Home › Learn › Lesson 15

Higher highs and lower lows — reading structure before anything else

By TradingPrimer · Published 2 Sep 2026

This is the vocabulary the rest of technical analysis is written in. Break of structure, change of character, trend continuation, reversal patterns, wedges, triangles — all of them are sentences built from these four words, which is why we teach them before the tools that use them. The words themselves take about ninety seconds to learn. What takes longer, and what almost no course covers, is that four labels made from two yes/no questions produce four arrangements rather than two, that a single higher high with a single higher low is a one-in-four coin flip and therefore evidence of nothing, and that each extra swing you wait for before believing the trend has a price you can calculate in advance. This lesson does all three, and finishes with the two things you actually do with structure once you can read it: where the trade goes, and where the stop moves to next.

What do HH, HL, LH and LL actually mean?

Two comparisons, repeated forever. Is this peak above the last peak? Is this trough above the last trough? That is the entire method.

Price does not move in a line; it moves in legs. Up a bit, back a bit, up a bit more. The turning points at the top of each leg are swing highs, the ones at the bottom are swing lows. Market structure is nothing more than reading those turning points in order and comparing each one to the one before it of the same kind:

• Higher high (HH) — this swing high is above the previous swing high.
• Higher low (HL) — this swing low is above the previous swing low.
• Lower high (LH) — this swing high is below the previous swing high.
• Lower low (LL) — this swing low is below the previous swing low.

The phrase to hold on to is of the same kind. A high is compared only to the last high. A low only to the last low. Beginners routinely compare a high to the low just before it, decide price is “higher”, and mark an HH where there is none. Keep the two chains separate and you cannot make that mistake: one chain of peaks, one chain of troughs, each read on its own.

The course this site learned from states the definitions in one line each — each low and high is higher than the one before for HH and HL, and the mirror for LH and LL — and then makes a claim that is worth flagging now and answering properly later: higher high and higher low together form a durable bullish structure, and the higher the timeframe the structure forms on, the more durable the trend. That is true, and section 8 shows exactly why, with a number attached.

Here is the same reading in the other direction, so the mirror is not just asserted:

One deliberate omission before we go on. Marking a swing point requires a rule for what counts as a turning point, and that rule is a setting with real consequences — two traders using different settings will mark different structure on the same chart. We cover that in full, with the arithmetic, in lesson 17 on BOS and CHoCH. For this lesson we use the obvious visual turning points, which is what the course charts above do, and section 5 gives the convention to write down.

Why are there four labels but only two bits of information?

Because “are the highs rising?” and “are the lows rising?” are separate questions with separate answers — and two yes/no questions make four combinations, not two.

Almost every treatment of market structure presents the four labels and then immediately pairs them off: HH with HL is an uptrend, LH with LL is a downtrend, the end. That pairing quietly assumes the two questions always agree. They do not. Highs can rise while lows fall, and highs can fall while lows rise, and both of those happen constantly.

Lay it out properly and the missing half of the map appears:

Now the reason this matters more than a tidy diagram should. In both unnamed boxes, one chain still reads bullish. In a widening range the highs are genuinely making new highs; in a narrowing range the lows are genuinely making higher lows. So the two most common shortcuts traders use — “is it making new highs?” and “is it holding higher lows?” — each return an encouraging yes in one of the two boxes where there is no trend at all.

That is the practical failure mode, and it is specific. A trader watching only the highs sees a broadening top as an uptrend, and is not wrong about the highs; the highs really are rising. What they have missed is that the lows are falling underneath, which means the range is getting bigger and the average trade in it is getting worse, not better. A trader watching only the lows sees a coil as an uptrend for the mirror reason. Neither is being careless in any way they could notice, because a single label always looks like an answer.

The fix is small and mechanical: never say one label, always say the pair. Not “it made a higher high” but “higher high, lower low”. The second version cannot be misread, and it takes the same breath.

One precision to keep the map honest, because it is easy to over-apply. The 2×2 sorts by direction only. A rising wedge — where the highs and the lows are both rising but the lows rise faster, so the shape converges — still sits in the uptrend box; it is a refinement inside a box based on relative slope, not a fifth box. The same is true of a falling wedge in the downtrend box. Our slide course describes exactly this: a pattern made of higher lows and higher highs where each high is not far above the last. Those shapes get their own lesson later in the path; the point here is only that the four boxes come first and the slopes come second.

How many higher highs before an uptrend means anything?

There is no number that makes it true. But there is arithmetic that tells you how impressed to be, and it is unflattering to the first one.

Use a deliberately neutral yardstick, the same kind we use elsewhere on this site: assume each new swing high is equally likely to land above or below the previous one, and the same independently for the lows. This is not a claim about how markets behave — real price series trend, and a trending chart will beat this baseline. It is a floor, and its only job is to tell you what a given pattern looks like when nothing at all is going on.

Under that yardstick, a single higher high with a single higher low is 0.5 × 0.5:

Read the top row again, because it is the sentence most beginners need. One higher high and one higher low is a one-in-four event. It is not a signal, it is the base rate. You would see it four times in sixteen on a chart with no direction in it whatsoever. When someone announces “we have a higher low, the trend is up”, they are reporting a coin flip that came up heads twice.

The second thing this table gives you is the shape of the improvement. Each extra pair divides the chance-alone explanation by four — 25% to 6.25% to 1.56%. That is a genuinely steep improvement, much steeper than most people's intuition, and it is the honest argument for patience. Three confirmed pairs is a 1-in-64 arrangement; that is worth something.

Which raises the obvious question, and it is the one nobody asks: if waiting improves the odds fourfold each time, why not wait for five? The answer is that waiting is not free, and the next section prices it.

What does waiting for one more swing cost?

About three points of break-even win rate per swing, in our worked case — and the cost is entirely mechanical, because every swing you wait for happens at a higher price.

Here is a structure in round numbers, chosen so nothing hides in the arithmetic. Bitcoin on the four-hour chart makes a low at $60,000, rallies to $66,000, pulls back to $63,000 (the first higher low), pushes to $70,000 (the first higher high), pulls back to $66,000 (second HL), pushes to $73,000 (second HH), and pulls back to $69,000 (third HL). You believe this run has $80,000 in it. The question is only where you get on.

Buy each pullback with the stop 300 below the higher low that precedes it — the same rule each time, so the comparison is clean:

Break-even win rate is 1 ÷ (1 + R:R) throughout — the identity this site uses in every lesson that prices a decision.

Now put the last two columns side by side, because together they are the answer to “how many higher highs do I need?” that no rule of thumb can give you:

• Waiting from the first HL to the second: break-even rises 2.8 points (16.3% → 19.1%). The chance-alone explanation falls from 25% to 6.25%.
• Waiting from the second to the third: break-even rises 4.0 points (19.1% → 23.1%). Chance-alone falls from 6.25% to 1.56%.

So the trade you are being offered is roughly: three points of required win rate for a fourfold improvement in the odds that this is not noise. Whether that is a good trade depends on something the chart cannot tell you — how often you are wrong about trends in the first place. If you take the first higher low every time and it turns out to be a trend one time in three, 16.3% break-even is comfortable. If your record says one in six, it is not, and the extra two swings are cheap at three points each.

Two honest caveats on that table. The risk column is constant at $3,300 only because the pullbacks in this example are the same depth each time; on a real chart they vary, and a deeper pullback widens the stop and eats the advantage of the earlier entry. And every figure ignores fees, funding and slippage, which is fine for a comparison between three entries but not fine as a description of your account.

The general shape survives both caveats, though, and it is the thing to remember: earlier entries are not braver, they are cheaper — and confirmation is not safer, it is more expensive. Those are two sides of the same subtraction, and knowing the size of it is what turns “wait for confirmation” from advice into a decision.

How do you set up a chart to read structure?

You do not add anything. That is the point, and it is worth saying plainly because it is the only tool in this course with no settings to get wrong.

Structure is read on bare price. No indicator, no subscription, no lag. But there are three choices you have to make before you start marking, and the mistake is not making them badly — it is making them differently each time you look at a chart.

Then the procedure, which takes about two minutes per chart:

1. Pick your frame and scroll back far enough to see roughly 100 candles.
2. Mark the obvious turning points only — the ones that stand clear of the candles either side. If you are squinting at it, it is not one.
3. Label the highs left to right against each other: HH or LH. Ignore the lows entirely while you do this.
4. Now do the lows the same way: HL or LL. Ignore the highs.
5. Read off the pair from the two most recent labels and place it in one of the four boxes.

Steps 3 and 4 are separate on purpose. Doing the chains one at a time is what stops you from reading a high against a low, and it is also what makes the two unnamed boxes visible — if you label both chains at once you will unconsciously make them agree.

What does structure tell you to actually do?

It tells you which side of price the next break is likely to happen on — and the course states the rule in a single sentence that most beginners have backwards.

For an uptrend to continue, it has to make higher highs. So resistance tends to break in an uptrend. And the mirror: for a downtrend to continue it has to make lower lows, so support tends to break in a downtrend. Read those twice, because they invert the instinct. Most people meet support and resistance as levels that hold, and then treat every approach to resistance as a reason to take profit or to short. Structure says the opposite: in a confirmed uptrend, the level above price is the one that is supposed to fail, because its failure is the definition of the trend continuing.

That gives you the trade, and it is deliberately unexciting:

• Direction. Take trades in the direction the pair says. HH+HL: long only. LH+LL: short only. Off the diagonal: neither — you are looking at a range, and range trading is a different tool set covered in lesson 20.
• Entry. Buy the pullback into the forming higher low, not the break of the higher high. The break is where the structure confirms; the pullback is where it is cheap. That is the whole content of section 4's table.
• Stop. Just below the previous higher low. Not below the swing you are buying — below the one before it, so that a normal wick cannot take you out.
• First target. Project the previous leg. In our example the leg from $63,000 to $70,000 was $7,000, so from an entry at $66,000 the projection is $73,000.
• Invalidation, written before entry. A close below the previous higher low. At that point the higher-low chain is broken and the reason you were long no longer exists — whether or not the stop has been hit.

The last point deserves its own sentence. The condition that gets you in and the condition that gets you out should be the same condition, read forwards and backwards. You are long because the lows are rising. When a low stops being higher than the last one, you are not long any more. That is a cleaner exit rule than any indicator will give you, and it costs nothing.

How do you manage the trade once you are in it?

Structure hands you two things most methods make you invent: a place to move the stop to, and a reason to take some money off. Both are printed on the chart already.

The trailing stop is free. Every time the market makes a new higher low, it has published a level that, if broken, means the reason for your trade has expired. That is exactly the definition of a stop. So the management rule writes itself: when a new higher low completes, move the stop under it. You are not guessing a distance, not using a multiple of anything, not picking a percentage — you are following the same chain that got you in.

What that is worth is easy to show, and it is the difference between two ordinary trades:

15.3 points of required win rate, for one decision about which low to hide behind. The wider stop is not more careful; it is a different trade with the same idea in it, and a much worse one. And note that this gap grows every time the chain adds a link — the original low keeps getting further away while the most recent higher low keeps arriving underneath price.

Take profit in pieces, not in one go. This is the part of the course deck that took us longest to notice, and it changed how we write these lessons. Where the slides show a completed trade — on a bullish engulfing candle, for instance — they do not mark a single target. They mark the entry, the stop, and then take partial profit at the first resistance and the rest at the second. Structure trading fits that naturally, because the chain of previous highs is a ready-made list of resistances in order.

So the management sequence for the worked example:

1. Long at $66,000, stop $62,700 (under the last higher low). Risk $3,300.
2. Price reaches the previous high at $70,000 — the first resistance overhead. Take part of the position off here. This is 1.21R banked on that portion, and it also removes the outcome you least want, which is watching an open profit become a loss.
3. When the next higher low prints above $66,000, move the stop under it. The remainder is now a trade that cannot lose money.
4. Second target at the leg projection, $73,000, for the rest.
5. If instead price closes below the last higher low before either target, exit whatever is left. The chain is broken; you are not in a trend trade any more.

Be clear about what scaling out does and does not do, because it is often sold as free improvement. It lowers your average exit compared with holding everything to the second target, and in a strong trend that is a real cost. What you buy with it is a much narrower spread of outcomes and a stop that reaches break-even sooner. That is a preference, not an edge — but for anyone learning structure, a preference for surviving to read the next chart is the right one.

Why is structure on a higher timeframe more durable?

Because it is further away, and you can put a number on how much further. This is the mechanism behind the course's rule, and the rule is worth more once you can see it.

Take a standard random-walk model of price: over n bars, the typical distance travelled scales with √n. It is a model, not a law, and real markets deviate from it in both directions. But it is enough to answer the question, because the question is comparative.

A swing on the daily chart is built from 96 fifteen-minute bars. A swing on the weekly is built from 672 of them. So, taking a fifteen-minute swing as one unit:

So a weekly higher low sits roughly twenty-six times further below price than a fifteen-minute one. To invalidate it, the market has to move twenty-six times as far, which takes a correspondingly larger flow of orders and a correspondingly longer time. Nothing mystical is happening on the big chart; the level is simply harder to reach. That is the entire content of “higher timeframes are more reliable”, and stating it as distance rather than as authority makes it usable.

It also makes the cost visible, which the usual phrasing hides. The same √t that makes the level durable makes your stop wide. If your risk per trade is fixed — and it should be — then trading weekly structure means a position roughly one twenty-sixth the size of the same risk on a fifteen-minute chart. Higher-timeframe structure is not a free upgrade. It is a trade of position size for durability, and the exchange rate is √t.

The practical reading, then: use the higher frame to decide which box you are in, and a lower frame to find the entry inside it. That is the agreement principle the course builds toward and that lesson 21 on multi-timeframe analysis works through properly.

When is this advice wrong?

Three conditions, and the first is the ordinary one.

When the market is not trending at all. The course is explicit that a sideways market is the market's main state rather than the exception. In a range, the chain flips constantly: a higher low forms, then a lower low, then a higher high that fails. Every one of those labels is correctly marked and none of them means anything, because in a range the turning points are produced by the boundaries rather than by direction. Structure is a trend tool. Applying it to a range does not fail loudly — it fails by generating a steady stream of correct labels that lead nowhere, which is much harder to notice.

When the structure is so old it is no longer live. A higher low from four months ago on the 4-hour chart is a real level and a useless stop; price has moved a long way since, and hiding behind it means a stop so wide that the 15.3-point calculation in section 7 runs in reverse. Structure decays. Use the most recent completed swing, and if the most recent one is far away, that is information — it usually means volatility has expanded and the frame you are on is now the wrong one.

When the “low” you are hiding behind is the obvious one. A clean higher low visible to everyone is also the level where every stop in the market is sitting, and clusters of stops are worth reaching. This is not a reason to abandon the method — there is no version of trading where you get a stop nobody can see — but it is a reason to place the stop with a little room beyond the level rather than exactly on it, and to expect the wick that pokes through and closes back inside. Distinguishing that wick from a genuine break is the subject of lesson 17.

And one thing that is not a limitation but is often described as one: structure is lagging. Of course it is. A swing low cannot be confirmed until price has moved up from it, which means every label you draw is a statement about the past. That is true of every method that uses completed price, and it is the price of using facts instead of predictions. The question is never whether structure is late — it is whether the part of the move that remains after it confirms is worth the risk you have to take to capture it. Section 4 is how you answer that in advance, per trade, in about twenty seconds.

Common mistakes

Frequently asked questions

What do HH, HL, LH and LL mean in trading?

They are four shorthand labels for two comparisons. HH means higher high — the newest peak is above the peak before it. HL means higher low — the newest trough is above the trough before it. LH and LL are the same two comparisons coming out the other way. An uptrend, in this vocabulary, is the pair HH and HL together; a downtrend is LH and LL together. Note carefully that each label compares like with like: a high is only ever compared to the previous high, never to a low. That is what makes the four labels mechanical rather than a matter of opinion, and it is also why you must read them in pairs. A single label on its own tells you about one half of the market's behaviour and says nothing at all about the other half.

How many higher highs do I need before I can call it an uptrend?

There is no threshold that makes it true, but there is arithmetic that tells you how impressed to be. On a deliberately neutral yardstick where each new high is equally likely to be above or below the last, and the same for lows, one higher high plus one higher low happens 25% of the time — a one-in-four coin flip, which is no evidence at all. Two such pairs in a row is 6.25%, three is 1.56%. So each extra swing you wait for divides the chance-alone explanation by four. It also has a price: in our worked case, entering at the first higher low breaks even at 16.3%, at the second 19.1%, at the third 23.1% — roughly three points of break-even win rate per swing waited. That trade-off, not a rule of thumb, is the real answer.

Why is market structure on a higher timeframe more reliable?

Because breaking it costs more money, and you can put a number on how much more. Under a random-walk model the typical distance price travels over n bars scales with √n. Taking a fifteen-minute swing as one unit, an hourly swing is about 2.0× larger, a four-hour swing 4.0×, a daily swing 9.8× and a weekly swing 25.9×. So a weekly higher low sits roughly twenty-six times further below price than a fifteen-minute one, and invalidating it requires a correspondingly larger flow of orders. That is the mechanism behind the course's rule that the higher the timeframe a structure forms on, the more durable the trend. The cost is symmetrical: your stop is twenty-six times wider too, so the position must be twenty-six times smaller for the same risk.

What are the two market structures that HH, HL, LH and LL do not name?

Highs rising while lows fall, and highs falling while lows rise. Because the two comparisons are independent, there are four combinations, not two. HH with HL is an uptrend and LH with LL is a downtrend — those get names. HH with LL is a widening range, the broadening or megaphone shape. LH with HL is a narrowing range, the coil that becomes a symmetrical triangle. Both are skipped by most teaching, and both are dangerous for the same reason: in each of them one half of the structure still reads as bullish. A trader who only checks whether price is making new highs will call a broadening top an uptrend, and be right about the highs and wrong about the trade. The fix is to say the pair out loud every time rather than the single label.

Where do I put the stop when I trade market structure?

Under the most recent higher low, not under the low that started the move — and the difference is large enough to change which trades are worth taking. In our worked example, buying a pullback at $66,000 with a target of $73,000 and a stop 300 under the last higher low at $63,000 risks $3,300 to make $7,000: 2.12R, breaking even at 32.0%. Dropping the same stop under the original low at $60,000 risks $6,300 for the same $7,000: 1.11R, breaking even at 47.4%. That is 15.3 points of required win rate for one decision about stop placement. It is also why structure is worth learning even if you never trade a break: each new higher low hands you a fresh, non-arbitrary place to move the stop to.

Educational content only — not financial advice, and not a trade recommendation. The definitions of higher high, higher low, lower high and lower low, the statement that HH with HL forms a durable bullish structure, the rule that structure on a higher timeframe is more durable, and the observations that resistance tends to break in an uptrend and support in a downtrend all come from the slide course this site learned them from; the charts in this lesson are that course's own. Everything numerical on this page is a worked model built for this lesson and each figure is reproducible from the numbers given. First, the chance figures use a deliberately neutral yardstick in which each new swing high is equally likely to sit above or below the previous one, independently of the lows: 0.25 for one HH+HL pair and 0.25 raised to the power k for k pairs, giving 25%, 6.25% and 1.56%. This is a floor for comparison, not a claim about real price series, which trend and will beat it. Second, the entry ladder uses a structure of $60,000 → $66,000 → $63,000 → $70,000 → $66,000 → $73,000 → $69,000 with a common target of $80,000 and a stop 300 below the preceding higher low, giving 5.15R, 4.24R and 3.33R and break-even win rates of 16.3%, 19.1% and 23.1% via break-even = 1 ÷ (1 + R:R). Third, the stop-placement comparison uses an entry of $66,000 and a target of $73,000, the latter being the previous leg of $7,000 projected forward: risk $3,300 gives 2.12R and 32.0%, risk $6,300 gives 1.11R and 47.4%, a difference of 15.3 points. Fourth, the timeframe table applies √n scaling to 1, 4, 16, 96 and 672 fifteen-minute bars, giving 1.00, 2.00, 4.00, 9.80 and 25.92 — a random-walk model, stated as such. All R figures are before fees, funding and slippage. No hit rate, success rate or historical frequency for structure signals is quoted anywhere on this page, because we have not tested one. The prices visible on the course charts are real market history but are not quoted as figures here, because we could not reach a data source to verify them to the tick. Sources: our own arithmetic, stated inline. Published 2 Sep 2026.