How we test

Here is our recipe. No secret sauce. Just a lot of patience.

1. Quote the claim

Word for word, with the type of source. No names. We test the claim, not the person.

2. Write the rules down first

Before we see a single result, we write down the main rules and freeze them with a digital fingerprint (a hash). That's called preregistration. Anything we add later gets flagged as an extra check. (Small print: the timestamp comes from our own computer, so it's our word, not hard proof.)

For Case 01 we publish a redacted copy of the preregistration. Paths and third-party names are removed. It is not the original frozen file, so no hash is shown. New cases going forward will publish the unchanged frozen file with a public hash, written without personal names or absolute paths from the start.

3. Test on real data

Real historical prices. We split them into a practice period and a fresh period (in-sample vs out-of-sample).

Want to look at the same markets on a chart? See the trading toolbox (for looking, not a trade idea).

4. Include costs

Fees, slippage and, for crypto futures, funding. Our cost model is an assumption, not your exact broker bill. A strategy that only wins before costs is like a lemonade stand that forgets to pay for lemons.

5. Compare with random entries

Each trade gets random twins with the same stop and target. If the rule can't beat chance, it's not an edge.

6. Give a plain verdict

HOLDS UP The claim holds up in our test, after costs and after our checks for lucky flukes.

FAILS AFTER COSTS Before costs it was zero or better and it beat random entries, in both periods. After costs: a loss.

FAILED OUR TEST Every other failure: negative before costs, no better than random, or the claim itself failed.

Grey cards show descriptive numbers (how price behaves) or extra checks. They get no stamp, because they are not a trading rule.

7. Say what we don't know

Every case ends with the limits of the test. Backtests are hypothetical. Real trading, with real nerves, is usually worse.

Education only, not financial advice. No signals. Backtests are hypothetical: real trading can be worse. Past results don't predict future results.