Clients Are Asking About AI Trading. Here Is What the Profession Should Be Saying.

turned on monitoring screen

If you work anywhere near personal finance, you will have had the question this year: should I be letting AI trade for me? It arrives from clients who have seen the adverts, colleagues who have downloaded the apps, and occasionally from the one relative at a barbecue who has found a bot with a suspiciously smooth performance chart. The honest professional answer is more nuanced than yes or no, and it is worth having ready.

What the question actually means

When a client says AI trading, they usually mean one of three very different products: software that suggests trades for a human to approve, automation that executes fixed rules quickly, or genuinely adaptive systems that adjust their own strategy. The retail market is overwhelmingly the first two dressed in the language of the third. That distinction matters professionally, because the risk conversation is different for a research tool than for software with discretion over live positions.

The professional framing

Three points put the subject on solid ground with clients. First, the software layer is unregulated; the protection sits with the executing broker, so an FCA-authorised venue is non-negotiable whatever sits on top. Second, vendor backtests are marketing, not evidence – the only meaningful number is forward performance in a live funded account, net of all costs, and vendors who publish it are rare enough to be notable. Third, total cost of ownership decides most outcomes: subscription plus spreads plus financing regularly exceeds the fee load of a conventional managed alternative, which is the honest benchmark.

Independent testing helps here. Reviews that fund real accounts and measure AI stock trading tools on live results rather than vendor claims give the conversation a factual anchor, and the pattern in that testing is consistent: a small number of tools are useful within narrow limits, and the gap between the best and the rest is wide.

Where the risk really sits

The failure mode professionals should worry about is rarely the algorithm itself; it is position sizing. Clients treat automated systems as safer than their own judgement and allocate accordingly, sometimes moving pension-sized sums into products designed for satellite allocations. The steadying advice is old-fashioned: treat any automated strategy as an experiment sized for tuition, demand a probation period against a passive benchmark, and be unsentimental at review.

For those who want the homework done independently, research houses such as The Investors Centre publish plain-English, live-money testing of these platforms – a useful thing to be able to point a curious client towards before their enthusiasm outruns their allocation policy.

A checklist for the client meeting

For advisers and finance professionals who want something concrete to hand over, a five-line checklist covers the ground. One: is the executing broker FCA-authorised, and have you checked the register entry matches the firm reference number on the website? Two: has the vendor shown live, funded, forward performance – not a backtest – and over what period? Three: what is the all-in annual cost at your intended allocation, including spreads and financing, expressed in pounds? Four: what is the exit process, and are there withdrawal fees or lock-ins? Five: what specific slice of the portfolio is this for, and what result at review would trigger switching it off? A client who can answer all five is making an informed choice, whatever they decide.

It is also worth naming the behavioural risk that automation introduces: over-trust. The same client who would monitor a human manager monthly will leave software unattended for a year, because checking feels like meddling with the machine. Scheduled reviews – in the diary, against a passive benchmark – are the antidote, and suggesting them is squarely within the profession’s value.

AI will keep moving into retail investing, and the profession does its best work here as translator: separating capability from marketing, and making sure the exciting part of the conversation never gets ahead of the suitability part.