Skip to main content
Track System optimization
17 / 23
Search the archiveLibrary
Library

2004issue C071-6

Band construction as a testable trading procedure

A trading band is assembled from a smoothed center and a distance, then restated as one Mechanical trading system. Bandwidth, a volatility link, and a Trend filter become the controls, rather than extra exception rules.

  • A usable center is made by averaging open, high, low, and close, then smoothing that series, because raw bar midpoints are too jumpy.
  • The same band family can be written as a lower-bound entry, a short holding window, and a small profit target under explicit position-size and commission constraints.
  • Narrower bands raise trade count, while profit and profit-per-bar typically peak at different widths, so width is a comfort tradeoff rather than a secret number.
  • A later construction mixes a constant with live range and trend terms so a Trend filter can widen or tighten the band instead of sitting as a separate overlay.
Entries in this reading3 entries
Current trackSystem optimization131 readingsTrend filter184 readingsMechanical trading system322 readings

Center and distance

A trading band is assembled from a center line that tracks the middle of the price series and a distance that places upper and lower boundaries around that center.

Raw bar midpoints are too jumpy for a usable center, so the construction first averages open, high, low, and close and then applies a smoother such as a simple, exponential, weighted, or asymmetrical moving average.

A constant-distance band can be drawn as a fixed percentage offset from a smoothed center, illustrated with a six-period exponential average of average price and a six-percent offset.

The band as a procedure

The same band family can be restated as a fully specified procedure: a lower-bound entry, a short holding window, and a small profit target, then executed under explicit position-size and commission constraints.

Width as a control

Bandwidth is an explicit control: a narrower band raises trade count, while profit and profit-per-bar typically peak at different widths, so the width can be chosen as a comfort tradeoff rather than a one-time secret number.

A single fixed optimal width is impractical because that width can change with the series and with time. Linking distance to a volatility estimate such as range or true range is presented as an automatic alternative to continual System optimization.

When a sample of names is compared at their chosen widths, higher measured volatility lines up with larger recorded profit, but the table does not yield a simple formula that maps volatility onto one universal bandwidth.

NXTL range-band profit versus bandwidth

On Nextel, net profit peaks at a 110 bandwidth setting and then eases, while the same sweep shows fewer trades as the band widens. Width is therefore a knob for dollars versus activity, not a new exception rule. The dollar figures were read from the labeled OptVar1 bars on the article’s NXTL range-band optimization chart.
On Nextel, net profit peaks at a 110 bandwidth setting and then eases, while the same sweep shows fewer trades as the band widens. Width is therefore a knob for dollars versus activity, not a new exception rule. The dollar figures were read from the labeled OptVar1 bars on the article’s NXTL range-band optimization chart.NXTL · Daily

Wealth-Lab run on 1,000 daily bars with $100,000 start, $10,000 per trade, $10 one-way commission, a 1% profit target and a two-day timeout. Win/loss stayed between 78% and 82% across this sweep. Chart x is OptVar1, i.e. master bandwidth in percent. Printed bar labels use three decimal places (35,882.992 at 110 matches the 35882 NXTL figure in the article’s results table).

Trend inside the band

A later construction mixes a constant with live range and trend terms so that a trend reading, defined as the daily change of an optimal tracking filter, can widen or tighten the band instead of sitting as a separate overlay.

Editorial. That trend reading is the Trend filter in this construction. It changes bandwidth from inside the band rather than adding another overlay.

Exception rules as a poor habit

Patching isolated failure patterns with extra exception rules is described as a poor construction habit because a local fix often creates new failures elsewhere. Further work is directed at dynamic weights, extra scaled components, and nonlinear mixing rather than more special cases.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
17 of 23 in the System optimization track
20061-5 pp.Next on System optimizationLayered day-session automation from brackets to rulesDay-session automation can be built in layers: broker-side conditional orders, software that automates selected legs such as exits, or a full mechanical procedure that covers both entry and exit.
All readings on this track · 23 readings
  1. 1984Evaluating mechanical systems beyond net profit
  2. 1987Mechanical systems need more than pretty charts
  3. 1987Evaluating silver moving averages and RSI by scoring rule
  4. 1988Simple systems over optimized complexity
  5. 1988Cycle rule inputs until the parameter search stops
  6. 1990Split-sample tests for stochastic parameter grids
  7. 1993Four-subsystem trading system construction
  8. 1993Randomized price paths for stock system parameters
  9. 1994Mechanical timing with out-of-sample and filter checks
  10. 1995Evaluating asymmetric percent-reversal trading rules
  11. 1996Evaluating a daily volume-strength and new-high new-low timing procedure
  12. 1997Rule-based intraday system design with genetic search
  13. 1997Evolving three-rule systems with genetic templates
  14. 1997Audit the holdout and the exposure before the fitted entries
  15. 2001Inner weights are part of the candlestick signal
  16. 2003Mechanical short-term rules for news-sensitive rallies
  17. 2004Band construction as a testable trading procedure
  18. 2006Layered day-session automation from brackets to rules
  19. 2006Why mechanical rules must be retuned per security
  20. 2006A linear-regression next-bar forecast needs a searchable cutoff
  21. 2015Construct if-then rules, then backtest and optimize
  22. 2015Mechanical system design as a critique of discretion
  23. 2025Joint channel design with a trend filter and stop-loss orders
By fieldSystem design203 readingsBy year2004180 readings
All 131 readings tagged System optimization
Also on System optimization5 readings
1994Mechanical timing with out-of-sample and filter checks2015Optimizing virtual bar size as a trend filter2025Joint channel design with a trend filter and stop-loss orders1984Evaluating mechanical systems beyond net profit1987Mechanical systems need more than pretty charts