Skip to main content
Track Chi-square test
17 / 17
Library

1998issue C101-3

Timed breakout rules after a nested-bar contraction

A nested-bar-contraction is only the setup. The archive freezes the oldest bar as the reference range, accepts the geometry when the latest midpoint sits in a stated central band, and treats a later break of that frozen range as the entry. The same instance is then retired after one fill or after a timeout.

  • A nested-bar-contraction is a three-bar setup, not an order. The base-bar high and low become the only frozen reference range for later confirmation.
  • Acceptance also requires the apex, the midpoint of the latest bar, to sit inside a stated symmetry-band of the base-bar range.
  • A long or short signal is a later close through that frozen high or low, and only while the validity-window is still open.
  • The one-use-rule lets the same contraction open at most one position. Later crossings of that instance are ignored.
Entries in this reading3 entries

A setup, not an entry

The archive treats a nested-bar-contraction as a setup rather than a trade. The procedure freezes the oldest bar as the only reference range, accepts the finished geometry only when the latest midpoint sits in a stated central band, and waits for a later break of that frozen range before any entry is counted. The same instance is then retired after one fill or after a timeout.

How the contraction is defined

A nested-bar-contraction is a three-bar sequence in which each later bar's high and low sit inside the high and low of the bar immediately before it. The latest high and low sit inside the prior bar, and that prior bar sits inside the bar two periods earlier.

The oldest bar in that sequence is the base-bar. Its high and low become the frozen reference range for later confirmation.

Accepting the apex

Acceptance also requires the apex, the midpoint of the latest bar, to fall inside a symmetry-band of the base-bar range. That band is the central fraction of the oldest bar. One worksheet encodes the fraction as 0.28 and another as 0.12.

The apex is used only to accept or reject the completed contraction. It does not place the order.

The later break is the signal

A long signal is not the pattern print itself. It is a later close through the frozen high of the base-bar. A short signal is a later close through its frozen low.

One fill or a timeout

One encoding keeps the breakout eligible for 25 later bars. If that validity-window expires unused, the signal is withheld.

The same contraction is allowed to open only one position. The one-use-rule counts the first valid cross of the frozen base and ignores later crosses of that instance.

What happens after a fill

After a fill, the same one-position procedure can reverse on the opposite base break. It can also exit on close when price is 2 beyond entry, 2.5 beyond the best price, or more than 5 beyond entry.

A later score, not an order

A separate worksheet applies a forward-horizon-score. It takes the close 30 bars later minus the pattern close, then records how often the pattern appears, how often that later close is higher, and the maximum, minimum, and mean of those point changes. That tally is used to count events rather than to place an order.

A regime label around the same geometry

A regime-filter can sit around the same geometry without entering the contraction test. A 5-bar versus 20-bar simple average of the close is attached as a bullish or bearish regime label. It labels the surrounding trend and does not define the pattern itself.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
17 of 17 in the Chi-square test track
1982Track finished · Next track: Commodity Channel IndexThree gates on a 1982 pork-belly short78 readings
All readings on this track · 17 readings
  1. 1987Testing price-volume agreement after percent reversal filters
  2. 1988Constructing chi-square tests for two-way price counts
  3. 1988Building consensus indicators with correlation and the chi-square test
  4. 1988Test edges against chance, not story
  5. 1988Constructing an advance-decline divergence oscillator
  6. 1989Evaluate a contrary put-call premium ratio at a stated horizon
  7. 1990A weekly resistance-index from hourly volume-per-point
  8. 1990Testing breadth above moving averages by horizon
  9. 1990Evaluating member versus odd-lot breadth
  10. 1990A chi-square test of split frequency histograms across price aggregations
  11. 1990Evaluating smoothed secondary counts with a chi-square test
  12. 1991Treat session high and low times as codes, then require a chi-square check
  13. 1991A signed hourly swing catalog as a next-session chi-square check
  14. 1992Constructing a chi-square test as a gate for two-way market records
  15. 1992Percent filters, log point-and-figure, and breadth residuals
  16. 1997Build a chi-square stationarity screen before you forecast
  17. 1998Timed breakout rules after a nested-bar contraction
All 32 readings tagged Chi-square test
Also on Chi-square test5 readings