Diagnosing index tops with breadth divergences

A stacked internal audit

Market breadth means participation series such as advance-decline and new-high versus new-low measures that show whether an index move is widely or narrowly supported. The advance-decline line is a running total of net advancing minus declining issues, charted with support, resistance, and pattern rules as if it were price. The high-low differential is a moving average of new highs minus new lows that can turn before the parent index.

The TradersWeek reading, which is editorial, is to teach an equity top as a stacked internal audit. Apply the same trendline, failed-retest, and moving-average grammar to advance-decline, high-low, and oscillator panels that you would apply to price. Treat the high as a falsifiable case only after several of those panels lose confirmation together.

Oscillator and average panels

A McClellan Oscillator can be specified as the difference between a 19-session exponential average of net advances minus declines and a 39-session exponential average of the same series. That spread is the oscillator panel in the cases below.

A moving average is a fixed-lookback smooth of ordered price or breadth observations, used to mark trend breaks and crossovers on internal series. The Arms Index is a short-horizon internal ratio of advancing versus declining issues and volume, read for overbought or oversold conditions and trend breaks.

The 1986 NYSE Composite path

On the 1986 NYSE Composite path, a third higher high formed near 146 while the advance-decline line printed three lower peaks, a negative price-breadth divergence. Divergence here means a split between an index extreme and a related internal or oscillator series that fails to confirm that extreme.

Just before that 1986 high, the advance-decline line broke a marked support line and the following rebound stayed well below the prior peak.

As that NYSE peak formed, a 10-session Arms Index series returned to an overbought reading of 0.70 after a run of lower highs, and an Arms Index oscillator shifted into a designated negative state one session off the high without confirming the new price high.

Both 10-session and 30-session advance-decline differentials, and both 10-session and 30-session high-low differentials, turned down in April 1986 and then made lower peaks while the index continued higher.

The late-summer 1986 intermediate high

Ahead of the late-summer 1986 intermediate high, the 10-session advance-decline differential topped near +395 in mid-August, and that 10-session series crossed under the 30-session series more than three sessions before a 100-point drop in the industrial average.

The early 1987 advance

During the early-1987 advance, the advance-decline line traced a rising wedge through February and March that completed when early-April support failed, and the later rebound did not reclaim the prior peak.

The McClellan Oscillator crested in January 1987, then posted a sequence of lower highs. The early-April decline broke a marked support line, after which the oscillator only rebounded to the descending trendline.

A 30-session moving average of the advance-decline differential broke its uptrend in mid-March 1987 while prices were still rising. Matching high-low differential averages had already peaked, and a later break of prior lows on the 30-session high-low series confirmed the internal turn.

The mid-1988 snapshot

In the mid-1988 snapshot through late July, new recovery highs in the NYSE Composite were not confirmed by the advance-decline line, a 10-session advance-decline average failed to confirm at a secondary peak, and the McClellan Oscillator had already fallen for six weeks and broken both a support line and the zero line.

When several panels fail together

The editorial synthesis is that no single panel closes the case. The 1986 high combined a negative advance-decline divergence, a failed retest of broken support, an Arms Index that did not confirm the new price high, and differentials that had already turned down. The 1987 and 1988 sequences repeated that stack: pattern failure on the advance-decline line, lower oscillator highs, and moving-average breaks on the differentials while price was still firm.

That is the repeatable chart condition offered for a later hypothesis. It is not a forecast of a later market and it is not a trading rule.