Open beta · first 7 days free, no commission · then 50% off (£5 a trading week or £17.50 a month) until 1 September · start free
blank
hard rails, not suggestions

Can an AI set stops and manage risk?

Yes, and the automatic stop loss is the least interesting part of the answer. Any script can attach a stop. What decides whether an account survives a bad month is how large each position was allowed to be in the first place, how exposure changes when volatility rises, and whether the stop can be overridden by the same reasoning that got the trade wrong. Those are the questions to ask a risk engine.

join the beta the full safety answer →

The four layers that matter.

01

Position sizing

The first and largest decision. blank sizes with a Kelly fraction capped well below the theoretical optimum, so no single idea can take an outsized share of the account regardless of how confident the reasoning is.

02

The stop itself

A stop placed at a round number is a stop placed where everyone else's is. blank uses ATR-based stops, set from the instrument's own recent volatility, so a quiet stock and a violent one do not get the same distance.

03

Exposure by regime

Risk is not constant. When the market regime turns volatile, blank reduces total exposure rather than continuing to size positions as though nothing changed.

04

Rails the advisor cannot cross

The important one. Limits enforced outside the reasoning loop cannot be talked out of by the reasoning loop. Cash only: no leverage, no CFDs, no shorting, and a separate price monitor watching the stops.

Why an automated stop is better than yours.

Not because it is cleverer. Because it is not there when the price approaches it.

The most common way a retail account is damaged is not a bad entry. It is a stop that was moved. Price comes close, the position is already down, and moving the stop a little wider feels like patience rather than panic. Repeat that three times and a small planned loss becomes the loss that matters. An automated stop is executed by something with no feelings about the position and no memory of what it paid.

What automation does not protect you from.

Gaps. A stop is an instruction to sell when a price is reached, and if a stock opens far below that price after news, the fill happens where the market is, not where the stop was. No tool can prevent that, and any tool implying otherwise is misleading you.

It also does not protect you from committing more money than you can afford to lose. Risk management sizes the bets inside the account. It has nothing to say about how much you put in the account, and that decision remains entirely yours.

Questions people actually ask.

Can an AI set stop losses automatically?

Yes. blank arms a stop at entry on every position, sized from the instrument's own recent volatility using ATR rather than a fixed percentage, and a separate price monitor watches them.

What else does an AI risk engine do?

The stop is the smallest part. Position sizing decides how much damage any single idea can do, exposure limits reduce total risk when the market regime turns volatile, and hard rails outside the reasoning loop stop the advisor overriding its own limits. blank is cash only: no leverage, no CFDs, no shorting.

Can the AI move its own stop?

No. The risk limits are enforced outside the decision loop, which is the point. A limit the reasoning can revise is a suggestion, not a limit.

Does a stop loss guarantee my loss is capped?

No. If a stock gaps down through the stop after news, the fill happens where the market opens, not at the stop price. No tool can prevent that.

Limits that hold when it matters.

Kelly-capped sizing, ATR stops, regime-aware exposure, cash only. Watch them work in practice mode.

join the beta