Skip to main content
Track Rank rotation
23 / 38
Search the archiveLibrary
Library

2015issue C0324-27

MACD crossover evaluation by trend rank rotation

A default 12/26/9 signal-line crossover is held fixed as a long-only hypothesis, first on an unfiltered equal-weighted daily equity set, then rotated onto names ranked by isolated trend range and uptrend count.

  • The evaluation treated the 12/26/9 MACD line versus its nine-period signal average as a long-only signal-line crossover, with entry on an upward cross and exit on the next downward cross, both at that day's close.
  • About 1,816 eligible names from major U.S. listings entered an equal-weighted historical test; the author treated the unfiltered summary as too weak to support the crossover as a standalone procedure on a randomly assembled set.
  • An isolated trend required a 20 percent range. Names were then ranked by trend range and uptrend count, and the same unchanged rules were rotated onto three mutually exclusive 25-name buckets.
  • The highest combined-rank bucket led the next two, the selected buckets improved on the unfiltered summary, and the study changed only the ranked universe rather than the 12/26/9 settings.
Entries in this reading3 entries
Moving-average crossover3 readingsMACD4 readingsCurrent trackRank rotation50 readings

A fixed long-only crossover

The evaluation treated a 12/26/9 MACD line versus its nine-period signal average as a moving-average crossover. The MACD line is the spread between the 12-period and 26-period exponential averages of closes. A signal-line crossover recorded a long entry when that line crossed above its nine-period signal average and an exit when it crossed back below. Both marks used that session close.

Eligibility and the unfiltered summary

Eligibility required common stock, a price of at least two dollars, and continuous daily open-high-low-close history from 1 January 2005 through 31 August 2014 with no unadjusted split errors. About 1,816 names from major U.S. listings entered an equal-weighted, long-only historical test of those unchanged crossover rules. The equal-weighted summary gave every eligible name the same weight so price level did not dominate the evaluation.

The author treated that unfiltered equal-weighted summary as too weak to support the MACD-signal crossover as a standalone procedure on a randomly assembled set.

Average close-to-close percentage change on crossover days exceeded the average change on ordinary isolated uptrend or downtrend days. The study used that comparison to tie the signal to larger momentum bursts. An advance or decline was counted as an isolated trend only when its isolated range reached at least 20 percent.

Trend range is the measured extent of one isolated advance or decline once that 20 percent threshold is met. Uptrend count is the number of isolated advances of at least 20 percent observed for a name inside the study window. Each name received a combined rank from those two traits, with higher rank assigned to larger ranges and more frequent 20-percent-plus advances.

Rank rotation of the unchanged rules

Rank rotation sorted the universe by that combined rank, then retested the same signal rules on successive, mutually exclusive rank buckets instead of on the full unfiltered list. The same crossover rules were reapplied, without changing the 12/26/9 settings, to three mutually exclusive 25-name buckets covering ranks 1 to 25, 26 to 50, and 51 to 75.

Grouped retests followed the rank order. The highest combined-rank bucket led the next two, and the selected buckets improved on the unfiltered summary. The author read that pattern as a positive link between those trend traits and the MACD evaluation.

Suitability screen, not a rewrite

The study kept MACD parameters fixed and changed only the ranked universe. A suitability screen holds the indicator settings fixed while restricting the test to names whose isolated trend traits match the momentum bursts a MACD crossover is built to notice. The study presented that screen as an implementation step rather than a rewrite of the indicator.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
23 of 38 in the Rank rotation track
201534-38 pp.Next on Rank rotationPersistence and strength as one close-to-close switchThe persistence-switch enters just before the 3:59 close on the first up session and exits at the same clock on the first down session, so each cycle donates two days and a streak-break refuses a second consecutive down session.
All readings on this track · 38 readings
  1. 1987A mechanical rank-rotation sleeve for monthly fund leaders
  2. 1989Rank rotation in a five-name no-load sleeve
  3. 1990Cycle-tested five-year fund rank rotation
  4. 1991Blue-chip rank rotation by relative-strength-index slope
  5. 1992Currency rank rotation and intermarket timing
  6. 1992Rank rotation and relative strength for portfolio construction
  7. 1994MACD crossovers then short-horizon rank rotation
  8. 1994A comparable group-trend ledger from published ranks
  9. 1997Normalized yield rank rotation as a full portfolio procedure
  10. 1997Constructing an investor preference index from two capitalization-weighted series
  11. 1998Constructing anchored momentum from a centered average
  12. 2000Rank rotation, a stop-loss order, and Relative Strength Index in fund switching
  13. 2003A one-fund daily rank is a two-sleeve construction problem
  14. 2004Evaluate rank rotation only where persistence already exists
  15. 2004Sector fund rank rotation with regression and trailing stops
  16. 2006Evaluating equal-weight annual yield-rank rotation
  17. 2007Weekly preferred-symbol reselection for mechanical trend systems
  18. 2011Portfolio capacity and entry pacing for mechanical systems
  19. 2011Rank rotation as a testable ETF construction procedure
  20. 2011The inverse-Fisher stochastic is a forecast layer until the book rules can be disabled
  21. 2012An underwater stretch is a sizing test for rank rotation
  22. 2015Rule-based ETF rotation as one testable procedure
  23. 2015MACD crossover evaluation by trend rank rotation
  24. 2015Persistence and strength as one close-to-close switch
  25. 2015Evaluating rank rotation after a persistence screen
  26. 2016Evaluating an annual valuation rank rotation
  27. 2017A two-step yield and price rank rotation for a five-name sleeve
  28. 2018Smoothed volatility and the missing rank-rotation exit
  29. 2018A five-condition scorecard that ranks stocks and can refuse the trade
  30. 2018Small-cap growth sleeve eligibility with trend and rank rotation
  31. 2018Evaluating rank-rotation momentum across fund wrappers
  32. 2019Evaluating an annual equity-gold momentum rank rotation
  33. 2019Evaluating equity-gold momentum on funds versus indexes
  34. 2020How a signed comparative-strength oscillator is built for rank rotation
  35. 2020Which calendar clock changes a gold-versus-equity rotation test
  36. 2020Four-dimension relative strength as rank rotation
  37. 2020Portfolio construction as a ranked relative-strength problem
  38. 2020Two clocks for a Nasdaq put/call sleeve
By fieldIndicators and filters69 readingsBy year201516 readings
All 50 readings tagged Rank rotation
Also on Rank rotation5 readings
1994MACD crossovers then short-horizon rank rotationMoving-average crossover2003Rebuilding band distance as a z-score crossoverMoving-average crossover2018Small-cap growth sleeve eligibility with trend and rank rotationMACD2020Layering Relative Strength Index, MACD, and Parabolic SAR onto relative rotation mapsMACD1987A mechanical rank-rotation sleeve for monthly fund leaders