Risk controls are limits, not guarantees

A risk control is a programmed decision under stated assumptions. It can reduce the amount or frequency of intended exposure, but it cannot promise a maximum realised loss. CryptoMetric AI checks proposed entries against configured rules and records refusals. It does not insure the user, guarantee an exit price or remove the underlying risk of crypto markets.

What the controls do

The current pipeline requires a finite stop before approving an entry, sizes the position from the account balance and stop distance, caps intended risk per trade, limits single-asset exposure and blocks entries after configured daily or consecutive losses. It can also reject entries during high measured volatility and require a minimum reward-to-risk distance.

These checks are useful because they turn a user's configuration into consistent rules. They are not personalised risk advice. The defaults are templates; a percentage that is technically accepted by the software may still be too high for a particular user.

Why realised loss can be larger

Stops in this system are software checks, not guaranteed resting orders at the exchange. Price can gap past a stop. Slippage, fees, thin liquidity, delayed data, a worker or network outage, an exchange incident, credential failure or a software defect can delay or change an exit. A configured 1% risk input is therefore an intended sizing bound at entry, not a promise that realised loss cannot exceed 1%.

A daily loss limit and circuit breaker block new exposure after their conditions are observed. They do not undo losses already incurred, and they cannot guarantee that an open position will close. An emergency halt can also block automated closing actions, so users need independent exchange access and a rehearsed manual response.

Automation changes operational risk

Automation can apply a rule consistently, but it can also repeat an incorrect configuration quickly. Monitoring must cover data freshness, bot status, credentials and the exchange account itself. The user decides whether to start, pause or stop the bot and remains responsible for supervising the connected account.

Practical boundaries

• Commit only funds whose complete loss you can bear.
• Use paper mode first and treat results as tests, not forecasts.
• Set smaller limits than the maximum the software permits.
• Keep independent exchange access and know how to close a position manually.
• Stop the automation when behaviour, data or account state is not understood.

The safest reading of every control is narrow: it attempts the described check when the system is operating normally. It does not transfer market, exchange, configuration or operational risk away from the user.