How to Trade a Triangle
To trade a triangle, require at least two touches on each converging boundary, identify which type it is by which side is horizontal, and enter on a close beyond one boundary. Patterns that reach the apex without breaking have expired rather than resolved.
A triangle is price consolidating between two converging boundaries. There are three versions and they are distinguished by which boundary is horizontal, which is the part that carries the information about what is actually happening.
Before you start
At least two touches on each boundary, because two points is a line and not a pattern. Four touches minimum, and they have to be real reactions rather than passing bars.
A decision about which of the three types you are looking at. Flat top, flat bottom, or neither. The answer changes what the pattern claims.
An apex date, since a triangle that runs to its point has stopped being one. Converging lines meet. After that there is no pattern left to trade.
The steps
1. Require two touches on each side
A line through two points is arithmetic. A line touched twice and respected is a boundary. Without this rule, converging lines can be drawn on almost any consolidation.
2. Identify which side is flat
A flat top with rising lows means buyers are paying more each time while sellers hold one level. A flat bottom with falling highs is the reverse. Neither flat is a symmetrical triangle and says less.
3. Mark the apex date
Where the two lines meet is a date, and it is computable as soon as both boundaries exist. That date is the pattern’s expiry.
4. Wait for a close beyond a boundary
On this site’s shared series the median bar range is 0.493 and the ninetieth percentile is 1.101. A wick through a converging boundary is well inside ordinary movement.
5. Expect a false break first
Price frequently pushes through one boundary, fails, and then resolves the other way. That is common enough that a plan for it belongs in the trade rather than being a surprise.
6. Take the stop from the opposite boundary
Price returning inside the triangle and reaching the other side has invalidated the break. That distance narrows as the apex approaches, which is one reason late breaks are traded smaller.
7. Abandon it at the apex
If the apex arrives without a break, the pattern is done. Extending the lines beyond it produces a shape that will eventually be broken, which is not the same as a pattern resolving.
How to tell it worked
Each boundary has at least 2 touches, counted rather than assumed.
The type was identified before any directional expectation was formed.
The apex date was computed, and 0 trades were taken after it passed.
And entries came on a close beyond a boundary, not on a wick through one.
What the three types actually claim
A flat top with rising lows says buyers are getting more urgent. Each pullback is bought higher while one price level keeps being defended. That is a directional claim with a mechanism behind it.
A symmetrical triangle claims nothing directional. Both sides are converging, which describes narrowing range and says nothing about which way it resolves — so it is a volatility observation rather than a directional one.
Why the apex date matters
A triangle is a claim about compression resolving. The compression is real and measurable: the range genuinely narrows toward the apex.
But the lines meet. Past that point price cannot be inside the triangle, so the structure resolves by arithmetic rather than by anything the market did.
Which means a break that happens near the apex carries much less information. Price had nowhere else to be. Breaks in the earlier part of the structure, while there was still room inside it, are the ones where the boundary was genuinely chosen over the alternative.
Sizing inside a shrinking structure
The stop distance narrows as the apex approaches, because the opposite boundary is getting closer. That is arithmetic, and it has a direct consequence most people never apply.
A fixed position size therefore risks less late in the pattern than early in it. Same number of units, shorter distance to invalidation, smaller loss if wrong — which sounds fine and quietly means the trades you take late are the ones carrying least weight.
Sizing from the distance corrects it. Risk figure divided by entry-to-opposite-boundary gives a larger position late and a smaller one early, at the same amount at risk throughout.
Whether you want that is a real question. Late breaks carry less information, so a larger position on them may be the wrong direction — in which case the honest answer is to stop trading the pattern past some fraction of its width, rather than to trade it small and pretend that is a filter.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 16 mention triangles in the title,
at a median of just 354 views across 16 channels, and 62% of those titles are instruction-shaped.
Double tops appear in 53 at 12,299 and flags in 37 at 2,702. The counts come from
site/corpus_count.py.
16 videos at a 354 median — the smallest audience per video of any named pattern measured here. A pattern requiring four touches, a type identification and an expiry date is considerably harder to demonstrate in a short video than two peaks, and the coverage numbers show exactly that.
The answer to the question on that chart is that one touch is not a boundary. A line through a single high can be drawn at any angle you like — which means the triangle it produces was chosen by you rather than found, and the break it eventually produces proves nothing.
When it fails
The failure is the triangle drawn on a consolidation that had no boundaries, and it is available on every chart. Any sideways stretch has a highest high and a lowest low, and lines through them will converge if you pick the right points. The resulting shape looks exactly like a real triangle, price eventually leaves it because price eventually leaves everything, and the break gets recorded as the pattern working — when what actually happened is that four arbitrary points were connected.
The second failure is trading after the apex. The pattern has expired.
A third is acting on a wick. That is inside ordinary movement.
A fourth is treating a symmetrical triangle as directional. It is not.
A fifth is no plan for the false break. It happens often.
And a sixth is a fixed position size near the apex. The stop distance has shrunk.
Related
Ascending triangle is the flat-top version and what it claims. Symmetrical triangle is the non-directional one. And breakout is the event all three resolve into.
The rule that stopped me drawing triangles everywhere was requiring two touches per side. With one touch each I can draw a triangle on any chart at any time, and I did — every consolidation had converging lines on it because converging lines can always be drawn through four arbitrary points.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.