Market structure terms

Break of structure

Also called: BOS.

A break of structure is price closing beyond the swing point that defined the current trend, confirming that the sequence of higher highs or lower lows continues.

In an uptrend, structure breaks upward when price closes above the previous swing high — a continuation signal. The term is often confused with its counterpart, the change of character, which is the break that goes the other way and warns the trend may be ending.

The value of the concept is that it makes "the trend is intact" testable. Instead of an opinion about direction, you have a specific price that either has or has not been exceeded, which is precisely what an automated system needs.

The subtlety that costs money is whether you require a close beyond the level or accept a wick. Requiring a close filters out sweeps but arrives late; accepting a wick is faster and catches every liquidity sweep as a false signal. We require the close, and accept the lateness, because on gold the wick version fired constantly and won rarely.

Breakout

A breakout is a move through an established support or resistance level, taken as evidence that the balance between buyers and sellers at that level has resolved.

The premise is that a level holding price back represents resting orders, and once those are consumed there is little immediate resistance to further movement. When it works, the move that follows is fast, which is why breakout strategies tend to have low win rates and large average wins.

They also have a characteristic failure mode. Levels attract orders precisely because they are obvious, which makes them the ideal place to trigger stops before reversing — the false breakout. On liquid instruments a majority of clean breaks fail.

The filters that separate the two are mostly about context rather than the break itself: whether the move is aligned with the higher timeframe, whether the range preceding it was compressed, and what time of day it happened. Our surviving breakout strategy, SESSION-ORB, keys on session structure for exactly this reason — the edge is in when, not just where.

Whether breakouts from the session open follow through on gold is tested in do gold session breakouts follow through?

Change of character

Also called: CHoCH.

A change of character is the first break of structure in the opposite direction to the prevailing trend, taken as the earliest structural warning of a reversal.

In an uptrend made of higher highs and higher lows, the first close below the most recent higher low is the change of character. It does not confirm a downtrend — it confirms that the uptrend's defining sequence has been broken, which is a weaker and more useful claim.

Traded directly it is a poor signal, because most trends produce several of these before actually reversing. Used as a filter it is a good one: refusing to add to a position after a change of character avoids the largest category of trend-following loss, which is holding a trend that ended some time ago.

This is the shape of most genuinely useful structural concepts. They are better at telling you when to stop than at telling you when to start, and our own results agree — the strategies that survived our cull were mostly those whose gating logic prevented bad trades rather than those whose entry logic found good ones. Our STRUCTURE-SHIFT strategy, which traded it directly, lost $60.42 at a 30% win rate and is disabled.

Fair value gap

Also called: FVG, Imbalance.

A fair value gap is a three-candle pattern where the middle candle moves so fast that its neighbours' wicks do not overlap, leaving a price range that traded only briefly.

The gap represents an imbalance: price moved through that range so fast that little two-sided trading occurred. The theory holds that markets tend to return and trade through such ranges properly, making the gap a magnet and a potential entry zone.

Identification is mechanical, which is the concept's main practical virtue — the high of the first candle and the low of the third simply fail to overlap. There is no discretion involved, so it can be coded and evaluated objectively.

Our SMC-FVG implementation did not survive evaluation and is disabled. That is not evidence the concept is worthless; it is evidence that our implementation of it on XAUUSD, at the timeframes we scan, did not produce an edge that persisted out of sample. Publishing that distinction matters more than defending the strategy, and the strategy list is built around it.

False breakout

Also called: Fakeout.

A false breakout is a move through a level that immediately reverses back inside the prior range, trapping traders who entered on the break.

The sequence is consistent: price pushes past an obvious level, triggers the stop orders resting beyond it, fails to attract follow-through, and returns inside the range. Traders who entered on the break are now offside, and their exits fuel the move back.

It is tempting to conclude that fading these is easy money, and equally tempting to build a strategy on it. We did, and it is worth being blunt about the result: our FAILED-BREAKOUT-FADE strategy looked like the strongest choppy-market performer in replay at +$220, and lost money live in both directions. It is disabled.

The lesson generalises beyond the one strategy. Patterns that are obvious in hindsight on a chart are frequently impossible to identify in real time, and a replay engine that knows how the bar closed can identify them trivially. That gap between hindsight and live identification is one of the main reasons backtests overstate performance.

How often gold’s first break of the overnight range at the London open followed through, and how often it failed, is measured in do gold session breakouts follow through?

Higher timeframe bias

Also called: HTF bias, Multi-timeframe bias.

Higher timeframe bias is the directional read taken from a longer chart and used to filter which direction of trade is permitted on a shorter one.

The logic is that a five-minute signal against a strong four-hour direction is fighting the dominant flow, while the same signal aligned with it has the larger move behind it. Rather than predicting, the bias narrows the set of trades you are willing to take.

It has a real and often unacknowledged cost. Requiring alignment cuts trade frequency substantially and guarantees you miss every genuine reversal, since a reversal begins precisely when the shorter timeframe disagrees with the longer one. This is a trade-off, not a free filter.

Our engine computes bias across several timeframes and refuses to trade at all when the result is neutral. Two of our strategies additionally require the bias to be strong rather than merely present, which cuts their frequency further — a deliberate choice, because in testing the marginal trades taken on weak bias were where the losses concentrated.

Liquidity

Liquidity is the presence of resting orders available to be traded against, determining how easily a position can be opened or closed without moving price.

In a liquid market there are orders on both sides at every price and large trades are absorbed without much impact. In a thin market the same order walks the book, producing slippage and erratic candles. Liquidity is not constant — it varies by session, by instrument and by proximity to news.

In smart-money terminology, liquidity also refers to specific pools of resting orders: stops above a swing high, stops below a swing low, and the orders clustered at obvious levels. These are the places where a large participant can execute size without moving price against themselves, which is the mechanical reason price so often visits them.

For gold, the practical implication is timing. XAUUSD is thinnest around the daily rollover and in the late Asian session, where spreads widen and the same strategy produces meaningfully worse fills. Our engine restricts trading to a defined session window rather than running around the clock, which is a liquidity decision rather than a stylistic one — see gold trading hours.

Liquidity sweep

Also called: Stop hunt, Stop run.

A liquidity sweep is a fast move through an obvious high or low that triggers resting stop orders, followed by an immediate reversal back through the level.

Stops cluster where they are easy to reason about: just beyond the recent high, just beyond the recent low, just past the round number. That clustering makes those levels the cheapest place for size to be filled, so price is drawn to them.

The identifying signature is speed plus rejection. A genuine break tends to consolidate above the level; a sweep spikes through, leaves a long wick, and closes back inside. The wick is the visible trace of orders being filled and immediately absorbed.

Two of our four live strategies key on this structure — LIQUIDITY-WICK trades the rejection wick itself, and SMC-OB uses the zone the move originated from. Both are deliberately low-frequency: they wait for a specific structure rather than manufacturing a setup, which is why they trade far less often than an indicator-based system.

Order block

An order block is the candle or zone from which a strong directional move originated, treated as an area where institutional orders were placed and may remain.

The conventional identification is the last opposing candle before an impulsive move — the final down candle before a sharp rally, or the final up candle before a sharp decline. The reasoning is that a move of that size required substantial orders, and that the area where they were placed retains significance if price returns.

It is fair to say the concept is a repackaging of supply and demand zones with more specific rules for drawing them. The rules are the useful part: they make the zone identifiable by code rather than by eye, which is what allows the idea to be tested rather than merely believed.

Tested is exactly what happened. SMC-OB is one of only four strategies enabled in our shipping configuration, and it carried the trend regime almost single-handedly, producing +$179.64 in out-of-sample 2025 testing while everything else in that regime was noise. Several other smart-money strategies in the same family were cut — the full list records which and why.

Range

Also called: Consolidation, Chop.

A range is a period in which price oscillates between a defined high and low without net progress, and it is the condition in which most trend strategies lose money.

Ranges are the market's default state — instruments spend far more time going sideways than trending. That asymmetry is why strategies tested only on trending samples look so much better in a backtest than in a live account that has to survive the other 70% of the time.

Inside a range the edges work and the middle does not. Mean-reversion approaches sell the top and buy the bottom, while trend approaches are repeatedly stopped out by a market that keeps reversing just after confirming. Nothing is wrong with either strategy; they are simply matched or mismatched to the condition.

This is the entire argument for regime detection. Our engine classifies conditions as trending, choppy or mixed and enables a different set of strategies for each. In testing, the choppy classification performed badly enough that it ships disabled by default — the honest conclusion being that we do not currently have a strategy set that earns money in a range, and pretending otherwise would cost users money.

Support and resistance

Support and resistance are price areas where a market has repeatedly stopped and reversed, marking levels where supply and demand previously changed the direction of price.

Support is an area below current price where buying has previously been sufficient to stop a decline; resistance is the same thing above, where selling has stopped an advance. They are areas rather than lines — treating them as exact prices produces stops that are hit by a single tick and entries that never fill.

The reason they persist is memory and positioning. Traders who bought at a level defend it, traders who missed it wait there, and resting orders accumulate around round numbers and previous extremes. The level is not magic; it is where the orders are.

A level that breaks frequently flips role, with old resistance acting as new support. On gold this behaviour is pronounced around whole-dollar and $10 increments because of how orders cluster, and several of our strategies key on the reclaim of a broken level rather than on the break itself — see liquidity sweep.

Whether round-number levels hold on gold is measured in do round numbers work as gold support and resistance?

Swing point

Also called: Swing high, Swing low.

A swing point is a local high or low with a defined number of lower highs or higher lows on each side, used to mark the turning points that define trend structure.

The definition requires a lookback parameter — commonly two or three bars either side. That parameter is not cosmetic: a two-bar swing identifies many minor turns, while a five-bar swing identifies only major ones, and the same chart produces completely different structure depending on which you choose.

Swing points are where stops go, which is why they are also where liquidity pools. Placing your own stop at the obvious swing point puts it exactly where everyone else's is, and a few points beyond it is meaningfully safer for the cost of slightly wider risk.

They also carry a confirmation delay that is easy to overlook in a backtest. A swing high is not confirmed until the required number of bars has closed after it, so a system trading swing points is always acting on information that is several bars old. A replay that treats the swing as known at the moment it formed is measuring a strategy nobody can trade.

When in the day gold tends to set its daily high and low is measured in gold makes its daily low in the Asian session.

Trend

A trend is a sustained directional bias in price, conventionally identified by a sequence of higher highs and higher lows, or lower highs and lower lows.

The structural definition — successive higher highs and higher lows — is more useful than an indicator-based one because it is unambiguous and does not lag. A trend is intact until the sequence breaks, and the moment it breaks has a name: a break of structure.

Every trend definition is timeframe-specific and the timeframes routinely disagree. A market can be trending up on the four-hour chart and down on the five-minute chart simultaneously, and neither reading is wrong. Systems that fail to specify which timeframe defines their bias produce contradictory signals for reasons that look mysterious.

Our engine resolves this with an explicit multi-timeframe bias check and will not open a trade when that bias is neutral. That single gate removed a large share of losing entries in testing — not because trend-following is superior, but because trades taken against a clear higher-timeframe direction consistently produced the worst results across every strategy in the set.