RSI on a one-minute chart mostly measures noise

[Amanhasnoname]

The relative strength index on a one-minute chart oscillates wildly. It hits overbought and oversold levels several times an hour, and acting on each signal produces a stream of marginal trades.

The default settings were designed for daily data, and applying them to tiny timeframes changes their meaning. A reading of 70 on a one-minute chart reflects a few pips of movement, not a significant extension.

Where it may help is in context. Extreme readings combined with a key level and a slowing tick rate can flag exhaustion. On its own, it is a coin flip with costs.

Adjusting periods and thresholds improves the fit, but also invites overfitting. Testing across several pairs and periods guards against that.

An alternative is using RSI on a higher timeframe as a bias filter while trading on the lower one. That reduces noise while preserving the information.

Plotting the indicator next to a plain price chart for a week usually makes it obvious how rarely the extremes lead anywhere useful.

Does RSI add value on short charts, and in what role?