2015issue C0660-61
Optimizing virtual bar size as a trend filter
A virtual bar of type time, range, volume, or Renko can be resized by an optimizer at the same time as other algorithm parameters. When a simple rule set stays fixed, bar type and size act as a zero-lag filter, and advancing-price size can be set independently of falling-price size.
- Time, range, volume, tick, and Renko bars can host trades inside a larger clock bar, so bar type is a testable parameter rather than a display default.
- Candidate range-bar increments run from 0.02 to 100, counted as 9999 sizes, and average true range does not collapse that search because the ATR reading changes with the clock interval used to measure it.
- Dual-direction bar size sets advancing and falling increments separately so one procedure can sample two volatility environments.
- The narrower procedure holds a simple rule fixed, treats optimized bar type and size as a zero-lag filter, and retunes only the bars after a gradual fade.
The bar as a trend filter
A trend filter is a testable gate on whether price action is aligned enough to act. Here that gate is implemented first by how bars are built rather than by a trailing overlay.
Virtual bars can be specified as time, range, volume, or Renko, exposed to every platform function, and resized by an optimizer at the same time as other algorithm parameters.
Bar type as a testable parameter
Time, range, volume, tick, and Renko bars can all host trades inside a larger clock bar, so bar type is a testable parameter rather than a display default.
Different bar constructions absorb noise differently, and different algorithms tolerate the residual noise differently. A Renko chart is a brick-style bar construct that can be generated as a virtual bar and sized by the same optimizer used on time, range, or volume bars.
Searching bar size
Candidate range-bar increments run from 0.02 to 100, counted as 9999 sizes. Average true range does not collapse that search because the ATR reading changes with the clock interval used to measure it.
For one stream of oil ticks, the preferred range-bar size is not unique: a different algorithm or a different data sample implies a different size.
A bar-length search on certain foreign futures selected 13-minute and 19-minute intervals rather than the conventional 5-, 10-, 15-, or 20-minute grid.
Dual-direction bar size
A bar size that fits a trending oil move can be a poor match in pullbacks because with-trend volatility is not the same as counter-trend volatility. Independent of trend direction, downside volatility was observed a little higher than upside volatility except just before a blowoff.
Advancing-price bar size can be set independently of falling-price bar size so one procedure can sample two volatility environments.
A narrower search
System optimization is a joint search over rule inputs and bar construction so entry, exit, and abstention can be evaluated as one procedure.
Extra formula variables plus a free bar-size choice expand degrees of freedom. The narrower procedure holds a simple rule fixed, treats optimized bar type and size as a zero-lag filter, and retunes only the bars after a gradual fade, whereas start-and-end confirmation still withholds a similar number of bars at both edges of a trend.
A zero-lag bar filter uses bar type and size as the filter so the sampling grid, not a delayed confirmation count, defines when a trend is visible.
All readings on this track · 14 readings
- 2007Renko brick construction and breakout rules
- 2014Modified Renko brick size from tick inputs
- 2014Renko brick increment and open construction
- 2014Modified renko construction with RSI bands and a trend filter
- 2014Construction checklist for a brick-chart crossover and same-session breakout
- 2015Optimizing virtual bar size as a trend filter
- 2015Constructing Renko energy bands so support and resistance arrive as pre-marked leans
- 2018Five-rule entries and exits with Elliott waves and Renko bricks
- 2018Modified Renko charts as a filter for fake breakouts
- 2018V-Trade reversal, retrace, and continuation rules
- 2018A finished impulse, a Fibonacci target, and a reversing Renko brick as one test
- 2019Brick sampling changes what a volatility envelope can show
- 2019A consecutive Renko brick count as a mechanical entry-exit procedure
- 2019Renko brick scale as a falsifiable trend hypothesis