Start here
This site is the public record of a testing machine. Every week we take one popular trading claim, run it through a backtesting engine that charges real spreads, real commissions and realistic fills, and publish the verdict — DIED,SURVIVED, or survived only in fantasy costs.
How to read a verdict
- The claim is stated precisely enough to be testable, and the test rules are fixed before the test runs. No hindsight, no tuning until it works.
- Everything is compared against the boring alternative — usually just owning the thing. In a rising market almost any strategy makes money, so "it was profitable" proves nothing on its own.
- Costs are charged in full, including the per-order commission minimum that quietly destroys small accounts, and the fact that a real account can only hold a few positions at once.
The archive so far
- RSI below 30: we tested buy-the-dip 2,473 times — DIED
The most repeated indicator rule in retail trading, tested on 2,473 real signals across 369 stocks with real costs. It died, and not for the obvious reason. - The golden cross: we tested it on 337 stocks — DIED
The most famous chart signal there is, tested on 661 real crossings. It lost to buying and holding, and then it lost to random entry dates.
Two things worth knowing before you read anything else
A backtest's trade count is a claim about your account. If a strategy fires two thousand signals and you can hold three positions at a time, the number on the screen is not the number you would have earned.
A rule that earns roughly what its market exposure is worth has not timed anything. Always ask what percentage of the time a strategy was actually invested.