Triple Bottom Pattern: Whose Definition Are You Using?

This page rests on five published descriptions of the pattern, each read end to end. Every one of them offers a test for deciding whether three lows qualify. No two of those tests are the same, and two of the five contradict each other outright on what volume at the breakout is worth.

That gap matters more than it looks, because on the platforms most readers use the decision is not made by the reader at all. It is made by a detector applying a rule the reader has never been shown.

Who Decided That Was a Triple Bottom

The usual description is short enough to memorise. Price falls, finds a level three times, and turns up through the high that separates those attempts. Three lows, two intervening highs, one break.

The description hides a decision. Nothing in a chart announces that three particular lows belong together, or how close to level they must be before the grouping counts. Somebody rules on that, and the ruling is where the published accounts stop agreeing.

On a platform with automatic pattern detection, the ruling has already been made and encoded. The tool decides which turning points are eligible, how far apart they may sit, when the pattern counts as complete, and what happens when two candidate patterns overlap on the same bars. What appears on screen is the output of those settings.

That is a different situation from the one every guide describes, where a reader compares a chart against a picture. It is closer to the position with the two-low version of the same shape, where the same question about tolerance arises and is answered the same way, quietly, by whoever built the tool.

What the TradingView Detector Actually Requires

TradingView publishes the rules its pattern indicator follows, in its own help documentation, which makes them checkable rather than assumed.

The search runs over a fixed window of the most recent 600 bars. Anything older than that window is outside the tool’s view, so the same chart on a different timeframe presents the detector with a different stretch of history.

Turning points have to qualify as 5/5 pivots. A pivot high, in that definition, is a bar with no higher high in the five bars on either side of it; a pivot low is the mirror of that. A low that is genuinely a low but has a slightly deeper neighbour four bars away does not qualify, and the pattern built on it does not exist as far as the tool is concerned.

The neckline is drawn along whichever of the two intervening highs is the higher one. The pattern is treated as formed only after the third low, and only when a close finishes above that line, so an intrabar spike through it changes nothing.

The documentation also gives a drawn pattern four possible states, covering whether the projected level was reached, whether price fell below the last low instead, whether the outcome is still open, and whether the tool cannot resolve it. Overlapping patterns have a published precedence rule rather than an arbitrary one.

None of that is a market fact. Every part of it is a design decision, published honestly, and a reader who has only ever seen the shape described in prose has no idea any of it is there.

Where the thinkorswim Reference Disagrees

The thinkorswim technical analysis reference describes the same pattern as three troughs of similar shape at roughly one price level, and then adds something the tolerance-based descriptions do not.

It states that the versions where the last trough sits above the middle one perform better, and that a larger advance can be expected where the first intervening high stands above the second. Neither claim carries data, a study, or any citation.

Set that beside a same-level test and the two do not fit together. A rising third low is a deviation from level, so a test built around equality within a tolerance treats it as a weaker instance or rejects it, while this reference treats it as the preferred instance. The same three lows can be the best case under one description and a marginal case under the other.

It also notes that the shape can appear after an uptrend, where what follows a break is unreliable, a caution the guides that assume a preceding downtrend never reach.

What none of the five states is that these are competing definitions rather than variations on one. A reader who learns the shape from one source and scans for it on another platform is applying two rules at once. The gap between them is one of the reasons a break past a level so often fails to hold, since the level itself was drawn by a rule that was never stated.

The Volume Question, and the Figures Nobody Sources

The clearest contradiction in the readable set is about volume at the breakout.

The ChartSchool entry holds that heavier trading as price climbs away from the third low, and again where the level gives way, makes the whole formation more trustworthy. The thinkorswim reference states that for this pattern, unlike its inverse, breakout volume does not much affect the result. Two established publishers, two opposite verdicts, no citation on either side, and no sign in either text that the other position is known to it.

On a spot forex chart the argument has a further problem, because the volume figure a MetaTrader chart displays for a currency symbol is not traded volume at all. That point belongs to our page on what the MetaTrader volume figure actually counts, which sets it out against the vendor documentation and is the right place to read it.

Here it is enough to note that a confirmation step two sources already disagree about is resting, on this market, on a quantity that measures something else.

The rest of the numbers in circulation get the same treatment on this page, which is to say none. The readable descriptions give formation periods that do not match each other, a reward ratio presented as a recommendation, and a claim that one arrangement of troughs performs better. None of it is attributed.

Where two or more publishers assert a number and no official document backs it, that number stays off the page entirely, which is why no formation period, ratio or success rate appears above. The wider habit of publishing pattern reliability percentages is taken apart on our page about the success rates published for chart patterns.

Three Equal Lows Are Equal in One Feed

There is a further wrinkle that applies to currencies and not to the equity charts most of these descriptions were written for.

A detector evaluates its rules against the bars in front of it. On a spot forex symbol those bars come from one broker’s price feed, and that feed is one venue’s view of an over-the-counter market rather than a single official record. Where prices actually come from is a topic of its own, covered on our page on price discovery in forex.

The consequence for this pattern is narrow and concrete. Whether three lows are level, whether a low satisfies a pivot condition, and whether a close finished above a line are all judged on the numbers in that feed. Two accounts at two brokers, looking at the same hour of the same pair, can be shown a pattern by one platform and nothing by the other, with neither of them wrong about its own data.

That is not a reason to distrust the chart. It is a reason to treat a detected pattern as a statement about one data set and one rule set, rather than as an event that happened in the market.

Reading the Pattern Without a Detector

Reading it by eye removes the hidden rule but not the decision underneath it.

The four points that carry any weight are the three lows and the higher of the two highs between them. The lows establish that selling stopped in the same area more than once. The higher intervening high is the level that has to give way before anything has changed, and among those four points it alone will carry a decision.

Spacing, symmetry and the general neatness of the picture are matters of taste, and not one of the five puts a defensible number on any of them.

Choosing a tolerance for level is unavoidable and best done in advance, before a chart is in front of you, because a tolerance chosen while looking at a chart tends to be the one that makes the pattern appear. The same discipline applies to reading price without indicators generally, which is set out on our page on reading a chart without an indicator.

When the Label Changes a Decision, and When It Does Not

The label earns its keep in one situation: when a scan is being run and its output has to be trusted or discarded. There, knowing which rule the tool applies is the difference between a filter and a lottery, and the rule is published for both platforms discussed above.

It earns nothing at all when a level is already on the chart. A price area that has held three times is the same area whether or not the arrangement gets a name, and the higher intervening high is the same line either way. Anyone whose decision would not change if the name were withdrawn is better served by studying the level.

Who This Page Is Not For

No reliability percentage appears anywhere above, and that absence is a decision rather than an oversight. Every figure of that kind in circulation for this pattern is unattributed, and repeating an unsourced number gives it a second source rather than a first.

It is also not for anyone wanting an entry rule, a stop distance or a position size to copy. Those depend on an account, an instrument and a tolerance for loss that no page can know, and a pattern whose definition varies by platform is a poor foundation for a rule stated to the pip.