The strategies

  • Keltner Channel Breakout (KC-01): a 20-day EMA midline with bands at 2× the 20-day ATR; goes long when price closes above the upper band and exits when it falls back below the midline.
  • Bollinger Band Reversion (MR-02): goes long when price touches the lower Bollinger Band and exits at the middle band, with a bandwidth filter.
  • Bollinger-Keltner Squeeze (COMBO-01): waits until the Bollinger Bands contract inside the Keltner Channel, then trades in the direction of the move when the squeeze releases.

All three run long-only on daily data with walk-forward optimization and no trading costs; the figures below cover the last five years.

Where breakouts won

The Keltner breakout had the higher Sharpe ratio on US stocks, gold, silver, bitcoin and ether. Crypto stands out: 14.7% a year on bitcoin (buy-and-hold 7.2%) and 13.3% on ether, where holding lost 8.8% a year. On SPY it made 4.9% a year with a −8.5% maximum drawdown and a Sharpe ratio of 0.75 — far less return than holding (13.2%), but a third of the drawdown.

Where reversion won

Bollinger reversion had the higher Sharpe ratio on oil, broad commodities, agriculture, EUR/USD and natural gas — markets that spent much of the period moving sideways. Its best result was oil: 11.5% a year with a Sharpe ratio of 0.91 and a −13% drawdown, while being in the market only 12% of the time. On EUR/USD it was the only one of the three that made money (1.3% a year, maximum drawdown −2.6%).

The squeeze: one good market

Combining both ideas did not combine their strengths. The squeeze strategy traded about twice as often as the other two and had the best Sharpe ratio of the three on only one market, gold (10.6% a year, Sharpe 0.99). On US stocks it lost 1.4% a year.

How sure can we be?

Not very. The 90% bootstrap ranges are wide: on SPY, for example, the breakout's annual return ranges from 0.5% to 10.8% and the reversion strategy's from −1.7% to 8.1%. After adjusting for the number of strategies we test, none of the 30 results reaches a Deflated Sharpe of 95%; the best is the squeeze on gold at 63%. The market split is the more useful finding than any single number.

Takeaways

  • Breakout and reversion are not better or worse — they suit different markets. Trending assets (stocks, metals, crypto) favored the breakout; range-bound ones (oil, agriculture, currencies) favored reversion.
  • Mean reversion with Bollinger Bands is in the market rarely (6–15% of days), so its returns are small even when its Sharpe ratio is good.
  • Neither beat buy-and-hold on return in the strong bull markets; their value is lower drawdowns.
  • Check the market before picking the strategy, and use the heatmap page to see this pattern across all strategies.