Double tops and bottoms on short timeframes
Posted: Wed Oct 28, 2026 4:18 am
Double tops and bottoms appear on every timeframe, and on a one-minute chart they appear constantly. A pair of similar highs separated by a dip is among the most common shapes in price action.
The pattern's claimed meaning, a reversal, is where the trouble starts. Because it appears so often, the information content per instance is low. Many double tops are followed by a third top, and a breakout.
Definitions affect everything. How close must the two highs be, within one pip, three, five? How deep must the dip between them be? How long may they be apart? Different answers produce wildly different statistics.
A fair test fixes those parameters and counts. Does price break the neckline and travel a target distance before returning? Often the answer is close to a coin flip, with the tilt depending on session, trend context and volatility.
Combined with a trend filter, such as taking only double tops after a strong run-up, the pattern might add something. Even then, costs must be included.
A hint from tests: the pattern improves when the two peaks are separated by at least 15 to 20 candles and the second peak is accompanied by a visible loss of momentum, like a lower RSI reading or a smaller range. Without those conditions, it's mostly noise.
What definition of double top has been tested, and did it beat a coin flip after costs?
The pattern's claimed meaning, a reversal, is where the trouble starts. Because it appears so often, the information content per instance is low. Many double tops are followed by a third top, and a breakout.
Definitions affect everything. How close must the two highs be, within one pip, three, five? How deep must the dip between them be? How long may they be apart? Different answers produce wildly different statistics.
A fair test fixes those parameters and counts. Does price break the neckline and travel a target distance before returning? Often the answer is close to a coin flip, with the tilt depending on session, trend context and volatility.
Combined with a trend filter, such as taking only double tops after a strong run-up, the pattern might add something. Even then, costs must be included.
A hint from tests: the pattern improves when the two peaks are separated by at least 15 to 20 candles and the second peak is accompanied by a visible loss of momentum, like a lower RSI reading or a smaller range. Without those conditions, it's mostly noise.
What definition of double top has been tested, and did it beat a coin flip after costs?