How Much of a One-Minute Candle Is Just Noise?

[Shadow Trader]

At the one-minute scale, a large share of price movement is noise: random fluctuation from order arrival, quote updates, and liquidity probing that carries little information about where price will be later. The signal, meaning the persistent component, is small by comparison.

A useful illustration is the efficiency ratio, which divides net displacement over a window by the total path length traveled. Values near zero describe choppy back-and-forth movement, while values near one describe a clean directional run. On one-minute data, typical readings tend to sit low, and they rise as the window lengthens.

Random walks also scale in a particular way: the typical size of net movement grows roughly with the square root of time, not linearly. Doubling the holding period does not double the expected range, it multiplies it by around 1.4. Targets and stops that ignore this scaling often end up too ambitious for the horizon.

The implication for scalpers is that any pattern spotted on a single one-minute candle has a high chance of being noise. Edges, when they exist, usually require confirmation across many observations, and they need to exceed costs by a comfortable margin. Testing how a pattern performs against randomly shifted entry times is a simple sanity check.

[FXS HongKong Scalper]

The efficiency ratio is a nice way to show it. We measured it on EURUSD M1 for the Asian session and found values mostly under 0.2, which means price travels five times the net distance. In the London open, values around 0.3 to 0.4 were common for the first hour.

That's a useful argument for not using M1 signals in Asia at all. The noise share is too high, and the spread is a bigger fraction of the moves. We use M5 or M15 for setups and only drop to M1 for timing in active sessions.