Out-of-sample test
Checking a pattern on data that wasn't used to find it. Patterns found by searching are partly luck, and luck doesn't repeat on new data, so this is where most market legends die.
In plain English
Split the data by date. Look for the effect in the earlier years, fix the rules, then test exactly those rules on the later years. If the effect was real, it should still be there. The order matters: an effect that appears only in the later data, and wasn't there when you looked first, is a reason to collect more data, not proof.
Uncle Gary explains
Uncle Gary's rule 'if Tokyo is up, Wall Street goes up' worked on 55.5% of days from 2015 to 2022. On the later years it worked on 49.9%, which is what a coin does.
Why it matters for your money
Any backtest can be tuned to look good on the data it was built on. Only results on data the rules have never seen tell you anything about the future.
Where we tested it
Related terms
All glossary terms. Educational content, not financial advice.