2002issue C051-4
Evaluating mechanical trend signals from ten-session average excursions
After each close-versus-ten-session-average crossing, evaluation is a frequency audit: how often the next signed excursion continues, how much of the sample is occupied by long one-sided runs, and whether the long side and the short side behave as one market or two.
- A century of daily industrial-average closes supplied 27,380 sessions and 4,098 even-split crossings of a ten-session average, so one mechanical crossing signal can be scored as a frequency table.
- Mean peak signed excursions were 1.6 percent on both sides of the average, and later half-percent, one-percent, two-percent, and three-percent distances became steadily less common after both upward and downward crossings.
- Continuation after a two-percent or three-percent threshold was more frequent on downward episodes than on upward ones, and the average stay above the line lasted nearly two sessions longer than the average stay below it.
- Occupying runs of ten or more consecutive closes began after only 24 percent of crossings, yet they accounted for 58 percent of all sessions and for more of the time spent above the average than below it.
Audit one mechanical crossing signal
A century of daily industrial-average closes supplied 27,380 sessions and 4,098 crossings of a ten-session average, split evenly into 2,049 upward and 2,049 downward penetrations. The ten-session average is a simple average of the latest ten daily closes, used only as a reference line for signed percentage distance and for counting consecutive closes on one side of that line.
The mechanical crossing signal is a same-side entry cue defined solely by the close moving from one side of the ten-session average to the other, with no discretionary overlay. After each such crossing, the signed excursion is the percentage gap between a daily close and the average during the uninterrupted stay, recorded to a peak distance and a duration in sessions.
Peak distance and time on each side
Across the full sample the mean peak distance above the ten-session average and the mean peak distance below it were both 1.6 percent, while the average stay above the line lasted nearly two sessions longer than the average stay below it.
Mean peak percentage distances in 1901 to 1910 were 1.4 percent above and 1.6 percent below, close to those in 1991 to 2000, which were 1.5 percent above and 1.4 percent below.
How often the next signed excursion continues
After an upward crossing, the series later reached at least 0.5 percent above the average on 74 percent of occasions, or 1,516 signals. After a downward crossing it later reached at least 0.5 percent below on 69 percent of occasions, or 1,414 signals.
A later distance of 1 percent or more from the average occurred on 58 percent of upward crossings and 52 percent of downward crossings. A later distance of 2 percent or more occurred on 30 percent and 29 percent. A later distance of 3 percent or more occurred on 13 percent and 16 percent.
Continuation after threshold is a later close that pushes at least another half percent beyond a pre-chosen two-percent or three-percent distance from the average, counted as further travel rather than as an automatic reversal. Once a plus-2-percent or plus-3-percent distance was first reached, that further half-percent continuation occurred on 65 percent of those episodes. Once a minus-2-percent or minus-3-percent distance was first reached, it occurred on 75 percent and 70 percent of episodes.
Occupying runs and the share of the sample they fill
When a short-term trend was defined as at least ten consecutive closes on one side of the ten-session average, 24 percent of the 4,098 crossings began such an occupying run, including 29 percent of upward crossings and 20 percent of downward crossings. An occupying run separates those longer one-sided phases from the larger set of all crossings.
Those ten-session-or-longer runs accounted for 64 percent of all days spent above the average and 50 percent of all days spent below it, or 58 percent of the 27,380 sessions in the sample.
All readings on this track · 38 readings
- 1986A Marshall Wave book is assembled one confirmed stance at a time
- 1987Operator gut check when two systems conflict
- 1988Visual trend rules need one testable procedure
- 1988Audit trend and breakout rules sold without tests
- 1992Weekly 4 percent close reversal case study
- 1992Write down the plan, then wait for the cross
- 1994Constructing a three-rule moving-average system
- 1995Weekly long-only trend rules with a gated retracement exit
- 1997Constructing rainbow charts with repeated smoothing
- 1997Three-line break as a one-parameter mechanical filter
- 2000Nested horizon trend alignment with a momentum stack
- 2001When certainty turns off the stop
- 2001Price must confirm every setup
- 2001Constructing a color stack from every lookback
- 2002Evaluating mechanical trend signals from ten-session average excursions
- 2004Trend pullbacks with a fifty-day average and relative strength index
- 2005Stochastic overbought as a trend-regime state
- 2005Trend-following as one testable signal procedure
- 2007One checklist for entry, exit and abstention after a market switch
- 2008Parallel listings and divergent trend behavior
- 2008Recovery headlines versus trend-regime rules
- 2008Constructing an intraday trend procedure with confirming close, retest, and planned exits
- 2010Mechanical trend-following system design, stops, and validation
- 2010A trading-plan that sets entry, stop, trail, and pause together
- 2010Multi-horizon cycle signals, trend filters, and stops
- 2012Weekly bear-market and shock signals as a mechanical procedure
- 2013A 2013 rates week as a structure-first exit drill
- 2013Trend following as one testable procedure
- 2013A forex veto checklist: trend, pair economy, and crowd bias
- 2015Relative-strength-index rules under a 20-percent trend filter
- 2015Trend-gated oscillator entry and exit rules
- 2016Trend process versus end-goal trading
- 2017Gold as a classroom for one midterm-trend procedure
- 2017Falling-knife FX longs wait for a four-hour trend turn
- 2018A closed procedure for quarterly sector rank slots
- 2018A recursive median as shared state for trend rules
- 2020Moving-average construction as an objective trend filter
- 2020Constructing first and second cross trend entries