Skip to main content
Track Price channel
4 / 55
Library

1988issue C081-10

Three-zone price channel from quadratic smoothing

A quadratic time-series projection with changing velocity and acceleration can be wrapped in a comfort-scaled expected-error band. The result is a timing envelope whose crossings partition the chart into deteriorating, neutral, and improving regions, unlike a moving-average cross that does not extrapolate.

  • The channel is a timing envelope for entering and exiting a security, not a device that projects how far price will travel.
  • Bounds are formed by adding and subtracting a comfort-scaled expected-error term from a quadratic forecast whose coefficients are allowed to change.
  • Price leaving the envelope marks a break in the assumed trend and partitions the chart into deteriorating, neutral, and improving regions.
  • The channel and a moving average both generate signals from price crossing a computed line, but the channel is designed to extrapolate and to place greater weight on recent observations.
Entries in this reading3 entries

A timing envelope, not a travel forecast

The channel is framed as a timing envelope for entering and exiting a security. It is not a device that projects how far price will travel.

The price channel is the pair of comfort-scaled bounds traced through time. It turns the forecast into an objective envelope instead of a point prediction.

A quadratic forecast with changing coefficients

Construction begins with a quadratic time-series projection whose coefficients represent trend velocity and acceleration and are allowed to change rather than stay fixed.

The quadratic forecast is a local polynomial path. Its coefficients stand for estimated price level, velocity, and acceleration at the moment of projection.

Level, velocity, and acceleration are updated recursively with analyst-chosen smoothing constants. Exponential smoothing is recursive weighting of ordered observations that refreshes level, velocity, acceleration, and error variance, giving more influence to recent prints.

Comfort-scaled expected error

Upper and lower bounds are formed by adding and subtracting a comfort-scaled expected-error term from the forecast line.

Expected error is the tracked gap between realized price and the model’s expected outcome. It is used to size the envelope rather than to revise a causal story.

In the illustrated construction, expected error is obtained by exponentially smoothing the squared differences between realized prices and the model’s expected outcomes.

The comfort factor is a risk-tolerance scalar applied to the expected-error estimate. It sets how far the upper and lower bounds sit from the forecast line. A comfort factor of 1 is presented as equivalent to one standard deviation on each side of the forecast, matching a conventional Student t-style interval around the best estimate.

Once coefficients, forecast horizon, and comfort factor are fixed, the channel is applied uniformly by formula. Freehand envelopes are treated as subjective and easily widened by a few large deviations.

Three regions from an envelope exit

Price leaving the envelope is used to mark a break in the assumed trend and to partition the chart into deteriorating, neutral, and improving regions.

The deteriorating region is the area below the lower bound, read as evidence that the assumed trend has weakened. The neutral region is the interior of the channel, read as consistent with continuation of the estimated trend. The improving region is the area above the upper bound, read as evidence that price is outrunning the prior trend.

Contrast with a moving-average cross

The channel and a moving average both generate signals from price crossing a computed line. The channel is designed to extrapolate and to place greater weight on recent observations.

A moving average, in this comparison, is a non-extrapolating average of past prices used as a baseline for crossing rules.

Editorial: the anticipatory three-zone channel is the object offered for checking. The moving-average cross does not extrapolate.

When the smoother is a poor fit

The quadratic smoother is described as unsuitable while a security is building a base, or while trend change is so small that the channel merely tracks price.

A worked daily initialization

A worked daily example initializes the smoothing constants and the error-smoothing constant at 0.5 and the comfort factor at 1. It then forms the next day’s bounds from a one-step forecast plus or minus the comfort-scaled error term.

DJIA one-step Holt-Winter channel, January 1988

The daily close drops through the lower comfort band on 8 January and then spends several sessions below the one-step forecast while the error bands stay wide after that gap. These numbers are the printed January 1988 Dow Jones Industrial Average worksheet, not a reading off a plotted curve.
The daily close drops through the lower comfort band on 8 January and then spends several sessions below the one-step forecast while the error bands stay wide after that gap. These numbers are the printed January 1988 Dow Jones Industrial Average worksheet, not a reading off a plotted curve.DJIA · daily · 1988-01-04T00:00:00.000Z to 1988-01-29T00:00:00.000Z

The source fixes smoothing weights a, b, c and d at 0.5 and a comfort factor of 1, and it projects only one trading day ahead. Upper and lower bands are not printed until the third session, once a variance estimate exists.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
4 of 55 in the Price channel track
19891-10 pp.Next on Price channelA variable-sensitivity stochastic built on three-sigma boundsConstruction places sigma-limits three standard deviations above and below a five-observation average after taking the square root of finite-lookback sample variance.
All readings on this track · 55 readings
  1. 1988Constructing price channels from trendlines
  2. 1988Three-point curved trend channel construction
  3. 1988Least-squares construction of channel trendlines
  4. 1988Three-zone price channel from quadratic smoothing
  5. 1989A variable-sensitivity stochastic built on three-sigma bounds
  6. 1989Close-minus-average oscillator for channel extremes
  7. 1989The six-stage hunt as a critique of one-click heroics
  8. 1990Fair-value gaps and a copper moving-average channel
  9. 1990Diversify markets, not systems, to cut trend-system variance
  10. 1991Constructing trendlines, price channels, and close-based breakouts
  11. 1991Constructing seasonal-cycle overlays with channel confirmation
  12. 1993Lag-compensated exponential trend channel construction
  13. 1993Constructing a lead-lag filter and price channel as one stack
  14. 1993Three stochastic warnings still need price-channel confirmation
  15. 1993Lead-lag smoothing for weekly trend-channel construction
  16. 1993Constructing zero-net-lag price channels
  17. 1995From a downtrend-line break to a regression channel
  18. 1995Validated trendline and price channel construction
  19. 1995Constructing price envelopes from averages, volatility, and regression
  20. 1996Constructing trendlines and channels from explicit swings
  21. 1998Fifty percent retracement as a channel regime test
  22. 1998Close-based channel rails as daily scenario maps
  23. 1999Constructing support, resistance, trendlines, and price channels
  24. 2001Cycle composites, price channels, and two-sided signals
  25. 2001Testing horizontal price channels with stops and scale
  26. 2002A two-stage momentum-shift and price-channel process
  27. 2002Wave-by-wave channel construction for Elliott counts
  28. 2002Affine channels as reusable trade hypotheses
  29. 2004Stress-test seasonal windows across regimes, then add channels
  30. 2004Regime permission from trendlines, channels, and range edges
  31. 2004Weekly-average and price-channel states on sector depositary baskets
  32. 2005Oil services catch-up after channel resistance breaks
  33. 2005Constructing a volatility-normalized cycle index
  34. 2005How a Darvas channel becomes a complete entry and exit procedure
  35. 2005Clustered Fibonacci and channel levels in news-driven forex
  36. 2005Treat a consolidating currency market as a time-frame problem
  37. 2005Channel walls that flip roles or recapture price
  38. 2006Stacking candlesticks, crossovers, and price channels
  39. 2006Failed uptrend channel breakout left the euro rangebound
  40. 2006Constructing a Wilson relative price channel from a range-bound strength index
  41. 2007Range bars change when a Bollinger squeeze counts as a breakout
  42. 2009One testable SPY procedure for a price channel, a trend rule, and a seasonal overlay
  43. 2010A gold-miner channel plan from value to false breakouts
  44. 2010A multi-timeframe channel from value to an overvalued zone
  45. 2010Asymmetric price channel construction for congested markets
  46. 2011Phasing many cycles at once with nested envelopes
  47. 2012Constructing adaptive horizontal price channels
  48. 2014Confirming support with trendlines, channels, and retracements
  49. 2015News-sentiment confirmation for support, channel, and volume tests
  50. 2015A three-layer permission stack: moving averages, a price channel, and weekly levels
  51. 2016Entropy-diff as a regime switch between trend following and a price channel
  52. 2017Competing rulers on a pound chart after Brexit
  53. 2017Test consolidation channel breakouts as one procedure
  54. 2020Constructing late-trend longs with a price channel, gap breakout, and trailing stop
  55. 2025Using IBM's multi-year price channel as a breakout teaching case
All 77 readings tagged Price channel
Also on Price channel5 readings