look competent. Eight fields in the same order on every listing, one score, one certificate. A buyer should understand what they are getting in thirty seconds; a new colleague should onboard without a call.
last, because clarity loses to honestyIII · What we are, what we aren’t
+ What Viroan is
Software is the product, a licence is the transaction, and the check is the reason to buy here rather than anywhere else. We mediate a sale; we do not trade.
Your money sits on your own account and stays there. We hold no keys and send no orders: the platform sells licences and nothing more.
The source — not a binary — is read against the declaration, and a person signs off every submission before it lists. The build runs on data and periods we pick and hold back from the creator. Out of sample, with costs counted, every time.
The creator sets a list price, receives 85% of it, and the buyer pays 115% of it, with VAT included in the one total they are shown. That is on the page, in numbers, before anyone signs anything.
− What Viroan isn’t
We take no deposits, quote no prices, margin-call nobody. We are not authorised to do any of it and we do not market as though we were.
No pooled capital. Nobody at Viroan decides what your account does. A licence is software you run, not a mandate you grant.
No alerts in a channel that you implement by hand at 3am, and no private group where the real information supposedly lives.
A high score means the build did what it claimed under our tests. It is not a forecast, not advice, and not a reason to size a position. Automation can lose money, including all of it.
IV · The line we won’t cross
Every company has them. Most do not publish them, because publishing one converts it from a preference into something you can be held to. That is the reason to publish.
No temporary safekeeping, no liquidity pool, no clever wrapper, and no withdraw permission on your exchange account — not even a whitelisted one pointing at us. If we ever need to touch your principal, we have failed.
No back door, no trusted-partner exemption, no expedited tier. Our own builds go through the same six steps. If we do not have the receipts, it does not list.
No screenshot-only equity curves, no vanity charts, and no figure whose source we cannot name. Where we have no data — conversion, expected returns, how the first cohort performs — the page says we have no data instead of estimating.
The creator-side share, the buyer-side share and the VAT are all visible before the first click.
A failed re-check re-opens review and may, after a human decision and a statement of reasons, lead to delisting; your existing licence keeps working unless revoked for a stated reason. Certificates go public; rejections never do, because a failed submission is not news about a person.
Featured placement and search priority are paid features of the marketplace, and they are labelled as paid wherever they appear. What is not for sale at any price: the score, a pass, a faster review, a softer review, or the wording of a certificate. One of those moves a product up a list. The other would make every certificate behind us meaningless retroactively.
the line is between attention and judgementV · How we work
Less theatre than you would expect, more receipts than you would want. A small team running like a small team, on purpose.
Monday we plan in sixty minutes. Friday we ship. Wednesday is the only call. Everything else is written, searchable and waiting for you.
Every review is readable by the whole team, including the ones that ended in a rejection. We argue with the report, not with opinions about the report.
Anything touching the check, the seal or money needs two signatures. Anything that does not ships when it is ready. No theatre either way.
If someone can make the call without escalating, they should. The fewer steps between a problem and its solution, the better the solution.
We do not glorify overwork. Evenings are for family, sport and rest. A professional finds the fastest clean solution, not the longest one.
Knowledge does not live in one person’s head. Every material decision is in the wiki, including the ones we later reversed and why.
VI · Decision filter
A feature, a hire, a partnership, a pivot. It has to pass all three, and the third one has failed more proposals than the other two combined.
If it adds screens, steps or jargon without reducing how much the user has to take on faith, it does not ship. That is the whole product, and it is the whole filter.
The buyer’s keys and funds, the creator’s code. If the answer needs a footnote, the answer is no. Safety here is a hard yes or a hard no.
Assume the thing shipped, a buyer lost money, and someone is reading this decision back to us. If the honest answer is “we would look like we had an interest”, it does not ship.
VII · Five years from now
The world we would like to wake up to in 2031. Not a promise and not a plan — a description of what we are optimising for on every product call. Our own roadmap is on the About page and it is much shorter than this.
2027
Expectation
When somebody says they have a trading bot, the first thing asked is who read the code and on what data — the way “is it open source?” became the first question for any privacy claim.
2028
Standard
Any venue that publishes a score also publishes what the score did not cover. A number without its own boundary stops being accepted as information.
2029
Enforcement
A creator’s declaration is checked on every order rather than in a document nobody reads, and other venues adopt the same per-order baseline.
2030
Economics
A creator with a real system earns more than a creator with a better screenshot — from licences, from renewals, from a record that accumulates across versions. Right now the incentive runs the other way, and that is the thing we are actually trying to break.
2031
Retirement
It becomes a joke in retail trading the same way “just give me your password” became a joke in security. If that happens and somebody else did it, fine. We would like to have helped.
VIII · Our motto
Verified software.
No blind trust.
This is version two
Version one was published in May 2026 and described a platform that sends client orders under enforced risk limits. What we built is narrower: a marketplace that checks software and sells licences to it, holding nothing. The manifesto had to narrow with it.
Three things were removed rather than softened. They are listed beside this paragraph instead of summarised in it, because a manifesto you can quietly rewrite is not a manifesto — and because two of the three actively contradicted the product that exists.
The next revision will be the interesting one. When the first cohort has been checked, the first licences have sold, and we know what our own rejection rate actually is, we will come back and see whether this document still describes what we built. If it does not, the change gets published here — same URL, same versioning, with the old version still readable. No quiet edits.