Skip to main content
Track Market breadth
52 / 71
Search the archiveLibrary
Library

1996issue C071-2

Constructing the McClellan oscillator and a calibrated summation index

The McClellan oscillator is the spread between 0.1 and 0.05 exponential smoothers of the daily advance-decline difference. The companion summation index is a running total of those readings and can be seeded to a +1000 neutral level before later values are added.

  • The same oscillator construction can track momentum in price, volume, or advance-decline counts, and the usual market-breadth form starts from the daily advance-decline difference.
  • Both exponential averages are seeded at zero, each trend is updated as the prior trend plus its smoothing constant times the gap to the latest advance-decline difference, and the oscillator is the 0.1 trend minus the 0.05 trend.
  • Starting the summation index at zero can keep it below zero for an entire illustrated window, so the layout uses a seed of 1000 minus 9 times the 0.1 trend plus 19 times the 0.05 trend.
  • After that +1000 neutral level is set, each later value adds the newest oscillator reading, and the calibrated index is treated as a bull-market reading above zero and a bear-market reading below zero.
Entries in this reading2 entries
Exponential smoothing117 readingsCurrent trackMarket breadth120 readings

Three steps to the oscillator

The same oscillator construction can track momentum in price, volume, or advance-decline counts. The usual market-breadth application uses the daily difference between advancing and declining listings. Market breadth is a participation signal built from how many listings advance versus decline, not from a single price series.

That daily series is the advance-decline difference: the count of advancing issues minus the count of declining issues. Building the McClellan oscillator is a three-step process. Form the daily advance-minus-decline difference, smooth that series with two exponential averages, then take the difference of those two averages.

The faster smoother is the ten-percent trend and uses a 0.1 constant. The slower smoother is the five-percent trend and uses a 0.05 constant. Both series are seeded at zero in the worked layout.

Exponential smoothing is a recursive average that blends each new observation with the prior average through a fixed smoothing constant. Each trend update equals the prior trend plus the smoothing constant times the gap between the latest advance-decline difference and that prior trend.

The McClellan oscillator equals the 0.1 trend minus the 0.05 trend. It is the point spread between the 10% trend and the 5% trend of the advance-decline difference.

McClellan oscillator from 10% and 5% NYSE trends

The tradeable line is only the spread between the slower 10% trend and the faster 5% trend of daily NYSE net advances. A mid-May breadth thrust lifts the oscillator near 96 before it fades through zero in late May and again in early June. All three series are the worksheet values from the construction table, 4 May through 12 June 1990, with both averages seeded at zero.
The tradeable line is only the spread between the slower 10% trend and the faster 5% trend of daily NYSE net advances. A mid-May breadth thrust lifts the oscillator near 96 before it fades through zero in late May and again in early June. All three series are the worksheet values from the construction table, 4 May through 12 June 1990, with both averages seeded at zero.NYSE advances and declines · daily · 1990-05-04T00:00:00.000Z to 1990-06-12T00:00:00.000Z

Both exponential averages are seeded at zero, as the McClellans recommend. The 10% trend uses smoothing constant 0.1; the 5% trend uses 0.05. Oscillator equals 10% trend minus 5% trend.

When the summation index starts at zero

The companion summation index is the running total of oscillator values. Starting that total at zero can keep the index below zero for an entire illustrated sample window.

Neutral calibration at +1000

A seed of 1000 minus 9 times the 0.1 trend plus 19 times the 0.05 trend places the summation index on a +1000 neutral level before later oscillator readings are added. Neutral calibration means seeding the cumulative index so a designated level stands for a balanced regime rather than an arbitrary spreadsheet start.

After that seed is set, each later summation value equals the previous calibrated total plus the newest oscillator reading.

With the 1000-neutral calibration, a summation index that remains above zero is treated as a bull-market reading and a drop below zero is treated as a bear-market reading.

A minimum run of 40 periods

The construction note recommends computing the oscillator for at least 40 periods, even though the illustrated calibration used 28 observations because of space limits.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
52 of 71 in the Market breadth track
19961-3 pp.Next on Market breadthDeclare the oscillator seed, then calibrate only the summation indexA breadth-oscillator seed computed on one date was about 2000, while the same construction computed on a later date was about 1000.
All readings on this track · 71 readings
  1. 1987How a failed rebound, weak breadth, and cycle dates broke the 1987 bull case
  2. 1988Diagnosing market bottoms with breadth, divergence and averages
  3. 1988Diagnosing index tops with breadth divergences
  4. 1988Record highs versus seven-day breadth and divergence
  5. 1989Constructing a percentage-scaled internals composite
  6. 1989Constructing a weekly block-tick breadth z-score
  7. 1989Constructing a dual-rate advance-decline oscillator
  8. 1989Normalize advance-decline series for a common-scale comparison
  9. 1990Unchanged-issue share as a narrow-breadth case study
  10. 1990Evaluating daily and weekly unsigned plurality breadth
  11. 1990Constructing paired new-high and new-low breadth indicators
  12. 1990Ten-day HI/LO extremes as a long-horizon breadth signal
  13. 1990Confirming index cycles with breadth, volume, and waves
  14. 1990Index cycle gates from breadth and volume
  15. 1990Constructing advance-decline breadth indicators
  16. 1990Weekly advance-decline oscillator: weight map, extremes, and spike cycle
  17. 1990Price-weighted construction distorts breadth, support, and trend
  18. 1991A peak-sequence test from the new-highs-to-advances-ratio
  19. 1991Fuzzy rules that turn daily market-breadth into a session consensus
  20. 1991From daily breadth tallies to a weighted consensus signal
  21. 1991Retesting market-breadth when market structure changes
  22. 1991Constructing TRIN as a breadth-volume ratio
  23. 1991Build the market clock before you read a price bar
  24. 1991A construction audit of the long-horizon trading index
  25. 1991Independent formula timers kept as a testable combination
  26. 1992When identical TRIN prints come from different pairings
  27. 1992Grade closing tick before a next-session breadth hypothesis
  28. 1992Noncumulative advance-decline swing confirmation
  29. 1992Five-day sum construction of the trading index
  30. 1992Daily closing-trin extremes and next-day direction
  31. 1992A three-layer audit: regime, breadth, and group RSI
  32. 1992Constructing a nine-state trend, momentum, and breadth score
  33. 1993Constructing a market-volume-impact rating from nested averages
  34. 1993When advance-decline confirmation counts the wrong universe
  35. 1993Constructing breadth momentum from advance-decline smoothing
  36. 1993Constructing a cumulative market-thrust line
  37. 1994Three-horizon construction of the Haurlan index
  38. 1994Checklist-gated session entry in 1993 index futures
  39. 1994Read one advance-decline pair through three windows
  40. 1994Constructing calibrated market-breadth summation indexes
  41. 1994Constructing a two-speed advance-decline oscillator and a calibrated summation
  42. 1995NYSE tick extremes and candlestick reversal entries
  43. 1995Assembling range, breadth, and a stored stop into one procedure
  44. 1995Restating market breadth timing rules as ratios
  45. 1995Constructing breadth ratio gates after lookback drawdowns
  46. 1995Building a short-range breadth and price oscillator
  47. 1996Constructing a smoothed advance-decline trend filter
  48. 1996Smoothed advance-decline alerts at the 1987 and 1990 turning points
  49. 1996Constructing breadth, RSI, and stochastic range filters
  50. 1996New-high and new-low counts as a breadth construction
  51. 1996Constructing the four-input breadth-volume ratio
  52. 1996Constructing the McClellan oscillator and a calibrated summation index
  53. 1996Declare the oscillator seed, then calibrate only the summation index
  54. 1997Three-gate centered strength in market-breadth construction
  55. 1997Daily advance-decline and new-high new-low breadth signals
  56. 1999Index-fund positions as a majority-vote committee
  57. 2000Tick, tiki and TRIN as a three-layer session confirmation stack
  58. 2000Constructing an advance-decline oscillator from one listed tape
  59. 2001Market breadth, beta, and volume-price confirmation
  60. 2001Regime context from relative venue volume, breadth, and intermarket spreads
  61. 2002When NYSE breadth misreads operating-stock participation
  62. 2003Two-gate breadth divergence and a trend filter for rally tops
  63. 2003Market internals confirm or diverge from the index
  64. 2004Constructing the McClellan oscillator and summation index
  65. 2005Intraday index-futures divergence as a three-part session hypothesis
  66. 2005Breadth summation levels as a short-term signal filter
  67. 2005Checking trend versus range with breadth and divergence
  68. 2011Constructing a Nasdaq hi-lo index from highs, lows, and issues traded
  69. 2013Cumulative advance-decline versus a one-year average
  70. 2013A one-year breadth average as a participation gate
  71. 2015Falsifying a healthy correction with breadth and support
By fieldMarket breadth101 readingsBy year199680 readings
All 120 readings tagged Market breadth
Also on Market breadth5 readings
1989Constructing a dual-rate advance-decline oscillatorExponential smoothing1989Constructing a weekly block-tick breadth z-scoreExponential smoothing1989Normalize advance-decline series for a common-scale comparisonExponential smoothing1990Evaluating member versus odd-lot breadthExponential smoothing1990Weekly advance-decline oscillator: weight map, extremes, and spike cycleExponential smoothing