Skip to main content
Track Parabolic SAR
1 / 21
Search the archiveLibrary
Library

1987issue C051-9

Constructing parabolic stops and cycle-window averages

Build each overlay as a written contract. Name the inputs, state the update rule and the chart-scale assumptions, then check a tightening parabolic stop or a pair of cycle windows against actual highs, lows, and closes rather than against the shape the plot happens to draw.

  • The parabolic overlay is a trailing protective stop that tightens after favorable extremes and traces a curved path when a market is trending.
  • From a long start, the next stop equals the current stop plus the high-to-stop gap multiplied by an acceleration factor; a new high steps that factor up, and a touch reverses the side.
  • In chart-scaled coordinates the long-versus-short test can reverse because the screen origin sits at the top, so the plotted path has to be checked against the underlying highs and lows.
  • A two-window overlay belongs only on prices with cyclic content: a full-cycle window estimates the cycle trendline, and a half-cycle window is the more sensitive pair member.
Entries in this reading2 entries
Moving average18 readingsCurrent trackParabolic SAR21 readings

Write the overlay as a contract

Editorially, a plotted overlay is only as useful as the contract behind it. Name the inputs, state the update rule, and state the chart-scale assumptions. Then treat a tightening stop or a pair of cycle windows as a hypothesis you can check against actual highs, lows, and closes, not against the curve that happens to appear on the screen.

The construction below is the historical workflow. It does not score the overlays as a trading system.

Build the parabolic stop

The parabolic overlay is built as a trailing protective stop. It tightens after favorable extremes and traces a curved path when a market is trending. That parabolic stop starts at a recent extreme, ratchets closer after favorable highs or lows, and flips side when price touches it.

From a long start, the next stop equals the current stop plus the high-to-stop gap multiplied by an acceleration factor. A new high raises that factor by a fixed increment. A stop touch reverses the side and sets the new initial stop to the highest high of the prior long. The short side uses the counterpart rules.

The acceleration factor shrinks the remaining gap between the trailing stop and the latest favorable extreme. It steps up after each new extreme while the same side is held. One implemented parameterization starts the acceleration factor at 0.02, adds 0.05 on each new extreme, and caps it at 0.30.

Read the side from chart-scaled coordinates

After prices are scaled for plotting, a short is identified when the midpoint of the scaled high and low sits beyond the scaled stop. That rule follows from the chart-scaled coordinates: the screen origin is at the top and values increase downward, which can reverse the inequality used to tell long from short.

The plotted stop path can look like entry and exit markers even when those points are not the actual trade prices. The construction has to be checked against the underlying highs and lows.

Advance a uniform-window average

The average used here is a uniform-window average: an equally weighted mean of a fixed count of recent closes. It advances one bar at a time by dropping the oldest observation and adding the newest, instead of recomputing the full sum.

Size two windows from a dominant cycle

The two-window overlay is intended only for prices with cyclic content. A full-cycle window estimates the cycle trendline. A half-cycle window is the more sensitive pair member. The two theoretically cross at cyclic peaks and troughs.

Window lengths come from a user-entered dominant cycle. Half-cycle and quarter-cycle defaults are computed from that input. The calculations run over the last 50 stored records.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
1 of 21 in the Parabolic SAR track
19891-4 pp.Next on Parabolic SAREvaluating always-in parabolic SAR trailing stopsThe evaluated procedure is a trend-following design that stays continuously positioned by reversing when a trailing stop is hit.
All readings on this track · 21 readings
  1. 1987Constructing parabolic stops and cycle-window averages
  2. 1989Evaluating always-in parabolic SAR trailing stops
  3. 1993Constructing parabolic time-price trailing stops
  4. 1995Constructing parabolic SAR as an accelerating trail
  5. 1995Constructing a noise-buffered parabolic trailing stop
  6. 1997Constructing a parabolic trailing stop that only tightens
  7. 1998Triangle breakouts filtered by an exponential average and exited with parabolic stops
  8. 2000Treat volume-price imbalance as a hypothesis and let Parabolic SAR hold the exit
  9. 2002Constructing volatility stops from average true range and parabolic SAR
  10. 2002Parabolic SAR construction from stop outputs to reversal signals
  11. 2002Always-in-market SAR trail with directional confirmation
  12. 2004Forex trend confirmation with Average Directional Index, Parabolic SAR, and trendlines
  13. 2006Permission and fill gates for mechanical systems
  14. 2008A Relative Strength Index channel for profit lock and a trailing stop for capital protection
  15. 2010Building loss limits from the parabolic stop-and-reverse plot
  16. 2015Dual-zone currency indexes filter parabolic SAR signals
  17. 2016A parabolic trailing stop is not a complete system
  18. 2019Assigning jobs in a stochastic, parabolic, and moving-average stack
  19. 2019A three-filter stack as a redundancy test
  20. 2020Mechanical Parabolic SAR as an always-in flip after a breakout
  21. 2020Layering Relative Strength Index, MACD, and Parabolic SAR onto relative rotation maps
By fieldIndicators and filters56 readingsBy year19872 readings
All 21 readings tagged Parabolic SAR
Also on Parabolic SAR5 readings
1998Triangle breakouts filtered by an exponential average and exited with parabolic stopsMoving average2019Assigning jobs in a stochastic, parabolic, and moving-average stackMoving average2019A three-filter stack as a redundancy testMoving average1989Volume confirmation windows and exponential average constructionMoving average1992Equity-curve average as a live-capital gateMoving average