Do AI trading bots actually make money?
Most do not, and nobody who says otherwise in a headline can prove it. Trading is negative-sum after costs, so any tool has to beat both the market and its own fees. The size of the number a tool advertises tells you nothing. Whether you can see the losses tells you a lot.
The short answer.
No tool can promise you a profit, and in the UK it is illegal for one to try. Under the FCA rules on financial promotions, a firm cannot advertise guaranteed returns or present past performance as an indicator of future results. If you see a tool advertising a fixed monthly percentage, you have learned something about the seller, not the software.
What automation does well is remove the parts of trading that people reliably get wrong: hesitating on an exit, sizing a position by feel, skipping the stop because it feels close, and revenge-trading after a loss. That is a real edge. It is a behavioural edge, not a forecast.
How to judge a performance claim.
Five questions dispose of most of them.
Is it live or backtested?
A backtest is a strategy fitted to data it has already seen. It costs nothing to produce a good-looking one. Live results, on a real account, with real fills and real slippage, are the only kind that mean anything.
Are the losing trades shown?
A results page with no red on it is a selection, not a record. Every real strategy has trades that went against it.
Is the period long enough?
Six weeks of a rising market proves nothing. You want to see the tool through a drawdown. That is where the difference between discipline and luck shows.
What are the costs?
Subscription, spread, commission and slippage all come out of the same number. A tool that beats the market by two percent and charges three percent has lost you money.
Who is paid, and when?
A flat monthly fee is paid whether you win or lose. A fee attached to realised profit is not. That does not make one honest and the other dishonest. It does tell you where the incentive points.
What blank does about it.
blank does not publish a return figure. There are two reasons, and the second matters more.
The first is regulatory. blank is not FCA-authorised to give investment advice, so it does not advertise performance and does not present itself as a way to get rich. It is a tool that trades on your instruction, on your own broker account, with your risk settings.
The second is that a single number would mislead anyway. Results depend on when you started, how much you committed, what the market did, and which risk settings you chose. What blank gives you instead is the full journal: every trade it took, every trade it closed, the reasoning it wrote at the time, and the losses alongside the wins. You can read it in the app before you go live, because practice mode runs the same engine on real market data with virtual money.
What to do instead of believing a number.
Run it yourself, on virtual money, and judge the reasoning rather than the result. A week of watching an advisor explain what it is doing tells you more than any performance page. You find out whether its logic is something you would have agreed with. If it is not, no backtested return will help when it is your money.
Common questions.
Do AI trading bots make money?
Some do and most do not, and no honest tool will promise it. Trading is negative-sum after costs, so a tool has to beat the market and its own fees. The reliable benefit of automation is behavioural: it removes hesitation, emotional sizing and skipped stops. Treat any advertised return figure with suspicion, especially a backtested one.
Are AI trading bot results usually real?
Often they are backtests rather than live results. A backtest is fitted to data the strategy has already seen and costs nothing to make look good. Ask whether the record is live, on a real account, over a period that includes a drawdown, and whether the losing trades are shown.
Why does blank not publish returns?
Two reasons. blank is not FCA-authorised to give investment advice, so it does not advertise performance. And a single figure would mislead anyway, because results depend on your start date, your capital and your risk settings. blank publishes a public ledger instead: two matched fleets of practice accounts, each started with £1,000, updated every trading day against the S&P 500 and the FTSE 100, losses included. In the app you see your own full trade journal.
Is there a way to test it without risking money?
Yes, and the first seven days cost nothing. blank's practice mode runs the same engine on real market data with a virtual balance, so you can watch it decide and read its reasoning before anything goes live.
Judge the reasoning, not the marketing.
Practice mode runs the real engine on real market data with virtual money. Read what it decides and why.
join the beta