Noncumulative advance-decline swing confirmation

The running total is a slow comparison

A conventional advance-decline series is built from a starting base by adding that day's net advances when advances exceed declines, or subtracting that day's net declines when declines exceed advances, then plotting the running total with the Dow Jones Industrial Average. That running sum is the cumulative-advance-decline-line.

Under the conventional reading, a new high in the Dow Jones Industrial Average that is not matched by a new high in the running advance-decline total is treated as negative divergence and a candidate top condition.

The running advance-decline total can separate from the price average for a long interval. One cited case has the breadth total peaking on August 8, 1989 while the average's final high arrived on July 17, 1990, eleven months later.

Plot each session as a residual

The alternative construction plots only that day's advances minus declines as a discrete value. 1,000 advances and 600 declines are recorded as +400, and 900 declines and 400 advances are recorded as -500. That single-session residual is the net-advance-print. Market-breadth is the daily count of advancing versus declining issues, used here as that residual so it can confirm or fail a price swing.

A sell hypothesis is defined only after three higher peaks in the Dow Jones Industrial Average align with three lower peaks in the daily net-advance print. That alignment is the three-peak-sell-sequence. A buy hypothesis is defined when the Dow Jones Industrial Average makes two lower swing lows while the daily net-advance print makes two higher lows. That alignment is the two-trough-buy-sequence.

A breakout is a new closing extreme in the reference average for the current swing, and it must be checked against that day's net-advance print before it is treated as confirmed participation. Divergence is a mismatch in which the reference average makes a new swing high or low while the daily net-advance print fails to make a corresponding new extreme.

Three weaker peaks, then a repeated high

In the May-June 1990 walk-through, the average printed 2801 with +943 net advances on May 11, from 1,282 advances minus 339 declines. It printed 2870 with +673 on May 29, and 2935 with +628 on June 4, completing the three-peak sell sequence.

After that sequence, the average declined to a closing low of 2842 on June 26, 1990, then made a new closing high of 2969 on July 12, 1990 with only 442 net advances, which is treated as a repeated sell sequence.

Validate each swing on the same day

A completed sequence is treated as valid only if, on each of the three top peaks or two bottom troughs, the average makes a new hourly extreme on the same day as the new closing extreme for that swing. That same-day check is the swing-hour-close-filter. It does not require a full bull- or bear-market extreme.

The daily net-advance sequence is presented as a complement to other short-horizon timing tools rather than a standalone decision rule.