How to define a trading hypothesis

A useful hypothesis describes a condition, an expected behaviour, and a reason it might exist. ‘An RSI strategy makes money’ is not a hypothesis. ‘After a sharp move during a defined session, a measured pullback may continue when volatility and trend conditions agree’ is a question that can be specified and challenged.

The explanation does not need to be certain. It needs to be clear enough that another person can understand what would support or weaken it.

Define the instruments, timeframe, date range, entry timing, exit timing, costs, risk model, and exclusions before looking at the final curve. These boundaries stop the experiment from expanding every time an inconvenient result appears.

A research process becomes more honest when it includes a stopping rule. If the idea fails out of sample, requires too many exceptions, or only works at one fragile parameter value, record that conclusion instead of endlessly searching for a rescue.

Good hypotheses can be rejected. A rejected idea is useful evidence when the test design was fair.