Test a trading idea in plain English

The problem

You have an idea, such as buying a share after a sharp fall and selling once it recovers. To see what it would have done, you need years of daily prices, the trading costs, the dividends, and a fair comparison. Most people stop there, or try it with real money.

And one good-looking result can mislead: it might be a single lucky stretch.

Step by step

1. 1

   Type the idea

   Name the share or fund, what makes it buy, what makes it sell, and how long to test. Your own words are fine.

2. 2

   See the price first

   The button shows what the test costs before anything is spent: 300 tokens, whatever its length.

3. 3

   Answer any question

   If a word could mean two things, the agent asks before anything runs. Questions are free.

4. 4

   Read the result

   A short card says whether the rule beat simply holding, what the starting amount became, and the worst fall on the way.

5. 5

   Go further, or stop

   Run Prove it to see whether luck could explain it, or press "Track it with pretend money" to follow it from today.

A worked example from a stored test

A new account's own test, stored on 7 Oct 2026 (the result the beginner card's tests use). Its owner typed: "RSI strategy on Tesla, buy under 30, sell over 70."

Read it honestly: Only 2 round-trip trades in 3 years, so one decision can swing this result. A rule that traded so rarely could have landed on either side of holding.

A stored test the day before went the other way: a golden cross on a FTSE 100 fund ended at £1,564 against £1,764 for simply holding, from £1,000, 6 Oct 2021 to 5 Oct 2026. Both are shown because both happened; neither says what a rule does next.

What it cannot tell you

• What happens next. A test replays one stretch of the past.
• Whether the result was luck. A rule found by trying many ideas can look good by chance alone, which is what checking a result for luck is for.
• Only shares and funds that trade today are in the data. A company that failed, was taken over or was delisted is not in it. So a test that picks from a list of shares is flattered, by an amount the test cannot measure.
• What your own broker would have charged, unless you chose it in the composer. Otherwise the test uses standard costs: Each order in a test pays 0.05% of its value in trading costs (4.5 basis points of slippage and 0.5 of commission), so a buy and a sell together cost 0.1%, or 10 basis points.

Try it

Each one opens the Strategy Agent with the words typed in. Nothing is sent until you press its button, and the price shows first.