The Advance-Decline Line Manual

Research note. Every figure computed at render from our reconstructed daily breadth dataset (from 2016-11-01) — the same file behind the live A/D line tracker. Methods in How we checked it; editorial standards in How we use AI.

What is the advance-decline line?

The simplest breadth measure there is: count the stocks that rose today, subtract the stocks that fell, add the result to yesterday's running total. That running total is the A/D line. When it rises with the index, the advance is broad; when the index climbs while the line stalls, fewer and fewer stocks are doing the lifting. Ours counts eligible US-listed common shares — as of September 18, 2026, 1,658 advanced and 2,807 declined, putting the cumulative line at 18,774 against its all-time high of 84,809 set June 11, 2021.

Is breadth confirming the market right now?

As of the September 18, 2026 session: 1,658 stocks advanced, 2,807 declined — a net of -1,149. SPY sits 2.1% from its high. This comparison uses the 500 most-liquid stocks: their line last peaked in August 2026, 24 sessions ago, so our read is price and breadth retreating together — an ordinary pullback. The less-liquid tail's line peaked January 2017. The tracker updates all of this after every close.

How the line is built — and why levels lie

Each day is one number: advancers minus decliners, regardless of size — Apple and the smallest Russell name each get one vote. That equal vote is the whole point (it measures participation, not capitalization) and also the line's oldest weakness: the universe itself drifts. Stocks delist, IPOs arrive, and the classic NYSE version famously counts bond funds and preferreds among its "stocks." Ours restricts to US common stocks to cut that noise, and still inherits drift — which is why the line's level against 2021 is partly a statement about universe composition, not only about markets. Missing historical stocks and invalid prices can affect the shape as well as the level. The corrected dataset retains available delisted histories and excludes invalid or nonconsecutive price comparisons. It starts 2016-11-01, because earlier Nasdaq listing metadata is incomplete.

The divergence record — selected covered episodes

These selected SPY tops fall within the corrected dataset and have a full prior year of observations. The table compares each with the A/D peak in that preceding year:

These examples were selected because a market top followed, which limits what they establish. The line has also stalled without a top following. The tracker page runs the systematic version across every unconfirmed high, famous or obscure, and the honest answer is that divergence is a condition that precedes trouble. It does not schedule it.

Where the weakness lives — the cohort decomposition

We divide the daily votes by trailing 63-session average dollar volume: the 500 most-liquid names, ranks 501 through 2,000, and the remaining tail. Membership is re-ranked each session. The top-500 line peaked on August 14, 2026; the tail peaked on January 25, 2017 and now reads -80,683 from the common starting date. Liquidity correlates with company size, but these are liquidity ranks. Missing historical coverage can affect each group differently, so the gap alone does not establish its economic cause.

Where this gauge will mislead you

• Levels across eras mean nothing. The line's value is a running sum from an arbitrary start; universe drift compounds it. Compare direction and divergence, never the number itself.
• Divergences don't time tops. Three weeks to seven months of lead in our own record. A divergence says conditions have narrowed — it does not say when narrowness matters.
• Every A/D line is a universe choice. NYSE-composite lines include bond CEFs and preferreds; ours excludes them but inherits common-stock delisting drift. Cross-site comparisons are usually comparing universes, not markets.
• One vote per stock cuts both ways. Equal weighting is the feature — and it means the line can diverge for structural reasons (a bleeding micro-cap tail) while the investable market is healthy. The cohort chart above is the antidote.

The last 12 sessions

Full daily dataset: ad_line.json.

How we checked it

Advancers and decliners are counted daily across our common-stock universe (the same computation the tracker charts); the cumulative line and the cohort lines come straight from that file. Divergence leads are computed by finding the line's highest close in the year up to each named market top — the dates and session counts in the table recompute at render, so data revisions flow through. The named tops are the S&P 500's closing cycle highs.

Frequently asked questions