Triangles are pauses. One boundary is flat, the other slopes toward it, and the range gets tighter until price has to pick a direction.
Ascending triangle
Flat resistance on top, rising lows underneath. Buyers are willing to pay progressively more; sellers keep defending a single price. Each test of the ceiling is made from a higher starting point, which means the buyers are gaining ground.
It usually resolves upward — "usually" being the operative word. The tradeable event is the close above the flat line, and the invalidation is a close back below the most recent higher low.
Descending triangle
Mirrored: flat support underneath, falling highs above. Sellers accept progressively lower prices while buyers hold one line. It usually resolves downward.
Same rules inverted: trigger on the close below the flat line, invalidate on a close back above the most recent lower high.
Why "usually" matters
These are directional biases, not certainties. An ascending triangle in a strong downtrend is just as likely to break down as up — it may simply be a bear flag in disguise. Read the direction of the trend the triangle appeared in before you decide which way to expect the break.
Common drawing errors
- Forcing the flat line. If the "flat" side slopes noticeably, it is a symmetrical triangle, which has no directional bias at all.
- Too few touches. You need at least two touches on each side; three is better.
- Trading inside the triangle. Buying every touch of support and selling every touch of resistance works right up until the breakout, which then costs you everything the chopping made.
How to check this yourself
Find ten triangles and record, for each: which way the trend was running before it formed, which way it broke, and how many touches each side had. Pay particular attention to the ones that broke against their bias — those are the trades that hurt, and seeing them in advance is the point of the exercise.