Home › Learn › Lesson 29

Wedges — falling, rising and expanding

By TradingPrimer · Published 4 Sep 2026

Most guides hand you the picture and the direction and stop there: rising wedge bearish, falling wedge bullish, wait for a close beyond the edge with volume. All of that is true, and all of it is in our course notes. None of it answers the two questions that decide whether the trade is worth taking — how much does it pay, and when should I take it. Those turn out to be the same question, and its answer is a single number you can read off any chart in about four seconds. This lesson works that number out, uses it to explain why the expanding wedge is the one pattern the course never gives you a rule for, and then shows the shape that looks almost identical to a rising wedge, converges by exactly the same amount, pays exactly the same 2.00R, and is a buy.

Schematic only — deliberately drawn without a price scale, because nothing in it is meant to be measured. Every number in this lesson is on the charts further down, which are drawn from the prices themselves.

What makes a shape a wedge rather than a triangle or a flag?

Both of its boundaries lean the same way. That is the whole definition, it is the only converging shape for which it is true, and it is what every other question in this lesson depends on.

Once price starts making swings inside two straight lines, there are only so many ways those two lines can sit. Lesson 13 taught you to draw them; this table says what you have drawn once you have.

Read the last three rows together and the family makes sense. All three wedges have boundaries leaning the same way; what separates them is which of the two leans harder. Lean harder on the bottom and the gap closes upward-tilted — a rising wedge. Lean harder on the top and it either closes downward-tilted — a falling wedge — or, if both are rising, prises itself apart into an expanding wedge.

Two boundaries of the same shape can be a long way apart in what they mean. That is worth keeping in mind before the next section, because the difference between the left panel above and an ascending triangle comes down to the slope of one line — and it reverses the trade.

Triangles get their own treatment in Lesson 32. Flags, pennants and rectangles — the shapes with a flagpole in front of them — are in Lesson 31, which also introduced wedges as members of the same “no pole” family. This lesson takes that introduction and prices it.

Why does a rising wedge fall when everything about it is going up?

Because the thing it measures is not direction, it is the rate at which direction is being bought. Our course notes put it in one line: the pattern is made of higher lows and higher highs, but each high is not far above the last — the uptrend is running out of strength.

Turn that into the two numbers on the left panel above and the sentence stops being a metaphor. Across that wedge the lows climb $1,667 every swing, three equal steps. The highs gain $939, then $831 — smaller each time. Buyers are committing more between each attempt and winning less ground with each attempt. That is the same effort-versus-result reading you met in Lesson 14, except it is drawn in price rather than in volume bars, which makes it easier to check and harder to argue with.

Lesson 31 introduced that reading as part of the no-pole family and this lesson takes it as given; what follows here is what the reading is worth. The falling wedge is the mirror in both geometry and meaning: sellers keep dragging price to lower lows, but the rebounds are giving up less and less ground each time, so the selling is the thing running down.

Now the part that confuses people, and it confuses them for a good reason: our own course materials file wedges in two different places. The reversal-patterns index puts the rising and falling wedge alongside double tops and head and shoulders, under trend reversal. Yet the schematic on a later slide labels them “often appears during an up-leg in the trend” and “during a down-leg” — which is a description of a continuation pattern. Both filings are common in the literature and both are defensible.

The resolution is that the argument does not touch the trade. A falling wedge breaks upward because of the way it is shaped, not because of what came before it. Find one at the end of a downtrend and the break upward is a reversal. Find the same shape inside an uptrend, forming as a pullback, and the identical break upward is a continuation. Same geometry, same direction, same entry, same stop — only the label you write afterwards changes. Filing is bookkeeping. Slope is the trade.

What is a wedge actually worth?

Less than the shape suggests, and exactly how much less is calculable. The classical measuring rule has two halves, and both are ordinary:

• The target is the height of the wedge at its widest point — the width it opened with — projected from where price breaks out. Call that width W₀.
• The stop goes past the opposite boundary at the moment of the break, plus a buffer so that ordinary noise does not take you out. Call the width still left at that moment W_t and the buffer b.

Which means the trade is fully described before you place it:

Take that chart line by line. The wedge opens with its lower edge at 60,000 and its upper edge at 64,000, so W₀ = $4,000. Price makes its swings; the lower edge climbs faster than the upper one; by the time price closes back below the lower edge, that edge has reached 65,000 and the upper edge is at 66,600. The width left is W_t = $1,600. Add a $400 buffer and the stop is 67,000.

Now write it in general and the pattern's own dimensions disappear:

Check it against the chart: c = 1,600 ÷ 4,000 = 0.40, β = 400 ÷ 4,000 = 0.10, so R:R = 1 ÷ 0.50 = 2.00. It holds.

The useful thing about that form is what has vanished from it. The size of the wedge is gone. A $4,000 wedge and a $40,000 wedge that have both closed to 40% of their opening width and both use a buffer worth 10% of it are the same trade, priced identically, on any instrument and any timeframe. So is a wedge on gold and a wedge on a small-cap altcoin. Only two ratios matter, and both are things you can measure with the cursor before you click anything.

Which gives a test short enough to run in your head. Ask R:R ≥ 2, and the algebra says c + β ≤ 0.5:

So the four-second version, for the buffer most people end up using: if the wedge is not down to about two-fifths of the width it started with, it is not a two-to-one trade. Eyeball the gap at the left-hand end of your two lines, eyeball it at the right-hand end, and if the right-hand gap is not clearly under half the left-hand one, the arithmetic has already answered you.

Why does waiting make a wedge pay more — and where does that stop being true?

Because the reward is fixed the moment you finish drawing the pattern and the risk keeps shrinking after that. The target is the opening width, which stops changing as soon as the widest point is behind you. The stop is tied to the current width, which is still closing. Every bar the wedge survives, the same reward is bought with less risk.

That is the honest mechanism behind the course's remark that price tends to break out near the end of the wedge. It is not only an observation about where breaks happen. It is also the reason a break near the end is a better trade than the identical break in the middle.

The catch is the buffer, and it is the part nobody mentions. W_t shrinks. b does not. Your buffer exists to survive noise, and noise does not know or care how narrow your wedge has got. So as the pattern converges, the buffer becomes a larger and larger share of the risk you are actually taking, and the textbook figure — opening width divided by remaining width, which is what Lesson 31 gives for the whole no-pole family — drifts further and further above the truth.

Read down the last column and there is a line worth naming. When the width still left equals the buffer — c = β — you keep exactly half of what the textbook formula advertises. Call it the half-value point. Down to c = 0.40 the two figures stay within a fifth of each other, so patience is cheap. Past the half-value point the pattern has become smaller than the noise you are budgeting for, and although the real figure does keep creeping up, you are giving up more than half of every additional R the formula shows.

This is the same ceiling Lesson 31 puts on the whole converging family, arrived at from the other end. That lesson caps a converging pattern at its height divided by the average candle range, and takes $900 as a fair 4-hour Bitcoin figure — so $4,000 ÷ $900 = 4.44R. Set your buffer from noise, as the next section insists you should, and β becomes roughly one candle range over the opening width: 900 ÷ 4,000 = 0.225. Feed that in and 1 ÷ (c + β) tends to 4.44R as the wedge closes. The two rules are one rule. The $400 buffer used in the worked example is a deliberately tight one, so the figures in the table above are the best case for a quiet instrument rather than a general promise — on a normal 4-hour Bitcoin chart the ceiling is lower and it is Lesson 31’s.

In this example the half-value point sits at $400 of remaining width, a little under half an average candle on a chart like this. The instruction that follows is unglamorous and easy to apply: work out your buffer first, then stop waiting once the gap between the two lines is down to roughly that size. The apex is not a target. It is a place where the pattern stops existing and the formula stops meaning anything, and the last stretch before it is priced far worse than the multiplier suggests.

Why does the course never give you a rule for the expanding wedge?

Because the same measuring rule that makes the other two work turns it into a losing proposition, and it does so before any question of whether the pattern “works” comes up.

Notice first what our own course does with it. The rising wedge gets a definition, a direction and an entry rule. So does the falling wedge. The expanding wedge gets a name, a diagram and a chart — and then the slide simply ends. No direction. No entry. That silence looks like an omission. It is not.

Apply the identity. An expanding wedge has boundaries that lean the same way but pull apart, so W_t is bigger than W₀, which means c > 1. Take the panel on the right of the first chart in this lesson: it opens $4,000 wide and by the time you would be acting on it the gap is $7,200. Same buffer, same rule:

Nothing changed except the sign of the convergence, and that alone moved the bar you have to clear by 32.2 percentage points. And this is not one unlucky example. Any shape whose boundaries pull apart has c > 1, so its measured-move reward-to-risk is below 1 ÷ (1 + β) — under 1R — always, at every stage of its life, whichever way it eventually breaks. A pattern cannot be rescued by a good win rate when it needs a 65% one to break even. So the course does not hand you an entry rule for it, and the reason is arithmetic rather than oversight.

That does not make the shape useless. It tells you something specific and worth knowing: participants are disagreeing more as time passes, not less, and any stop that respects the pattern is growing while you watch it. The right response is to size the trade off something other than the pattern — a level, a structural low from Lesson 17 — or to leave it alone. What you should not do is use the wedge rules on it, because those rules are what produced the 0.53R.

If the convergence sets the size, what sets the direction?

The slope of the slower boundary — and this is the trap in the whole subject, because that slope is not what your eye goes to.

Here is the problem in one pair of charts. Both patterns below open $4,000 wide. Both have closed to $1,600 by the break. Both therefore have c = 0.40 and both pay exactly 2.00R with a $400 buffer. One is a sell and the other is a buy.

Why the convergence cannot help you here is a small piece of algebra with a large consequence. Over the pattern the lower boundary rises by some amount and the upper boundary rises by some other amount, and the gap closes by the difference between them. In the rising wedge the lows rise $5,000 and the highs rise $2,600: difference $2,400. In the ascending triangle the lows rise $2,400 and the flat top rises $0: difference $2,400. The convergence is the same because convergence only ever measures the difference — and the difference is blind to whether the second number was $2,600 or nothing at all.

So the two measurements do genuinely different jobs, and conflating them is the mistake this section exists to prevent:

Which raises the practical question: how much slope is enough to call it? A boundary is drawn through candle extremes, and those extremes scatter. If the whole rise of the upper edge across the pattern is smaller than an ordinary wick, you have not measured a slope, you have measured noise — and you cannot tell a rising wedge from an ascending triangle. Ask that the slower boundary move by more than about two typical wicks across the entire pattern before you commit to calling it a wedge. On the worked example the upper edge rises $2,600 across the pattern, which on a chart whose average candle range is around $900 is far more than any single wick, so the reading is safe by a wide margin. When it is not safe, the honest conclusion is that you do not have a trade, because the two candidates point in opposite directions.

There is also a reason the flat version behaves differently, and it is worth having, because it is the thing that makes the whole taxonomy make sense rather than just be memorised. A flat boundary is a stack of orders at one price. A sloping boundary is not. Resting orders are finite: buyers eat into them, and when the stack is gone price gaps through — which is why an ascending triangle breaks up through its ceiling. A sloping edge has no stack. Nothing is parked there to be consumed, nothing gets exhausted, and so nothing gives way; instead the side doing the chasing runs out of appetite first, which is why a rising wedge breaks down through its own steepening floor. Same convergence, two completely different things happening.

Treat that as a mechanism that explains the convention, not as a proven fact about order books — it is the reasoning behind the rule rather than evidence for it. When the “flat” level is simply where nobody happened to trade rather than where somebody parked size, there is no inventory to exhaust and the triangle reading is on much weaker ground.

What does waiting for the confirming close cost?

Between three and a half points of win rate in your favour and nearly twelve against, depending on one decision you probably have not thought of as a decision.

Our course rule is explicit for both wedges: wait for price to break the edge and close beyond the pattern, and the break must come with volume. It is good advice and this lesson is not arguing with it. But every wait has a price, and this site has now measured that price for four other patterns, so it is worth doing here too.

Same trade as before. The break happens at 65,000. Suppose the confirming candle closes 700 lower, at 64,300, and its high came back to 65,300.

The middle row is the one to look at, because it is what most people actually do. They wait for the close — sensible — and then leave the stop where the pattern originally put it, because that is where the pattern says it goes. The result is that they pay for the confirmation twice: once in the worse entry and once in the wider distance to an unchanged stop. Confirmation now has to improve their strike rate by 11.7 percentage points just to break even on the decision.

The bottom row is the same wait, done properly. Once the candle has closed beyond the pattern, its own high is a legitimate anchor for the stop — the level that has to be wrong for you to be wrong is no longer the far boundary. Move the stop to that high plus the same $400 buffer, which is 65,700 here, and confirmation stops costing anything at all; it earns you three and a half points. What you buy that with is a stop 30% narrower, which will be hit more often by ordinary noise. That is a real trade-off rather than a free lunch, and it is a trade-off you should make on purpose.

Set against the rest of the course, this is the cheapest confirmation we have measured:

There is a reason wedges come out cheapest, and it follows from everything above. A converged wedge is a narrow pattern by the time you trade it, so the confirming candle cannot travel far relative to a stop that is already close. On a double top, the confirming move is measured against a stop the full height of the pattern away, which is why it costs about two and a half times as much.

How do you check a wedge, in order?

Four steps, and the order matters because each one can end the check for free.

1. Do both edges lean the same way? If one is flat you have a triangle and the trade points the other way. If they lean opposite ways it is a symmetrical triangle and it has no direction of its own. Only if both lean the same way do you continue.
2. Does the slower edge move more than about two wicks across the whole pattern? If not, you cannot tell wedge from triangle and there is no trade to take — not a smaller one, none.
3. Measure c. Gap at the widest point, gap now, divide. Above 0.40 with a normal buffer, the pattern is not paying 2R yet. Below your buffer as a fraction, it has passed the half-value point and is paying much less than it appears to.
4. Only now look at direction and confirmation. Both edges up means you are looking for a close below the lower edge; both edges down means a close above the upper one. Require volume on the break, as the course does. And decide before the candle closes whether you will move the stop to its high — because that decision is worth 15 points of break-even win rate, and it is much harder to make honestly afterwards.

When is everything above wrong?

The arithmetic in this lesson is exact. What it rests on is not, and the difference matters.

The measured move is a convention, not a law. Every reward figure here assumes the target is the width the pattern opened with. That convention is what the classical literature and our course notes use, and it is the reason the numbers are comparable across patterns — but a market has never promised to travel the height of a shape drawn on it. Change the target rule and every R figure above changes with it. The relationships survive: narrower still pays more, diverging still pays less, the buffer still eats a growing share.

β = 0.10 is an example, not a constant. Your buffer should come from the instrument's noise, not from a lesson. That is why the 2R table is given for four different buffers rather than one, and why the half-value point is stated as c = β rather than as a number.

Real boundaries are not straight. Two lines through scattered wicks are a simplification of what price did, and c inherits every bit of that imprecision. If moving one line by a plausible amount changes c from 0.35 to 0.55, you do not have a measurement, you have a preference — and the honest response is to widen the buffer rather than pick the flattering line.

Thin markets break the stop assumption before they break the pattern. The whole model assumes you exit near your stop. On an illiquid pair the spread and slippage can be a meaningful fraction of a $1,600 remaining width, and a converged wedge is precisely the setup where a small absolute cost is a large relative one. Converging patterns on thin instruments are the worst combination this lesson describes.

The inventory explanation is a mechanism, not evidence. The reason offered for why triangles break through their flat side is a plausible account of what resting orders do. It is not something this lesson has measured, and it fails wherever the flat level reflects an absence of interest rather than a presence of size.

And none of this is a claim about how often wedges work. Every figure here is a break-even threshold — the win rate you would need for the trade to be worth taking. Nothing in this lesson measures the win rate you would actually get. That is a separate question, it depends on your market and your execution, and the only way to answer it is to record your own trades.

What are the most common mistakes here?

Trading a rising wedge long because it is going up. The shape is an uptrend; what it measures is that uptrend getting more expensive per unit of ground. Direction of the pattern and direction of the trade are opposite here, and that is the point of the pattern.

Using the textbook ratio on a very narrow wedge. Width divided by width left is close enough while the pattern is wide, and badly wrong once it is not: at c = 0.10 it says 10.00R and the trade is 5.00R. The error grows exactly where the numbers get exciting.

Waiting for confirmation and leaving the stop where it was. The most common way to make a good habit expensive. Either move the stop to the confirming candle, or accept that you have raised your own break-even win rate by 11.7 points.

Calling a triangle a wedge because the top edge looks a bit tilted. If the tilt across the whole pattern is smaller than a wick, it is not a tilt. This mistake does not cost you some of the trade — it reverses it.

Trading an expanding wedge with the wedge rules. Those rules generate a sub-1R trade on any diverging shape, every time, whichever way it breaks.

Holding on for the apex. Past the half-value point you are giving up more than half of every additional R the formula shows, and at the apex the pattern has no width left to break out of.

Arguing about reversal versus continuation. It changes what you call the trade afterwards. It does not change the entry, the stop, the target or the direction.

PRACTICE CORNER

Ten minutes, and it makes the number stick. Open a 4-hour chart of any liquid pair on the exchange you already use and scroll back three months. Every time you find two boundaries leaning the same way, do this in order: measure the gap at the widest point, measure the gap where price left the pattern, and divide the second by the first. That is your c. Write it next to the date. Then decide what buffer you would actually have used on that instrument and check whether c + β came in under 0.50. Most of them will not — and knowing what fraction of the wedges on your own chart were never 2R trades is worth more than another article about the shape.

Affiliate disclosure: the links below are partner links. We may earn a commission at no cost to you. It does not change what this lesson says. Full disclosure.

Binance→ OKX→ Bybit→

What else do people ask about wedges?

Is a rising wedge always bearish?

That is the classical reading, and it is the one this course teaches: a rising wedge is defined by a lower edge climbing faster than the upper one — buyers committing more per attempt and gaining less per attempt — and the convention is that such a shape resolves downward. It is a convention about geometry rather than a measured frequency, and nothing in this lesson tests how often it holds. What varies is what happens afterwards, and that is where the reversal-versus-continuation debate lives: a rising wedge inside a downtrend usually resumes the fall, while one at the top of an uptrend may end it. Both break down. The much more useful question is not whether the shape is bearish but whether the shape is a wedge at all, because an ascending triangle looks almost identical, converges by the same amount, pays the same 2.00R and breaks up. The tell is one line: if the upper edge does not climb by more than about two typical wicks across the whole pattern, it is flat, and you are looking at the bullish pattern rather than the bearish one.

Where exactly do the stop and the target go on a wedge?

The target is the height of the wedge at its widest point, projected from where price broke out. The stop goes past the opposite boundary as it stands at the moment of the break, plus a buffer for ordinary noise. On the worked example in this lesson, a wedge that opened $4,000 wide and had closed to $1,600 by the time price broke below 65,000 gives a stop at 67,000 — the upper edge at 66,600 plus a $400 buffer — and a target at 61,000, which is the entry less the $4,000 opening width. That is a $2,000 risk for a $4,000 reward: 2.00R, break-even 33.33%. The general form is worth memorising because it removes all the specifics: reward-to-risk is 1 divided by (c + β), where c is the fraction of the opening width still left and β is your buffer as a fraction of that same opening width. The size of the wedge, the price of the asset and the timeframe all cancel out.

Should I wait for the wedge to get as narrow as possible before trading it?

Up to a point, and the point is calculable. Waiting works because the reward is fixed once the widest part of the pattern is behind you while the risk keeps shrinking with the pattern — so the same target gets bought for less. But your stop buffer does not shrink, so it becomes a larger and larger share of the risk. The crossover is when the width still left equals the buffer: at that width you keep exactly half of what the textbook figure of width-divided-by-width-left advertises. On the example in this lesson, that is a $4,000 wedge closing to $400, at which point the formula says 10.00R and the real number is 5.00R. Before that crossover, waiting is cheap and worth it. After it, you are giving back more than half of every extra R you appear to gain, and at the apex the pattern has no width left to break out of at all.

Can you trade a broadening or expanding wedge?

Not with the rules that make the other two work, and the reason is arithmetic rather than opinion. A converging wedge pays because the width you risk is smaller than the width you target. An expanding wedge reverses that: the pattern is wider at the break than at the start, so the same measuring rule gives a reward smaller than the risk. Worked through with the same numbers used elsewhere in this lesson — a pattern that opens $4,000 wide and has widened to $7,200, with a $400 buffer — the trade is 0.53R and needs a 65.5% strike rate to break even, against 33.3% for the converged version. That is a 32.2-point difference produced by nothing but the sign of the convergence, and it holds for any diverging shape at any stage of its life. It is almost certainly why our own course names the expanding wedge, draws it, and then gives no entry rule for it at all. The shape still tells you something — disagreement is widening, and any pattern-respecting stop is growing while you watch — but if you want to trade around it, size the trade from a level or a structural low instead.

Educational content only — not financial advice, and not a trade recommendation. Every figure on this page comes from one worked model built for this lesson and each can be reproduced from the numbers given: a rising wedge whose lower edge climbs from $60,000 to $65,000 while its upper edge climbs from $64,000 to $66,600, so it opens W₀ = $4,000 wide and has W_t = $1,600 left at the break; a $400 stop buffer; a target of the opening width projected from the break; and, for the diverging case, the same pattern widening to $7,200 instead. The identity is arithmetic: risk = W_t + b and reward = W₀, so R:R = 1 ÷ (c + β) with c = W_t ÷ W₀ and β = b ÷ W₀, and break-even win rate is 1 ÷ (R + 1) throughout. The $400 buffer and the $900 average candle range are stated modelling assumptions, not measurements of any particular day; readers should substitute their own ATR reading, and the tables give the result for four different buffers precisely because the choice is theirs. Every percentage on this page is a break-even threshold — the win rate a trade would need in order to be worth taking. No historical hit rate or pattern reliability percentage is quoted anywhere, because none was measured. The measured-move target rule and the bearish-rising-wedge convention are classical conventions taught by this course, not laws of the market. Sources: the chart-pattern section of the TradingPrimer slide course, plus the arithmetic above.