Risk Warning

Please read carefully. This text forms part of the Terms of Service and you confirm it with a separate tick box when you register.

KVERX is not an investment service

KVERX is a simulated laboratory. It is not an investment service, not investment advice, not investment research, and not a recommendation to buy or sell any financial instrument. The operator is neither a securities dealer nor an investment intermediary, never accepts or holds client money, and nowhere in the application does it tell you what to buy or sell. Nobody on the team is licensed to advise you — and we do not try.

The same goes for the Data chatbot, the Story of the Day, the Daily Dilemma and the leaderboard. They are teaching and entertainment features, not signals.

Every result is simulated

Kverks trade on the internal KVERX Market. Prices come from real market data delayed by roughly fifteen minutes, but no order is ever sent to a real exchange. A profit or loss shown in the application does not exist outside the simulation.

The simulation is deliberately harsher than reality: fees, the bid-ask spread and slippage carry a surcharge of roughly 15 %. Even so, a simulation does not capture everything. It does not model broker outages, rejected or partially filled orders, trading halts, taxes, or market behaviour under extreme illiquidity. The simulator is a model, not a copy.

Most kverks will not be remarkable

We say this up front, because it is the most likely outcome. A kverk can run for months, reach 100 % maturity and end up with a Score near zero. That is not a bug and not clumsiness on your part — it means that in the data it saw, there was no usable signal for its temperament.

Many kverks will end up losing. We do not hide losses: they are visible in the detail view, in the journal and on the public leaderboard. The point of a laboratory is that the result is true, not that it is pleasant.

Leverage multiplies both directions

A kverk may trade with borrowed capital: total exposure up to 200 % of account value for equities and 300 % for crypto, depending on its DNA. Anyone who does not want leverage sets the cap at 100 %.

Leverage is not a profit amplifier, it amplifies both sides. Double the exposure means double the move in the account, up and down alike, plus a daily financing cost on the borrowed part and, for shorts, a borrow fee on top. If equity falls below the maintenance margin, the simulator force-reduces positions (a margin call) — at the worst moment and at the worst price.

That is why leverage never looks like a bonus in the application: a kverk above 100 % exposure carries a label on its card, and the academy has a chapter on why it hurts in both directions.

Fuses are not a guarantee

A kverk has four layers of protection: the stop-loss, the daily loss limit, the margin call, and the main fuse, which puts the kverk to sleep and closes its positions once the drawdown passes the threshold you set. In practice they almost always fire before the account reaches zero.

But there is a scenario where one price gap jumps over all of them at once — a weekend move in crypto, an overnight move in equities. Between two prices there simply was no price, and a stop has nothing to trigger on. The account can then end at zero and the kverk “burns out”: it stops trading and stops learning, and its track record closes permanently. In the laboratory the capital is clipped at zero, never below; with a real broker, however, a debt could arise.

Past performance does not guarantee future results

That applies to simulated past performance too. A kverk that did well for ninety days may lose for the next ninety. The Score, the leaderboard and the unseen-period exam promise nothing — they describe what has already happened.

A model that learns from its own trades can overfit: learn the accidental regularities of one particular period instead of the principle. It can also meet a market regime it has never seen — the next crisis will not look like the last one. That is why an exam on an unseen period is a mandatory part of every training, and why even the best exam verdict is a statement about one unseen stretch, not about the future.

The “Why” card on each trade shows which inputs contributed to the decision. It is an honest approximate reconstruction, not a reason in the sense in which you have reasons. A neural network has no reasons.

Data can be delayed or missing

Market data comes from external providers. It can be delayed, incomplete or temporarily unavailable. During an outage kverks continue with fewer senses, and the journal and the track record carry a note — we never rewrite a curve after the fact. The current state is on the Status page.

If you take the result elsewhere

The upcoming desktop application will let you run a trained kverk against your own broker with your own money. That is entirely your decision and entirely your responsibility.

Real trading in financial instruments, especially with leverage, carries the risk of losing your entire invested capital; with some instruments a loss can exceed the amount deposited. A model trained in a simulation will also meet things the simulation knows nothing about: real latency, real fills, outages, taxes, and your own psychology. Never invest money you cannot afford to lose, and consider consulting a licensed adviser before taking such a step.

The Daily Dilemma and the Story of the Day

The Daily Dilemma is a game about knowing yourself, not a trading signal: the question concerns a historical situation and other users' answers are a statistic. The Story of the Day describes what happened yesterday — it never contains a view of what comes next. Neither is a recommendation to trade.

In one sentence

Simulated results. Not a promise of returns. Not investment advice.

Who operates the service

Caroline Technologies s. r. o., Korunní 2569/108, Vinohrady, 101 00 Prague 10, Czech Republic

Company number 29727430, entered in the commercial register kept by the Municipal Court in Prague, file no. C 450846

If you have a question about this warning, write to [email protected] or call +420 607 969 092. We do not answer “what should I buy” questions, though — the operator is not an investment adviser.

Document version

Last updated: 26 August 2026.