A peak-sequence test from the new-highs-to-advances-ratio

Daily construction

Daily construction takes the new-highs-to-advances-ratio and then applies a 10-day exponential moving average to that ratio. The new-highs-to-advances-ratio is the daily count of issues making new highs divided by the count of advancing issues, before any smoothing. The exponentially averaged form is the smoothed-breadth-series, which is read as a sequence of peaks rather than as a single-session value.

The recursive smoother updates yesterday's reading by adding 0.18 times the difference between today's new-highs-to-advances-ratio and yesterday's reading. The smoothing-constant 0.18 is taken from 2/(N+1) when N is 10, so the result approximates a 10-day exponential moving average.

Reading the peak sequence

A major advance is read from a rising-peak-sequence: a run of successively higher highs in the smoothed-breadth-series. A failed-peak-signal is a material high in the smoothed-breadth-series that stays below the prior high. That failed peak is the stated signal that the advance has ended.

Each peak in the smoothed-breadth-series is described as generally corresponding to an intermediate-term market high. The ratio is expected to contract late in an advance because issues can still print as advances on a bounce without making new highs.

Low readings and unconfirmed rebounds

In a prolonged decline the smoothed-breadth-series is described as shrinking toward zero and remaining there until a later advance lifts it, often abruptly, from those low values. A low-base-lift, a sudden rise in the smoothed series after a stretch of low readings, is treated as the start of a new advance and as the onset of a major uptrend.

When price rebounds while the smoothed-breadth-series does not confirm, that mismatch is treated as breadth-price-divergence. The mismatch is used to argue that the rebound is not durable.

Other published counts

The same new-highs and advances inputs can be assembled for other markets that publish those counts. A two-average crossover of the series was examined without adding interpretive insight.