Key takeaways · 5
- The Hindenburg Omen flags an internally split tape: an unusual number of stocks at fresh 52-week highs and at fresh 52-week lows on the same day, while the index is still in an uptrend.
- It descends from Norman Fosback's High-Low Logic Index. Jim Miekka assembled the omen in the 1990s and Kennedy Gammage gave it the name, after the 1937 Hindenburg airship disaster.
- The textbook criteria: the smaller of the new-high or new-low count tops about 2.2% of issues, new highs run no more than twice new lows, the index sits above its level of 50 days ago, and the McClellan Oscillator is negative.
- As a trigger its record is poor. The omen fires far more often than the market falls, single readings are close to noise, and even clustered signals over-warn. Most crashes had one beforehand; the great majority of omens had no crash after.
- tickerstance publishes no Hindenburg flag. Its ingredients, new highs minus new lows, the McClellan Oscillator, and the advance/decline line, all live inside the Breadth subscore, so the split is legible without the omen's false alarms.
What the Hindenburg Omen is
Most breadth warnings are one-sided. They fire when the new-low list fills up and the new-high list empties out, the ordinary signature of a market rolling over. The Hindenburg Omen looks for the opposite oddity: a day when both lists are long at once. A meaningful slice of the market is printing fresh 52-week highs and, at the same time, a meaningful slice is printing fresh 52-week lows. The tape is pulling apart.
That only counts as an omen if it happens while the index is still rising. A split tape in the middle of a decline is just a decline. The signal is meant to catch the split happening under a calm or climbing surface, when the headline index looks fine but the internals have stopped agreeing with each other. The reading is a question about coherence: is the advance still one market, or has it quietly fractured into a group ripping higher and a group breaking down?
The figure shows the shape of a qualifying day. Both the new-high and new-low counts clear the threshold, and the four boxes that turn an odd day into a named signal are ticked. The name is deliberately theatrical. Kennedy Gammage borrowed it from the 1937 Hindenburg, the disaster you are supposed to read as avoidable in hindsight. The name is built to summon that dread, and as the record shows, to oversell it.
A qualifying day: both the new-high and new-low counts clear about 2.2% of issues while the index is still in an uptrend and the McClellan Oscillator is negative. All four conditions on one session make one omen day.
Where it comes from
The root idea is Norman Fosback's. In the 1970s he built the High-Low Logic Index, which takes the lesser of the new-high or new-low count, divides it by the number of issues traded, and smooths the result. The logic is in the word lesser. In a healthy market one list is long and the other is short, so the smaller of the two stays low. The index only rises when both lists get long together, which is the unusual, incoherent state worth flagging.
Jim Miekka, a physics teacher who traded, took that core and bolted on conditions through the 1990s to sharpen it into a dated signal: the 2.2% threshold on both lists, the cap that keeps new highs from running more than twice new lows, the requirement that the index still be in an uptrend, and the negative McClellan Oscillator. Kennedy Gammage, who wrote a market letter, supplied the name. The branding outran the math, which is part of why the omen is famous out of proportion to how well it works.
It helps to see the lineage as a single thread. Fosback noticed that simultaneous extremes are strange. Miekka turned strange into a checklist. The financial press turned the checklist into a headline that reappears every time the conditions trip, usually with a chart of 1987 or 2008 attached and the false alarms left off.
Why a split tape is supposed to matter
The reasoning is intuitive even if the record is not kind to it. A market in a clean uptrend should have its leadership pointed one way. Lots of new highs, almost no new lows. When the new-low list swells at the same time as the new-high list, money is leaving one part of the market as fast as it is crowding into another. That kind of rotation, the argument goes, often shows up late in an advance, when the obvious winners get more crowded and everything else quietly breaks.
Pairing the split with a negative McClellan Oscillator is what gives the omen its teeth on paper. The McClellan reading is a momentum-of-breadth gauge, and a negative value says the average stock is already losing steam under the surface. So the full condition is a market that is split at the extremes and soft in the middle while the index still prints green. Stated that way it sounds ominous, and once or twice a generation it has been.
The trouble is that an internally mixed tape is also just normal a lot of the time. Sector rotation, a rate move that hammers bond proxies while growth runs, a few blown-up names in an otherwise fine market: all of these light up both lists without anything bad following. The omen cannot tell a genuine top from routine churn, because at the moment it fires the two look identical. That ambiguity is the whole problem, and the historical record only deepens it.
Its record: fires far too often
The Hindenburg Omen's problem is the false positive, and it is not a small one. It triggers many times more often than the market actually falls, so the unconditional hit rate is low. A single omen day, on its own, carries almost no information; the conditions trip during plenty of perfectly ordinary stretches. This is the standard finding, not a fringe complaint, and even the omen's own followers treat a lone signal as close to meaningless.
The usual fix is to demand a cluster: two or more qualifying days inside about a 36-day window, with the signal treated as live for the next 30 trading days and only in effect while the McClellan Oscillator stays negative. Clustering does cut the noise, and it is true that the famous declines, 1987, 2000, 2008, came after clusters rather than isolated days. But the implication runs one way only. Most clusters still resolve into nothing. Pointing at the crashes that had an omen while ignoring the omens that had no crash is selection bias with a chart attached.
So the figure is the fair picture. The omen fires again and again across a long advance, and the market mostly keeps rising. Only the cluster near the actual top precedes the decline, and you cannot tell that cluster from the others at the time. That is the right way to hold it: a condition worth noting, not a trigger worth trading.
The base rate, schematically. The omen fires repeatedly through the advance and the market keeps climbing past most of them. Only the cluster near the top precedes a fall, and at the time it looks like all the others.
Titanic Syndrome and the divergence family
The Hindenburg Omen is one of a small family of signals built from the same new-high and new-low counts. The Titanic Syndrome, named by Bill Ohama, fires when new lows outnumber new highs within about a week of the index making a fresh 12-month high. The Hindenburg looks for both lists long at once; the Titanic looks for the lows overtaking the highs right at a peak. Different triggers, same underlying worry, that the extremes are quarreling with the headline.
All of these are cousins of the plainer breadth reads. The new highs minus new lows count is the raw material, just netted instead of compared. The advance/decline line asks the broader version of the same question across the whole tape rather than the extremes. The classic breadth divergence, the index at a new high while net new highs make a lower high, is the slow-motion relative of the Hindenburg's single split day. They are all trying to hear the same thing: participation thinning or fracturing while price holds up.
The value in knowing the family is mostly defensive. When a market scare is on, these names get quoted as if each were an independent alarm, when they are really several views of one breadth picture. Counting them as separate confirmations double-counts the same evidence. Better to read the underlying new-high and new-low data directly and treat the omen as one colorful framing of it rather than a verdict.
How tickerstance uses it
tickerstance does not compute a Hindenburg Omen flag or post a roster of omen days. It is a market-level signal with a weak record as a trigger, and flashing it would cut against the whole point of the dashboard. What the site does carry is every ingredient the omen is made from.
The split the omen looks for is visible in the Breadth subscore directly. New highs minus new lows nets the two counts the omen compares; the McClellan Oscillator gives the momentum-of-breadth reading the omen requires to be negative; the advance/decline line tracks the broader participation underneath. Read together on the breadth page, those three say whether the tape is coherent or fracturing without packaging it as a countdown to a crash.
So the useful move is to watch the conditions and skip the omen. If new lows are swelling alongside new highs while the McClellan reading turns negative under a calm index, the market is doing the thing the Hindenburg is built to catch, and the Breadth subscore already reflects it. Most of them fizzle. The omen cannot tell you which ones will not, and tickerstance will not pretend otherwise.
Frequently asked questions
What is the Hindenburg Omen?
It is a market-breadth warning that fires when an unusual number of stocks hit fresh 52-week highs and fresh 52-week lows on the same day, while the index is still in an uptrend. The idea is that a market splitting in two directions at once has lost its internal coherence, a state that has sometimes preceded sharp declines.
What are the criteria for a Hindenburg Omen?
The classic version: the smaller of the new-high or new-low count exceeds about 2.2% of issues traded, new highs are no more than twice new lows, the index is above its level of 50 trading days ago, and the McClellan Oscillator is negative. A signal is usually only taken seriously when two or more qualifying days cluster within about 36 days.
Why is it called the Hindenburg Omen?
Kennedy Gammage named it after the 1937 Hindenburg airship disaster, as a metaphor for a foreseeable catastrophe. The indicator itself was assembled by Jim Miekka from Norman Fosback's earlier High-Low Logic Index. The dramatic name is part of why the omen draws more headlines than its track record earns.
Does the Hindenburg Omen actually work?
As a trigger, not well. It fires far more often than the market falls, so most signals are false alarms, and a single omen day carries almost no information. Clustering several signals together improves it, and the famous crashes did follow clusters, but the great majority of clusters still lead nowhere. It is best read as a description of an unusual market state, not a sell signal.
What is the difference between the Hindenburg Omen and the Titanic Syndrome?
Both are built from new-high and new-low counts. The Hindenburg Omen fires when both lists are long at the same time in an uptrend. The Titanic Syndrome, named by Bill Ohama, fires when new lows overtake new highs within about a week of the index making a fresh 12-month high. Different triggers, the same underlying worry that the extremes disagree with the headline.
Does tickerstance track the Hindenburg Omen?
No. tickerstance does not publish a Hindenburg flag, because its record as a trigger is poor and the site reports conditions rather than calling tops. The ingredients are all on the dashboard, though: new highs minus new lows, the McClellan Oscillator, and the advance/decline line, which you can read together on the breadth page.