A backtest you can audit without writing code
A backtest you can audit is a list of rules another person can replay, plus the year the rules lost money. This guide uses a twenty-day average cross on one liquid stock and shows the 2022 stretch where the rule kept buying into a falling market.
The rule, written so someone else can run it
Buy when the close finishes above the twenty-day average after a close below it. Sell when the close finishes back under the average. One stock, NML, no leverage, no overnight claim about a different symbol. Chartloom can show you the average. The test is whether you can list the trades without the vendor's adjective "smart" anywhere in the sentence.
The year it lost
In the 2022 stretch we used for the example, the average kept sloping down and the rule bought three times into that slope. Each buy was "correct" by the rule and lost money over the next twenty sessions. A backtest page that shows only the full-period gain has hidden that year. Write the losing stretch on the page with the dates. If the software will not show the trades, you do not have a test. You have a summary.
What this is not
This is not a strategy to fund. It is one rule on one liquid stock, used so a reader can see what an auditable test contains. A test that cannot name the year it failed is advertising. The trading-tool reviews on this site are held to that standard, which is why a pretty equity curve with no trade list does not earn the higher score.
The losing stretch stays on the page
The rule is a close above the twenty-day average, then a close back under. In the 2022 example it bought three times into a falling average and lost over the next twenty sessions each time. Those trades stay on the page next to any full-period gain. No trade list, no test. No losing year, no test. Chartloom can show the average. It cannot supply the honesty if you crop the chart.
This is not a strategy to fund. It is the standard the trading reviews use. A pretty curve with no trades does not get the higher score.
The three buys in the falling stretch
The rule: buy when the close finishes above the twenty-day average after a close below it, sell when the close finishes back under. One symbol, NML, no leverage. In the 2022 example the average sloped down. The rule bought on 3 February, 14 March, and 2 May. Each buy was correct by the rule. Each was sold within twenty sessions at a loss. The page that shows only a full-period gain has hidden those three. Write the dates and the direction. If the software will not list the trades, you do not have a test. Chartloom can show the average on those dates. It cannot supply the honesty if you crop the chart.
This is not a strategy to fund. It is the standard the trading reviews use. A pretty curve with no trades does not get the higher score. Mintcanopy will not add a second symbol to make the example look like a portfolio. One liquid stock is enough to see whether the test can be repeated by someone who was not in the room.
What you owe the person replaying it
The person replaying the test needs the rule in one sentence, the symbol, the three buy dates, and the fact that each lost over the next twenty sessions. They do not need the word smart. They do not need a second symbol added so the curve looks like a book. Chartloom can show the average on those dates. It cannot force the losing trades to stay on the page. Mintcanopy holds trading reviews to this page. A curve with no trades does not get the higher score, however smooth the line in the vendor's deck.
Dates a second person can find
The rule buys a close back above the twenty-day average and sells a close back under, on NML, with no leverage. In the 2022 stretch it bought on 3 February, 14 March, and 2 May, and each buy lost over the following twenty sessions. A page that shows only the full-period gain has hidden those three. Mintcanopy does not add a second symbol to make the curve look like a book. Chartloom can display the average. It cannot force the losing trades to remain. No trade list, no test.
Three dates, then stop
3 February, 14 March, and 2 May, each a buy that followed the twenty-day rule on NML and lost over the next twenty sessions. That is enough to show what an auditable test contains. Mintcanopy does not add a second ticker to improve the curve. The person replaying it needs the rule in one sentence and those dates. They do not need the word smart. No list of trades, no test.
Hand the dates over
Another person should be able to take NML, a twenty-day average, and the buys on 3 February, 14 March, and 2 May, and see three losses over the following twenty sessions without calling you. If they need your memory of the demo, it is not a test. Mintcanopy stops at one liquid stock on purpose. A second symbol would start to look like a book and would hide whether the rule was even written down. Chartloom can show the average on those days. Cropping the losing trades out of the picture is the failure mode this guide exists to name. The trading reviews inherit the standard.