What is a double top and double bottom pattern?

A double top has four landmarks, and the order in which they matter is the reverse of the order in which they form. First the initial peak, made when an advance runs into enough supply to stop it. Then a pullback to a trough. Then a second rally to a level comparable with the first, which fails. And finally the neckline, drawn horizontally at the low of that trough, whose break is the only landmark that establishes anything at all.

Look at that list again and notice how weak the first three are as evidence. An advance that stops at a level is what every advance does; a market that never stopped would be a market with no sellers in it. A pullback is the ordinary consequence of that stop. A return to the same level is what price does when the level is where the recent business was transacted. Put them in sequence and you have a description that fits an enormous number of charts, most of which never become anything. That is the honest starting point, and it is the opposite of how the pattern is usually introduced.

The phrase that carries the most false precision is comparable height. The two peaks do not need to match to the rupee, and the conventions in circulation put the working tolerance at somewhere around two to three percent. Treat that as a rough band rather than a threshold: different sources set it differently, no exchange or standards body defines it, and a scanner that returns a tidy list of double tops is returning its author's cutoff and not a fact about the market. What matters is qualitative and it is not really about equality at all. The second attempt must fail to make a meaningful new high. A second peak that pushes well clear of the first is not a failed retest, it is a higher high, and the advance is intact.

The classical description asked for something else that modern retail versions almost always drop, and it is the more useful requirement. Edwards and Magee, writing about daily stock charts, put weight on separation rather than on equality: they described genuine double tops as developing with the two peaks two to three months or more apart, treated cases only two or three weeks apart as the unusual end of the range, and expected a substantial decline in between, noting the shallow extreme at a valley of roughly fifteen percent. Read that alongside the shapes people trade today, formed over eight bars on a fifteen-minute chart, and the mismatch is obvious. The classical pattern was a statement about a market that spent months failing at a level. Compress the time and the depth far enough and the picture survives while the thing the picture was evidence of does not.

The same logic runs through the other classical reversal structures. A head and shoulders pattern is the three-test version of the same argument, with the same dependence on a neckline that has to give way before the shape means anything, and reading one properly is most of the work of reading the other. If you want the level itself examined rather than the picture drawn around it, our guide to supply and demand zones covers why some prices keep producing sellers.

Here is where the money goes. Price rallies back to a level it failed at three weeks ago, stalls there, and prints a weak bar. On the screen the structure looks finished: two peaks, a clean level, an obvious short. The trade is taken at the second peak, with a stop just above it. It has an attractive shape on paper, because entering at the top of the range and targeting the projected move gives a flattering ratio between what is risked and what might be made. Everything about it feels like anticipating rather than chasing, and anticipating is what good traders are supposed to do.

The flaw is not in the execution. It is in the premise. A short taken at the second peak is a position that requires a pattern to form in order to be right, and the pattern has not formed. What has been observed is a level that produced sellers twice. What is being bet on is that price will now fall far enough to close beneath the trough, an event that has not happened and may never happen. The trader has converted a conditional into a forecast and then risked money on the forecast. It is worth saying this in the plainest possible terms, because the alternative framing is so seductive: two peaks that hold and then break upward are not a failed double top. They were never a double top at all. There was nothing to fail.

Follow the figure carefully, because the detail that matters is what the losing trader did right. The stop was placed at a sensible level, above the second peak, where the failed-retest idea would be genuinely invalidated. The position was sized from that stop rather than the stop being tightened to fit a bigger position. The loss, when it came, was the loss that had been planned. Nothing in the risk management was careless. The trade was still a mistake, and the mistake was upstream of every one of those decisions: the reason for being in it was a pattern that did not exist.

The trader who required a close below the neckline did not make a better prediction. That is the part people resist. The waiting trader had no idea the market would rally either, and gained nothing from foresight. What the rule did was make an unformed pattern unable to produce a position. The trigger never printed, so there was nothing to be right or wrong about, and the capital was still there for the next structure that did complete. Over a long sequence of charts that is the whole of the difference: not a better read, but a smaller set of situations in which a read can cost anything.

Anticipating the second peak is not an early entry into the pattern. It is a trade in a different instrument entirely: a bet on the pattern forming, priced as though it already had.

There is an honest objection to make here, and it deserves a straight answer rather than a dismissal. Waiting for the close does cost something. The neckline break often comes with a gap or a fast bar, so the entry is worse than the level a shape-first trader could have sold, and the stop that goes with it is wider. That is real, and it is the price of the rule. What the rule buys is that the losses which do occur are losses on structures that at least existed. The alternative buys a better average entry on a population of trades that includes every range that never resolved, and there are a great many of those. This is a trade-off between entry quality and premise quality, and premise quality is the one that compounds.

Everything above turns on one word, so it needs defining properly. Confirmation of a double top is a decisive close below the neckline. Three parts of that phrase are doing work.

Close, not touch. Price that trades below the trough during a session and settles back above it has not confirmed anything. It has demonstrated the opposite: that selling below the level could not be sustained to the end of the period. This distinction is not pedantry. It is the difference between a level that gave way and a level that was tested and held, and the two look identical for most of the session in which they occur. The close is where the market records which one happened.

Decisive, which is a judgement and should be admitted as one. There is no standard threshold. Common practice is to require the close to clear the neckline by some margin scaled to the instrument's own volatility, or to require the bar to close in the lower part of its range, or to ask for a second close beneath the line. Any of these is defensible; what is not defensible is deciding which one you meant after seeing the bar. Write the rule before it triggers, and the ambiguity becomes a parameter instead of an excuse.

Below the neckline, which means the trough between the two peaks, drawn horizontally, and nothing else. Not below the second peak. Not below a trendline connecting the lows. Not below some round number nearby. The trough is the level whose loss changes the structure of the chart, because it is the last point at which buyers stepped in on the way to the second attempt.

Because a break is a break, the machinery of failed ones applies here in full, and it is machinery worth understanding on its own terms: the clustering of resting orders around an obvious level, the sweep that takes them and reverses, the reclaim that turns the whole thing into evidence for the other side. Our guide to breakouts and how they fail works through that in detail, and everything it says about levels generally is true of necklines specifically. The point to carry back here is narrow: a neckline is the most watched line on the chart precisely because everybody reading the pattern is watching it, and the most watched line is the one most worth probing.

Set the trap aside and look at the version that does complete, because it deserves to be drawn honestly rather than idealised. Price makes its first peak, pulls back, returns, fails, and then closes decisively beneath the trough. At that moment three things are true simultaneously, and only the first is a fact about the market: the structure is now a double top; a projection can be calculated; and a population of traders who bought the second push are now holding losses with their protective stops sitting just under the line that has broken.

That third point is the mechanism, and it is why a genuine break tends to accelerate rather than drift. The buyers of the second peak were late longs, chasing what looked like a fresh push to new highs. Waiting for them at the level were two sources of supply: whoever capped the first peak and is still willing to sell there, and the earlier buyers who sat through the pullback and now take the chance to leave at breakeven. The retest fails because demand from people chasing the move is smaller than supply from people who already know the level. When the neckline goes, the stops of the late longs become market sell orders, and the pattern briefly feeds on the traders it created.

The volume pane in that figure carries the evidence most explanations mention in one sentence and then never use. The classical expectation is that the second push forms on lighter volume than the first, because fewer participants are willing to pay up at a level that has already rejected once, and that volume then expands on the break as trapped longs and fresh sellers arrive together. Both halves are readable in advance of the outcome, which is what makes them useful rather than decorative.

Be precise about what that evidence can and cannot do. Lighter volume at the second peak is corroboration that the retest is thin; it is not confirmation, and no volume shape has ever completed a pattern. Expansion on the break is consistent with real participation rather than a drift through an empty level, but volume on a single instrument is noisy, expiry and index rebalancing distort it, and a clean close beneath the neckline on unremarkable volume still counts while a beautiful volume profile with no close does not. The right hierarchy is simple: the close decides, and volume adjusts how much weight the decision carries. If reading turnover properly is unfamiliar ground, our guide to volume in trading covers what it does and does not measure.

A shape without volume is still a pattern if it closes through the line. Volume without the close is just a busy day at a level that held.

The projection attached to the pattern is the measured move: take the height from the peaks down to the neckline and project that same distance below the break. It is worth understanding where that comes from, because the derivation tells you exactly how much to trust it. The pattern height is a measurement of how far this instrument swung while it was building this structure. Projecting it forward is a statement that the move which follows may be of a similar size to the moves that just occurred. That is a reasonable prior drawn from the market's own recent volatility. It is not a promise, not a level the market has agreed to visit, and not a price at which anything is obliged to stop.

Drawing it as a single line is the source of most of the trouble, because a line invites the question of whether the target was hit, which is the wrong question. Drawn as a zone, as in the figure above, it says what it actually means: somewhere in this region is a reasonable expectation for follow-through, many moves fall short of it, some run well past, and the number is an input to a decision rather than the decision. The arithmetic below uses the pattern measured off that figure, and its real subject is not the target at all. It is the trade-off between where the stop goes and how large the position can be.

Read the two columns as a single sentence and the pattern's real lesson appears. The wide stop is placed where the idea is actually wrong, which is why it survives noise, and it caps the position so hard that the projected move barely covers the risk. The tight stop makes the arithmetic look wonderful and sits exactly where a false break would take it out. Neither column is the correct answer, and any page that tells you one of them is has stopped describing the market and started selling a method. The choice is a judgement about the specific chart, the instrument's own volatility, and how much of a knocked-out position you can absorb without changing behaviour. Deciding where an idea is genuinely wrong, and sizing from that distance rather than from the position you wanted, is the discipline the method we teach is built around.

One arithmetic point is worth stating explicitly because it survives every change of instrument and timeframe. A stop is a trigger, not a guaranteed exit price. If the market gaps through it, the loss is larger than the one in the table, and it is largest in exactly the conditions that produce the fastest neckline breaks. The planned risk is a plan, not a floor.

Everything above inverts cleanly. A double bottom forms after a decline: price falls to a low, bounces to an intermediate high, falls a second time to roughly the same area, and holds. This time it is the sellers who fail the retest, unable to force a meaningful new low at a level where they had been winning, while buyers who defended the first low return. The shape resembles the letter W, as a double top resembles an M, and that mnemonic is the only part of the pattern anybody remembers under pressure.

The neckline runs across the intermediate high between the two troughs, and the pattern is confirmed only by a decisive close above it. The height projects upward from the break as an expectation zone. The stop references mirror exactly: below the second trough, wider and harder to knock out, or just below the neckline, tighter and more exposed. Every sentence in this guide can be read in both directions by swapping the sign.

Two asymmetries are worth carrying across, and they are matters of texture rather than of logic. The first is that bottoms are often slower than tops. Markets fall on urgency and rise on the gradual return of willingness, so the second trough frequently takes longer to form than the second peak did and the structure looks less crisp while it is building. The second is that the volume evidence reads slightly differently: at a top, the tell is the absence of buying on the second push, while at a bottom the more informative signal is usually the expansion of buying on the break rather than the quietness of the second test. Neither changes the definition. The close through the neckline is still the whole of the confirmation.

If the neckline is what makes the pattern, then most of what goes wrong is a variation on the neckline not doing what the shape promised. Three cases account for the great majority of the damage. All three look like textbook double tops while they are forming, which is exactly why they are worth studying in the form they take before the outcome is known rather than after.

The third push is the most common and the least dramatic. Price returns to the level a third time, the neckline stays intact, and the structure that was going to be a double top is now a range with three touches, or possibly a triple top, or possibly the base of a continuation. The correct reading is that the level has not resolved. There is a strong pull toward calling it a top anyway, because two touches had already produced a name and the third feels like more evidence for the same conclusion. It is not. Repeated touches with the line intact are evidence that the market has not decided.

The shallow break that is reclaimed is the expensive one, and it is deliberate as often as it is accidental. Everyone reading the pattern knows where the trigger sits, so orders pile up around it: the stops of the trapped longs just below, and the fresh sell orders of pattern traders. A push that dips just beneath the line can take all of it and reverse, leaving the sellers who acted on the break stranded as price reclaims the level and holds above it. The response is not cleverness but definition. A break judged on a decisive close rather than an intraday poke filters most of these, and the reclaim itself is genuine information: a level that was lost and immediately taken back is a level with buyers behind it.

The shape too small to mean anything is the quietest failure and the most common in practice, because it is manufactured by the trader rather than by the market. Drop to a low enough timeframe and two-peak silhouettes appear continuously, in every instrument, all day. Each one is a real shape. Almost none of them is the structure the classical description had in mind, which asked for months of failure at a level and a substantial decline in between. Reading a fifteen-minute double top inside an intact daily uptrend is not applying the pattern at a smaller scale; it is applying the name to something the pattern was never a description of.

That last row deserves its place in the table even though it is not a failure of reading. A confirmed double top that goes nowhere is an ordinary result, and treating it as evidence that the pattern is broken leads to the worst possible correction: abandoning the confirmation rule because a confirmed trade did not pay. The rule was never a promise about outcomes. It was a filter on which situations are allowed to produce a position.

Once the neckline has actually gone, there are two defensible ways to act, and they are not interchangeable. A break entry takes the position on the confirming close. A retest entry waits for price to return to the broken neckline from the other side and takes the position only if the level now rejects it. The first is certain and expensive; the second is cheap and uncertain. That is the entire trade-off, and it is a real decision rather than a matter of taste.

Neither is superior in the abstract, and a trader who switches between them after seeing how the bar closed is running neither. What matters is that the choice is made in advance and then applied to every instance, because the two doctrines produce different populations of trades, and a rule evaluated across a mixture of both tells you nothing about either.

The lineage is worth getting right, partly because most articles get it slightly wrong and partly because the correct version explains why the pattern is described the way it is. The classical vocabulary, including the term double top, comes from Richard W. Schabacker, whose 1932 work Technical Analysis and Stock Market Profits set out the chart patterns as a system. What made those names standard was the book that followed: Robert Edwards and John Magee's Technical Analysis of Stock Trends, published in 1948, which codified and extended Schabacker's material into the reference that still anchors the field. Edwards was Schabacker's brother-in-law, and the early part of the 1948 book is built directly on his work. So the honest framing is that the language was coined in the 1930s and standardised in 1948, rather than invented from nothing by either.

The tradition of measuring these structures rather than merely describing them belongs largely to Thomas Bulkowski, whose Encyclopedia of Chart Patterns catalogues variants and their behaviour across large samples. It is worth citing for the discipline of the approach and worth reading carefully rather than as scripture: the figures come from specific datasets, mostly United States equities, over specific periods, and they are extremely sensitive to how a pattern is defined and identified in software. Change the tolerance on peak equality, the minimum separation, or the definition of a decisive close, and the numbers move. The value of that work is the method of asking, not a figure to memorise.

From it comes one genuinely useful refinement, the Adam and Eve taxonomy, which classifies each extreme by its shape. An Adam extreme is narrow and sharp, often a single-bar spike that comes to a point. An Eve extreme is rounded and wide, a dome or bowl built over several bars. Because either extreme can be either type, four combinations exist. The practical reading is modest but real: a sharp extreme records a fast, emotional rejection of a level, while a rounded one records a slower churn of supply against demand. It is a vocabulary for describing the two peaks more precisely, not a separate pattern to trade.

What survives all of that scepticism is the structure of the argument rather than any statistic. A level produced sellers. Price came back and the level produced sellers again. Then the last point where buyers had defended gave way on a close. That sequence is a legible account of supply beating demand at a specific price, and it comes with a natural place to be wrong and a rough sense of scale drawn from the market's own recent behaviour. None of that requires a win rate to be useful, and none of it becomes more useful with one attached.

Strip the picture away and the double top is a small, precise claim: an advance failed at a level twice, and then the market gave up the last price the buyers had defended. The first half of that sentence is common. The second half is the pattern. Every difficulty people have with this structure comes from spending their attention on the half that is ordinary.

Which gives the discipline its final, unglamorous shape. Learn the anatomy, because you cannot see the neckline without it. Then hold the name back until the close arrives, accept that a great many promising shapes will resolve upward and cost you nothing, treat the projection as a zone rather than a destination, and let volume adjust your confidence without ever letting it substitute for the event. That is not a system, and it will not turn a chart into a forecast. It is the difference between reading a structure and being read by one.

The pattern is defined by its break. Trading it before the break is trading a pattern that does not yet exist, and the market is under no obligation to create it for you.

• Richard W. Schabacker, Technical Analysis and Stock Market Profits (1932). The origin of the classical chart-pattern vocabulary, including the term double top. This is the attribution most guides get wrong, crediting the later and more famous text with coining names that were already in print sixteen years earlier.
• Robert D. Edwards and John Magee, Technical Analysis of Stock Trends (1948). The work that codified the double top and double bottom into the standard reference of technical analysis, and the source of the requirement, quietly dropped by most modern retail treatments, that the two peaks be separated by real time with a substantial decline between them. Their description of daily stock charts treats peaks two to three months or more apart as the typical case, cases two or three weeks apart as the unusual end of the range, and a valley of roughly fifteen percent as the shallow extreme.
• Thomas N. Bulkowski, Encyclopedia of Chart Patterns. The serious statistical tradition of measuring chart patterns across large samples, and the source of the Adam and Eve variant taxonomy. Cited here qualitatively and deliberately: the figures derive from specific, mostly United States datasets and are highly sensitive to how a pattern is defined in software, so they are not reproduced here as success rates. thepatternsite.com
• The peak-tolerance convention. The widely repeated allowance of roughly two to three percent between the two highs is a working convention in general circulation rather than a threshold this page can trace to a primary text. It is presented as a rough band for that reason. No exchange and no standards body defines a double top.
• Standard neckline, volume and measured-move mechanics. The failed-retest reading, the requirement of a decisive close through the neckline for confirmation, the lighter-volume-on-the-second-push expectation, and the height-projected move reflect the settled classical description of the structure as taught across the technical-analysis literature. Every price, rupee and share figure on this page is illustrative arithmetic generated for legibility, not a claim about any instrument.