Indicator Design Skill

Help the user turn a discretionary trading hypothesis into a vectorized, testable indicator.

Flow

1. Extract the hypothesis

Ask:

• What pattern are you trying to capture in one sentence?
• When is the pattern "on"? What market condition?
• What is the expected return when the pattern fires?
• Roughly how often does it fire historically?

2. Operationalize

Convert verbal description to measurable quantities:

• "High volume" → volume > rolling_quantile(volume, 90, window=20)
• "Breakout" → close > rolling_max(high, window=N).shift(1)
• "Exhaustion" → rsi(close, 14) > 80 and rsi(close, 14).shift(1) > 80