Backtest with your own amount and your UK broker's fees

Many people start with £50 or £100, but every backtest ran on 100,000 with fractional shares and percentage costs, and a "£100" in your prompt was ignored. From 29 September you can say how much you would put in and with which broker, for example "Golden cross on SPY, I have £100 on Trading 212", and the backtest runs at that amount with that broker's dealing charges, FX fees, platform charge and share rules. Fixed fees that barely register on 100,000 can swamp a small account, and the result now shows it.

Five UK broker presets. Trading 212, Freetrade (Basic), Hargreaves Lansdown, Interactive Brokers UK (fixed pricing) and Vanguard Investor UK, each dated 28 September 2026 and linked to the broker's published charges. Anything we couldn't confirm on the broker's own pages is flagged on the card, and a market a preset doesn't cover is refused rather than guessed: Vanguard Investor UK only sells Vanguard's own funds, for example, so a Vanguard run on Apple is refused, free. You can also enter your own fees. Slippage stays at 1 basis point for every preset. An amount in another currency is converted at the ECB reference rate of 28 September 2026; exchange-rate moves over the test aren't modelled, and the card says so.

Whole shares and fixed fees. With whole shares, each order is rounded down and the leftover cash earns the risk-free rate. If your amount can't buy one share of something the strategy wants, it isn't bought, and the card names it with the amount you would need. Top-ups whose fixed fee would be more than 5% of the trade are skipped, so you don't pay £6.95 to reinvest £4. On these runs the first day's charges count in the return and the drawdown.

The minimum sensible amount. Each run is repeated at £50, £100, £250, £500, £1,000, £2,500, £5,000, £10,000, £25,000 and £100,000 with the same rules, prices and fees. The card gives the smallest amount from which every larger one comes within 0.5 percentage points a year of the £100,000 result and can buy every position, and "How was this calculated?" shows every step. In our tests, on a monthly momentum rotation across five FTSE 100 shares, Hargreaves Lansdown's dealing charges took about 20 percentage points a year off the return at £100, about 9 at £1,000 and 0.19 at £100,000. A golden cross on SPY lost 0.17 percentage points a year to FX fees at Trading 212 and 1.10 at Freetrade; at Hargreaves Lansdown, £100 couldn't buy one SPY share, so that run held cash and said so.

A small change to every result. Every backtest, including the standard 100,000 test, now skips rebalancing top-ups and trims smaller than 0.25% of the portfolio. Results move very slightly: the five-year golden cross on SPY went from 11.03% to 11.02% a year, with 17 fills instead of 56, because the tiny daily top-ups of interest are gone. Old cached results are not reused.

Beginner and quant cards now differ. A bug meant every result card showed the beginner layout. Beginners now see the verdict first, a line about their amount and broker, the result in their own money, and how far their amount fell at its worst and how long it took to recover. Before a beginner's first run, the agent asks once how much and with which broker; it's one click, or you can skip it and run the standard test. Quants keep the 100,000 default and see the key numbers first, with a form to change the amount, broker or fees, "Reproduce this run" and the count of their tests of the idea up front.

"I have £100, what should I buy?" We can't tell you, and the agent now says so plainly: Quantis Trade tests investing ideas on past prices, it doesn't recommend investments, and nothing here is financial advice. It offers three example tests sized to your amount and links to our new risk warning page.

Not covered yet. Goal search candidates, the evidence explorer and Deep Dive still run the standard test, and paper trading uses the standard costs rather than your broker's. Regular monthly contributions aren't modelled; if your prompt mentions one, the agent asks how you want it treated rather than guessing.