Advance/Decline Line

Key takeaways · 6

1. 01The advance/decline measure has three useful forms (daily ratio, cumulative line, and smoothed McClellan or Zweig variants) that answer different questions about the same underlying participation count.
2. 02Cleveland Trust began compiling A/D data in 1926. Barron's published it weekly from 1931. Lyman Lowry founded Lowry Research in 1938 and added the operating-companies-only refinement that survives today.
3. 03The textbook divergence ran April 1998 to March 2000: nearly 23 months of breadth deterioration as a narrowing band of mega-cap technology carried the cap-weighted indices. The subsequent S&P 500 drawdown reached 49% by October 2002.
4. 04The 2007 case is where the operating-companies-only distinction earned its keep. NYSE All-Issues A/D held up because rate-sensitive bond proxies were still rising; Lowry's OCO line had been falling for months. The S&P 500 fell 57% to March 2009.
5. 05Breadth thrusts measure the opposite — a sudden surge of broad participation off a low. Genuine Zweig Breadth Thrust signals are rare (around twenty since World War II) and have preceded large 11-12-month rallies in nearly every instance.
6. 06Advance/decline divergence is a necessary condition for many major tops but not a sufficient one. Divergences appear at false alarms too. Yellow light, not red light.

The daily net advance/decline ratio comes in two common spellings. Subtracting decliners from advancers and dividing by total issues yields a value between minus one and plus one: that is what TickerStance computes. Dividing advancers by the sum of advancers and decliners yields a value between zero and one, which is what Marty Zweig used as the input to his Breadth Thrust math. The two are algebraically related; the choice is editorial.

The cumulative A/D line is the running sum of net advances day after day, starting from an arbitrary value. The starting number is meaningless. Only the slope of the line, and its visual relationship to a price index plotted alongside it, carry information. When the index pushes to a new high and the cumulative line refuses to follow, that is the divergence chart breadth analysts have been pointing at for ninety years.

Smoothing turns a noisy daily reading into a momentum gauge. Sherman and Marian McClellan in 1969 took the difference between a 19-day exponential moving average of net advances and a 39-day exponential moving average. The values they called the ten-percent and five-percent trends in the language of the time. The output oscillates above and below zero on a roughly minus-one-hundred to plus-one-hundred scale and answers a different question than the daily ratio: not 'was today broad' but 'is breadth accelerating'.

The figure below stacks the three forms over the same year of synthetic data so the relationship is visible. The daily ratio is jagged and hard to read in isolation. The cumulative line is the slow-moving trend gauge. The McClellan oscillator splits the difference. Each is the right answer to a different question.

The three forms of advance/decline data on the same input. Each panel is the same trading year, smoothed differently. The daily ratio is honest but noisy; the cumulative line is the divergence chart; the McClellan oscillator is the momentum gauge.

The earliest sustained analytical use of advance/decline data is attributed to Colonel Leonard Porter Ayres and James F. Hughes at the Cleveland Trust Company in 1926. Ayres was vice president and chief economist at the bank for twenty-six years and one of the few establishment economists to argue in print, after October 1929, that the crash was the leading edge of a depression rather than a passing dislocation. Cleveland Trust began compiling daily counts of advancing versus declining issues to test whether broad participation, rather than headline indices, characterized real bull markets.

Hughes is separately credited with one of the first formal definitions of a selling climax: a session in which decliners exceed seventy percent of total issues while advancers drop below fifteen. That early work foreshadowed the entire literature on ninety-percent days that Lowry Research would build out four decades later.

Barron's began publishing weekly advance/decline figures in 1931, putting the data into retail circulation for the first time. The technique stayed academically dormant through the 1930s, the 1940s, and most of the 1950s. The data existed but very few analysts were doing anything serious with it.

Lyman M. Lowry founded Lowry Research Corporation in Florida in 1938 and extended the framework along two axes. Lowry tracked advancing and declining volume rather than just the count of issues. He also drew the line between operating common stocks and the rate-sensitive instruments that had begun cluttering NYSE listings — preferreds, closed-end bond funds, real-estate trusts, ADRs. The Operating Companies Only methodology survives today as the cleanest universe definition for breadth analysis. Lowry's firm became the longest continuously published technical investment advisory in the United States.

Richard Russell began Dow Theory Letters in 1958 and Joseph Granville's Granville Market Letter ran through the 1960s and 1970s. Both writers moved A/D data out of dusty Barron's tables and into active interpretation. The 1962 Kennedy Slide, a 27% Dow drawdown that began with a textbook A/D divergence, was the demonstration case that turned breadth from a curiosity into a discipline.

Sherman and Marian McClellan published Patterns for Profit in 1969. Working before computers were widely available, with everything plotted by hand, they recognized that net A/D was too noisy to read directly and applied two exponential moving averages of differing length to surface the rate of change of breadth. The McClellan Oscillator made breadth a momentum indicator rather than a participation count. The McClellan Summation Index (the running sum of the daily oscillator) plays the role analogous to the cumulative A/D line, with smoothing baked in. Tom McClellan, Sherman's son, founded the McClellan Market Report in 1995 and continues the work today.

Every modern breadth indicator on every modern dashboard descends from those three vintages. Cleveland Trust's count. Lowry's volume and operating-companies refinement. McClellan's smoothing. The math has barely moved in fifty years.

The set-piece chart of advance/decline analysis is the period from April 1998 through March 2000. The NYSE cumulative A/D line peaked in April 1998 and trended down for the next twenty-three months. The S&P 500 peaked on March 24, 2000, the Nasdaq Composite on March 10, 2000, the Dow Jones Industrial Average on January 14, 2000. Twenty-three months of negative divergence — the cumulative breadth line falling while the cap-weighted indices kept making new highs on the strength of an increasingly narrow cohort of mega-cap technology.

The mechanism is intuitive once you see it. Cap-weighted indices reward concentration. As the Internet bubble inflated, capital crowded into a few dozen technology and telecom names whose market caps grew faster than the indices that contained them. The other 2,500 NYSE-listed stocks did less and less. The headline number kept rising on the strength of the index leaders. The breadth line fell because the breadth count is one-vote-per-stock, and most of the stocks were no longer participating.

The subsequent drawdowns: the S&P 500 fell 49% to October 2002. The Nasdaq Composite fell 78% over the same window. The bear market lasted thirty-one months and earned its place in the record books.

The figure below shows both lines rebased to 100 at April 3, 1998. The vertical separation grows steadily through 1998 and 1999, peaks at the March 2000 top, then collapses together as both lines fall through 2001 and 2002. The trader takeaway is uncomfortable: the indicator gave the right answer twenty-three months early. Anyone who acted on it in April 1998 missed two more years of mega-cap gains. The indicator was right; the ergonomics of acting on it cleanly were not.

NYSE A/D Line versus S&P 500, April 1998 to October 2002, both rebased to 100. The twenty-three-month divergence between the A/D top and the S&P top is the canonical breadth-warning chart.

The 2007 case is where the operating-companies-only distinction earned its keep. The NYSE All-Issues A/D line peaked in June 2007 with lower highs through July and October as the headline NYSE Composite made fresh highs. The S&P 500 topped on October 9, 2007. The standard A/D line gave only a weak warning — a few months of mild divergence.

The All-Issues line was being held up by rate-sensitive bond proxies. As Treasury yields fell on early-recession scares, business-development companies, mortgage REITs, preferred stocks, and closed-end bond funds rallied uniformly. The NYSE counts roughly fifteen hundred such issues alongside the operating-company common stocks. When yields drop and bond proxies rise, the All-Issues count looks confirmatory even if the actual stock market is rolling over.

Lowry's Operating Companies Only line, which excludes those rate-sensitive issues by design, told a much darker story. Their separate measures lit up early. The percentage of OCO issues above the 30-day moving average dropped from 80% in early June to 64% by mid-July. Two ninety-percent down days fired within three trading sessions in late July, and a third arrived in early August without an intervening ninety-percent up day. Paul Desmond at Lowry was telling clients to sell into rallies and pare positions through August 2007 — two months before the S&P 500 peaked.

The subsequent S&P 500 drawdown reached 57% to March 2009. The lesson is universe selection. The TickerStance signal computes the daily net A/D ratio against the broad eligible universe (close at or above five dollars, dollar-volume at or above one million), which excludes thinly-traded micro-caps and structurally low-priced bond proxies. Operationally close to OCO with a liquidity floor. The figure below shows the gap that opened up in 2007 between the two universe definitions.

All-Issues A/D versus Operating Companies Only A/D in the run-up to the 2007 top. The All-Issues line was held up by rate-sensitive bond proxies; OCO showed the real picture.

The opposite of divergence is a breadth thrust — a sudden, broad surge of participation off a market low that is rare enough to mark the start of a durable advance. Marty Zweig formalized the canonical version in his 1986 book Winning on Wall Street, revised in 1994. The math: a 10-day exponential moving average of advancers divided by the sum of advancers and decliners moves from below 0.40 to above 0.615 within ten trading sessions or fewer. Zweig's own framing: it is very rare for advances to lead declines by a ratio of two-to-one over such a span.

Roughly twenty Zweig Breadth Thrust signals have triggered since World War II. The list of years on practitioner audits is consistent: 1962 (twice), 1966, 1970, 1975, 1982 (the start of the great bull market), 1984, 1987 post-crash, 2009 post-financial-crisis, January 7, 2019, April 2020, and a handful of debated more-recent calls. The forward-return record is the reason the indicator gets its weight in the literature. Nearly every clean signal preceded an eleven-or-twelve-month rally, with average gains in the high twenties across the historical sample compiled by Kirkpatrick and Dahlquist.

Wayne Whaley's Planes, Trains and Automobiles paper, which won the Charles H. Dow Award, catalogs three thrust types over a five-day window: a breadth thrust on advancers as a share of total issues crossing 75%, a volume thrust on up-volume as a share of total volume crossing 77.8%, and a price thrust on the S&P 500 cumulative gain crossing 10.05%. Conjunctions of multiple thrust types in the same five-day window are statistically extraordinary.

Walter Deemer's Breakaway Momentum is a sibling concept with a slightly different mathematical structure. A 10-day total of NYSE advancers needs to exceed 1.97 times the 10-day total of decliners, or a 20-day total needs to exceed 1.72 times. Triggers very rarely. Nearly every signal sits within a few weeks of a major bull-market initiation.

Paul Desmond at Lowry, in his 2002 Charles Dow Award paper Identifying Bear Market Bottoms and New Bull Markets, documented the parallel pattern from the other end. Almost every major decline since 1933 contained at least one ninety-percent downside day where down-volume equaled at least 90% of the sum of up-volume and down-volume. Capitulation followed by a ninety-percent upside day was Desmond's signal of demand exhaustion of the selling.

Thrust signals are rare by design. Their value comes from the rarity. The prescription when one fires is straightforward: stay aggressive, do not fight the new tape. The harder discipline is recognizing when no thrust has fired — most of market history is the in-between, where breadth is neither thrusting nor diverging dramatically and the dashboard is asking you to read the regime more carefully than a single signal can.

The cleanest recent breadth divergence is also the messiest. The Russell 2000 peaked on November 8, 2021. The S&P 500 peaked on January 3, 2022. Two months of small-cap-leadership rolling over before the headline index turned. The S&P 500 / S&P 1500 cumulative A/D line had gone sideways for roughly six months while the cap-weighted index continued rising — a narrowing breadth divergence concentrated in mega-cap technology and the so-called Magnificent Seven.

The divergence was much shorter than the 1998-2000 set piece. Many breadth chartists argue the cumulative A/D line itself did not give a clean signal in 2021. The warning was visible mostly via Russell 2000 underperformance, percentage-above-200-day-moving-average readings rolling over, and the count of stocks making new fifty-two-week highs collapsing while the S&P 500 made new highs.

From November 2021 through mid-2024, the S&P 500 returned roughly 9% on a price basis while the Russell 2000 was down nearly 20% from its peak. The mega-cap-versus-everything-else divergence persisted for the better part of three years.

This is the case that made Russell 2000 A/D the modern analyst favorite for measuring speculative risk appetite. The small-cap universe is operating-company-pure: no closed-end bond funds, few real-estate trusts at scale, no preferred-stock contamination. Underperformance there is a clean read on whether the marginal dollar is moving toward risk or away from it. The 2021-22 sequence made the case for what to count more loudly than 2007 did, even though 2007 had the cleaner signal.

Advance/decline is a question about a specific universe, not a universal constant. Whichever universe you pick, you are answering a slightly different question. Pick deliberately.

The most damaging pitfall in breadth analysis is the always-divergent problem. Breadth divergences appear at most major market tops. They also appear at false alarms. Practitioners suffer from recall bias: they remember 1998-2000, they remember 2007, and they forget 1995-1996 (the A/D line briefly diverged with no top), 2014 (multiple divergences resolved by recovery), 2015-16 (more of the same). A divergence is a yellow light, not a red light.

The August 1987 case sharpens the point. The A/D line did diverge from the S&P 500 in late summer of 1987, two months before Black Monday on October 19. The divergence was real. The 22.6% single-session crash was a portfolio-insurance feedback loop that no breadth indicator could have predicted in advance. A trader who acted on the August divergence would have been right in direction and wrong in magnitude — early to a small correction, then blindsided by an event that breadth did not even register.

Survivorship in the issuance base is a quieter problem. The universe of NYSE-listed issues changes constantly. The 1990s biotech IPO wave, the 2000s ETF explosion, the 2020-2021 SPAC wave each created brief structural shifts in the count of issues with nothing to do with actual breadth. Cumulative A/D lines computed across decades depend on how the issuance base evolved. OCO partially corrects, but not fully.

Single-day reads versus smoothed reads is a real trade-off, not a settled question. The daily ratio is honest and zero-lag; its cost is noise. The 10-day smoothed reading is cleaner; its cost is a five-to-seven-session lag at turning points. The McClellan oscillator is the smoothest; its cost is being least interpretable as a same-day reading. A dashboard that exposes only one form is leaving information on the floor; a dashboard that exposes all three should explain what each is for.

Two exotic indicators in the breadth family deserve skeptical mention. The Hindenburg Omen (Jim Miekka, named for the 1937 disaster) triggers on simultaneous extremes of new fifty-two-week highs and new fifty-two-week lows. The Titanic Syndrome triggers when new lows exceed new highs near a fresh index high. The underlying logic is reasonable; the empirical record is poor. SentimenTrader once described one of them as a silly indicator with a deadly record. Tom McClellan has written sympathetically about both while noting the false-positive rate. Name them, gesture at them, treat with caution.

William J. O'Neil founded William O'Neil + Co. in 1963 and Investor's Business Daily in 1984. The M in CANSLIM is Market direction. O'Neil was emphatic that three out of four stocks follow the market, so the M dwarfs the other six letters. His primary tool for reading the M is the price and volume action of the major indices (the Follow-Through Day signal), not the A/D line directly. Investor's Business Daily has always published breadth statistics as supporting evidence, but breadth is a confirming indicator in CANSLIM, not the lead.

Stan Weinstein's Secrets for Profiting in Bull and Bear Markets, published in 1988, is the breadth-heaviest of the classic-school books. Chapter 8 is titled the Weight of Evidence and explicitly uses the NYSE Bullish Percent Index, the advance/decline line, and the new-highs-minus-new-lows differential as primary market-condition tools. Weinstein uses weekly charts and the thirty-week moving average to classify individual stocks into stages, but his market-timing overlay is breadth-heavy. Of the major swing-trading authors, Weinstein leans most explicitly on A/D-style measures.

Mark Minervini's Trade Like a Stock Market Wizard and Think and Trade Like a Champion both apply the Trend Template to individual stocks. Market direction is judged separately through what Minervini calls general market analysis. He watches the major averages relative to their fifty-, one-hundred-fifty-, and two-hundred-day moving averages, and he watches the count of stocks meeting Trend Template criteria. That count is functionally a leadership-quality breadth measure: in strong bull markets twenty to fifty stocks pass, in weak markets zero to five do.

Kristjan Kullamägi, the Estonian trader behind the Qullamaggie handle who compounded a small account into nine figures, is leadership-stock-driven and treats index posture as gate-keeping. His public framework is stay aggressive in good markets, raise cash in bad markets. The market-condition read that gates the system is anchored to a regime view that practitioners often build from breadth and percent-above-MA measures, even when not labeled that way.

The synthesis: breadth does not tell you what to buy. It tells you whether to buy aggressively or defensively. O'Neil and Minervini both use index price action as the primary regime read; breadth, especially through the McClellan family, confirms or undercuts what price is telling you. Weinstein is the closest classical analog for breadth as a primary market-condition tool. Different schools, same instruction: read the count before you read the chart.

The breadth.ad_ratio signal carries a 5% weight inside the Breadth subscore. It computes the daily net advance/decline ratio across the broad eligible US equity universe (close at or above five dollars, thirty-day median dollar-volume at or above one million). The eligibility filter is a deliberate echo of the Operating Companies Only methodology Lowry codified in the late 1930s, with a modern liquidity floor added on top. The signal is a single-day, unsmoothed reading: maximally honest about what one trading session looked like.

The smoothed sibling is breadth.mcclellan, the McClellan Oscillator, at 15% weight inside Breadth. McClellan does the heavy lifting for breadth momentum on the dashboard. The two signals exist together because they answer different questions: the McClellan reading tells you whether breadth is accelerating, the daily ad_ratio tells you what today's participation actually was. The pairing is the same one Sherman and Marian McClellan were addressing in 1969 when they invented the EMA-based smoothing — single-day reads were too noisy to act on, but the underlying participation count was the most important piece of market truth there was.

Breadth is 30% of the headline Stance composite, equal in weight to Trend, ahead of Leadership and Macro. The other Breadth ingredients (percentage of stocks above their 50- and 200-day moving averages, distribution days, new highs minus new lows, follow-through days, McClellan Oscillator, up-volume versus down-volume ratio) all share the load. The full math, every weight, every normalization rule, is on the methodology page.

A trader who has read this far does not need TickerStance to read breadth. Watching the NYSE A/D ratio daily and the McClellan Oscillator on StockCharts, free and unsubscribed, will give the same signal. The dashboard consolidates that read into a single reproducible number, weighted against three other regime lenses, with the ingredients exposed so the reader can disagree with the weighting. That is the trade. The methodology is the product.

Advance/decline ratio — net advances divided by total eligible issues on a single trading day. The daily participation snapshot. Range minus one to plus one in TickerStance's spelling.

Advance/decline line — the cumulative running sum of net advances day after day, plotted as a continuous curve. Compared visually against a price index to read divergence.

Net advances — the count of advancing stocks minus the count of declining stocks on a session. The raw input to every form of A/D analysis.

McClellan Oscillator — the 19-day exponential moving average of net advances minus the 39-day exponential moving average. The momentum-of-breadth gauge introduced by Sherman and Marian McClellan in 1969.

McClellan Summation Index — the running sum of the daily McClellan Oscillator. The smoothed analog of the cumulative A/D line.

Zweig Breadth Thrust — the 10-day exponential moving average of advancers as a share of total issues crossing from below 0.40 to above 0.615 within ten trading sessions. Marty Zweig's start-of-bull signal.

Operating Companies Only (OCO) — Lowry Research's universe definition: common stock of operating businesses only, excluding preferreds, ADRs, closed-end bond funds, REITs, and other rate-sensitive instruments. The cleanest universe for breadth analysis.

Ninety-ninety day — Paul Desmond of Lowry's term for a session in which down-volume equals at least 90% of the sum of up-volume and down-volume, and points lost equal at least 90% of the sum of points gained and points lost. Marker of capitulation selling.

Breakaway Momentum — Walter Deemer's breadth-thrust variant. A 10-day total of NYSE advancers exceeding 1.97 times the 10-day total of decliners, or a 20-day total exceeding 1.72 times. Triggers very rarely; nearly every signal sits at a major bull-market initiation.

Breadth divergence — sustained months-long divergence between the cumulative A/D line and a major price index. Most often appears as the A/D line falling while the index rises, signaling narrowing leadership.

Hindenburg Omen — Jim Miekka's exotic breadth signal that fires on simultaneous extremes of new fifty-two-week highs and new fifty-two-week lows. Reasonable underlying logic; poor empirical record.

Breadth thrust — any indicator measuring a sudden surge of broad participation off a market low. Zweig, Whaley, and Deemer all formalized variants.

Frequently asked questions

Keep reading

• Market RegimeWhy breadth is one of the four lenses for reading market regime.
• Follow-Through DayFollow-Through Days and breadth thrusts both confirm rally attempts. Here is how they differ.
• Sector RotationSector A/D and broad A/D answer different questions about who is leading.
• MethodologyEvery breadth signal, every weight, the full math.