Backtest evidence

Do double tops and double bottoms actually reverse?

Do Double Tops and Bottoms Reverse? 2 Years of USD/JPY Tested

Finding

It depended on the timeframe. With five bars on each side of a pivot and entry after a neckline close, the share moving in the reversal direction ten bars later was 49.71% in 2024 and 50.15% in 2025 on M15, 56.45% and 51.85% on H1, and 46.88% and 31.43% on H4 — only H1 cleared 50% in both years. That did not make the trade stable: H1 net pips changed from +613.3 in 2024 to −562.7 in 2025, and H4 produced only 12 and 13 trades with the same setting. Entry timing also mattered differently by timeframe: on M15, entering when the shape was confirmed returned −589.9 pips in 2024 and −195.4 in 2025, while waiting for the neckline returned +459.3 and +275.1.

Key results

Share moving in the reversal direction ten bars after the neckline
M15 49.71% / 50.15%, H1 56.45% / 51.85%, H4 46.88% / 31.43%
H1 net pips with five-bar pivots
+613.3 in 2024, −562.7 in 2025
Entry timing on M15
at confirmation −589.9 / −195.4 pips; after the neckline +459.3 / +275.1 pips

Scope

  • USDJPY, 1 January to 31 December 2025, repeated on 2024
  • M15, H1 and H4
  • 8 pivot strengths (Swing Period 5 to 75) x 2 entry timings (second pivot confirmed / close beyond neckline) = 16 settings; 192 rows over three timeframes and four windows
  • 0.3 pip spread, 0.1 lots, filled at the close

Method

Measured with Formiq's backtester. A double top trades from the bar whose close falls below the middle low, and a double bottom from the bar whose close rises above the middle high; the exit is the opposite pattern. A pivot is unavailable until all its right-side confirmation bars have closed. No stop or target in the base run.

Limitations

  • The sample is USDJPY from 1 January 2024 through 31 December 2025.
  • Trades fill at the close with a 0.3 pip spread and 0.1-lot size.
  • Bars before each annual window supply warm-up history, and no pivot is used before its right-side confirmation bars have closed.
  • When one bar touches both the target and the failure level, the order cannot be determined, so those cases are classified separately.