Filtering MACD false signals with trendline breaks

A warning is not a trigger

A 12-26-9 moving-average convergence/divergence oscillator is presented as a source of buy and sell readings that can fail to persist. The same oscillator is also used to spot disagreement with price.

The archive then applies a later test: a sloping line through successive highs or lows. A break of that line is treated as the later test of an earlier oscillator warning.

The TradersWeek editorial view is that MACD only names a disagreement with price. The drawn trendline is what later accepts or rejects that warning. Until the line breaks, the oscillator stays a draft, not a trigger.

Readings that failed to persist

In May 2007 a MACD reading is described as signaling a dollar purchase versus the rupee near 40.50. The rupee then moved toward 39.00, and a year later the pair was still near the May 2007 area.

In the second quarter of 2009 a MACD signal to buy the dollar near 46.5 is described as a short-lived bounce that reached about 49.00 before the rupee strengthened, with the pair later near 44.5.

Both crossings are false-signals in the archive sense: the oscillator named a new direction, and that direction did not continue.

A 2007 warning confirmed in May 2008

A mid-2007 rupee sequence of lower lows and lower highs against a MACD sequence of higher lows and higher highs is labeled a bullish divergence. That split between price structure and MACD direction is treated as a trend-change warning, not an immediate action.

The illustrated response is to wait for a break of a downsloping trendline. That break is dated May 2008, nearly a year after the May 2007 MACD disagreement. The later price break is the confirmation that upgrades the warning into a timed, checkable hypothesis.

Acting on MACD alone is described as selling the rupee too early. The May 2008 trendline break is said to have allowed a short near the same price as the May 2007 oscillator reading.

The early 2009 disagreement

After a December 2008 rupee print near 50.50, a January 2009 recovery to about 46.50, and a March 2009 move below 51.00, the March 2009 MACD peak is described as lower than the November 2008 peak.

That early-2009 MACD disagreement is treated as a warning confirmed only when a trendline broke in April 2009. The oscillator again named the split; the line break is what timed it.

A later warning still pending a second break

A later rupee decline of lower highs and lower lows against rising MACD lows is again treated as a warning pending a break of a second downsloping trendline, after a longer-term rising trendline had already given way.

The longer-term chart marks 43.50 and 41.73 as a historically significant support-and-resistance band that could contain further rupee strength.

The TradersWeek editorial lesson is the delay. Divergence is allowed to sit as a warning while price is still making lower highs and lower lows. The hypothesis becomes timed and checkable only when the relevant downsloping trendline breaks.