The Zweig Breadth Thrust Manual

Research note. Computed at render from our daily advance-decline data on a ~5,000-symbol common-stock universe, 2010+. Pre-2010 claims are the literature’s, cited as such — we publish only what we can recompute. See How we checked it.

What is the Zweig Breadth Thrust?

Martin Zweig’s rule for catching the moment a bear-market tape turns into a stampede: the 10-day EMA of the advance ratio — advancing issues divided by advancing plus declining — rising from below 0.40 to above 0.615 within 10 trading sessions. Two weeks from washed-out to overwhelming. The washout arms the signal; only the speed of the recovery completes it, which is why decades can pass without one. The broader idea has its own concept page, and this exact rule has a definition & spec page; this Manual is about what the signal has actually done and where it will fool you.

Is one firing now?

As of August 21, 2026: the EMA sits at 0.50 and the signal is not armed — participation has not washed out below 0.40 recently, so no thrust can begin until it does. The last completed thrust on our universe was January 8, 2019, 1,915 sessions ago. The live gauge updates this state after every close.

How the signal is built

Each session’s advance ratio uses advancing and declining common stocks only on our ~5,000-symbol universe — ETFs, preferreds and index pseudo-tickers excluded — smoothed with a 10-day EMA. A state machine tracks two things: whether the EMA has closed below 0.40 (armed), and whether it then crosses 0.615 within 10 sessions (completed). Zweig’s original work used NYSE issues, and the classic 0.40/0.615 thresholds were tuned to that universe; ours is broader, which moves marginal outcomes — the next two sections show exactly how much that matters.

Our record — every thrust since 2010

The 1945+ record, as the literature tells it

The ranking references agree on the lore, and it deserves both respect and quotation marks. The commonly cited record — originating with Zweig’s Winning on Wall Street and extended by NYSE-data studies — counts roughly 20 signals since World War II, averaging about +23% over the following year, with no negative 6- or 12-month outcomes since 1950 and a famous 25-year drought from 1984 to 2009. We cite those numbers rather than print them as ours: they come from NYSE issue data we cannot independently recompute, some versions count overlapping signals differently, and pre-1970s NYSE breadth included far fewer interest-rate-sensitive issues than the modern tape — the classic caveat the better sources state themselves. What we can verify is the shape: on our own universe and era, every completed thrust resolved higher a year later, consistent with the literature’s claim, on a sample of 6.

The near-miss problem — and the two famous 0.612s

The signal’s defining feature isn’t the thrusts; it’s the near-misses. Since 2010 our universe armed and surged to 0.55 or better 44 times while completing only 6 thrusts. And the two most famous recent ZBTs are the sharpest exhibit: November 3, 2023 and April 24, 2025 — both widely celebrated NYSE-data thrusts — peaked at 0.612 on our universe on those exact days, three thousandths short of the 0.615 line. Neither reading is wrong. The advance ratio is universe-dependent, the completion line sits inside the noise band between universes, and every cycle’s “did it fire?” argument is this arithmetic wearing different data. Both near-miss dates were, for what it’s worth, excellent buying moments anyway — which says the washout-plus-surge shape mattered more than the third decimal.

Where it will mislead you

Four traps. Universe dependence — thrust lists disagree at the margin by construction; treat any single list (including ours) as one lens, not the referee. Sample size — 6 graded events here, ~20 in the literature’s 80 years; a perfect record on twenty coin-sized samples is encouraging, not statistical proof. The absence fallacy — 1,915 sessions without a thrust is not bearish: the signal requires a washout first, and grinding bull markets never arm it (the 1984–2009 drought contained two of history’s great bull runs). And near-miss discretion — the temptation to round 0.612 up to a signal, or to widen the window to eleven days, is how a mechanical rule degrades into a story. We publish the misses instead.

How we checked it

Everything computed here comes from one pipeline artifact: the daily advance ratio and its 10-day EMA on our common-stock universe, the completed-thrust list with forward SPY returns, and the raw series the near-miss scan runs over (an armed day starts a window; the best EMA print in the next 10 sessions classifies it). Forward returns are price returns graded only where the full window exists. Pre-2010 numbers are quoted from the sources below, never recomputed, and labeled as theirs. The universe difference from Zweig’s NYSE data is disclosed wherever it changes a conclusion.

Frequently asked questions