Skip to main content
Track Breakout system
6 / 23
Search the archiveLibrary
Library

1998issue C041-8

A noise-offset breakout judged after walk-forward re-estimation

Treat a currency breakout as one procedure: a noise offset, locked lookback extrema, and market orders. Judge those same rules only by how they behave after each walk-forward re-estimate.

  • The procedure buys at market when price clears the lookback high by at least the noise offset, and sells at market when price clears the lookback low by that same offset.
  • While short, the lookback high is not allowed to rise as the window rolls; while long, the lookback low is not allowed to fall, so the unused extreme cannot move against the open position.
  • Walk-forward evaluation selects parameters on a multi-year in-sample window, applies those rules unchanged to the next unused year, then advances the window by one year and repeats.
  • The archive presents the method as a public-domain trend-following breakout and states that more years of earlier out-of-sample testing would be needed before treating it as ready for live use.
Entries in this reading2 entries
Current trackBreakout system74 readingsWalk-forward analysis100 readings

A public-domain trend-following breakout

The archive presents the method as a public-domain trend-following breakout for a currency futures market that tends to trend. It argues that secrecy is not what makes a procedure useful.

A continuous back-adjusted futures series is used so quarterly contract rolls do not break the price history. The archive still notes that actual roll execution cannot be reproduced exactly.

Noise offset and locked extrema

The procedure buys at market when price clears the lookback high by at least the noise offset, and sells at market when price clears the lookback low by at least that same offset.

Daily prices contain enough random movement that an unfiltered breakout can generate false entries and exits. A noise offset is added to require a defined excess beyond the lookback extreme.

While short, the lookback high is not allowed to rise as the window rolls. While long, the lookback low is not allowed to fall. The lookback extrema are constrained so they cannot move against the open position.

Walk-forward evaluation of the same rules

Parameter choice on historical data is treated as in-sample fitting. The evaluation question is how the same rules behave on later data that was not used to select those parameters.

Walk-forward evaluation selects parameters on a multi-year window, records the next unused year, then advances the window by one year and repeats. The chosen parameters are applied unchanged to that next out-of-sample year before the window moves on.

The archive states that more years of earlier out-of-sample testing would be needed before treating the procedure as ready for live use.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
6 of 23 in the Breakout system track
19981-6 pp.Next on Breakout systemGating a weekly average crossover with stored support and resistanceA moving-average-crossover is a weekly close crossing a simple moving average as the first, unambiguous trend signal, with the lookback period as its only fitted input.
All readings on this track · 23 readings
  1. 1995Range breakout rules with an expansion filter and moving-average exits
  2. 1995Write a weekly breakout as one parameterized entry and exit
  3. 1995Combining a trend rule, a breakout trigger, and a seasonal filter
  4. 1996Constructing a two-bar clearance breakout from a twenty-session exponential average
  5. 1996Two-bar exponential-average breakout as setup, stop, and flatten
  6. 1998A noise-offset breakout judged after walk-forward re-estimation
  7. 1998Gating a weekly average crossover with stored support and resistance
  8. 1998Moving-average candidates gated by support and resistance
  9. 2000Constructing next-close envelope targets for breakout stops
  10. 2001February soybean high breakout and June trailing stop
  11. 2005Box-and-breakout states written as ordered entry and exit rules
  12. 2007Match trend and breakout rules to the market condition
  13. 2010How a JM internal band becomes long and short entry and exit rules
  14. 2013Constructing a three-average trend-aligned breakout system
  15. 2016Volume-confirmed breakout entry rules
  16. 2017How to construct exponential standard deviation bands
  17. 2017Four-day green candle breakout as one swing procedure
  18. 2018Constructing inverse ETF breakouts above a 200-day average
  19. 2018Evaluating trend, breakout, and regression rules by average robustness
  20. 2019A crypto pair breakout after a sideways range
  21. 2019Next-session breakout rules after a high-volume close
  22. 2020Altcoin dual-stop breakout with a timed exit
  23. 2020Critiquing required stops in mechanical breakout systems
By fieldEntry and exit rules231 readingsBy year199860 readings
All 74 readings tagged Breakout system
Also on Breakout system5 readings
2000Fixed-horizon walk-forward tests of first-print breakouts2018Critiquing breakout systems through robustness tests1988Early entry as the session switch for an opening-range breakout1989The bull hook as a next-session opening-range breakout filter1989Evaluate an always-in ATR breakout as one procedure