5. Key Levels & HTF Criteria

6 min read

Maintained by Nix Young

5.1 Key Levels

Community insights

If the market hasn't produced a key level, treat the next significant swing, gap, or range as the one you build the read from. — Erik — Lathyrus Q&A, 6 Jul 2026
Look for a key level around the opening price. — Garrett — "GxT Invalidations | Continuations", 03:40
Really only three key levels matter in practice: highs and lows frame reversals and retracements, while fair value gaps and order blocks are what continuations are traded from. — Garrett — "4H Profiling", 4:07
Don't reverse off a failure swing that's just a stop-out of a larger relevant swing — wait for the relevant swing itself, or for a new fractal to form, rather than trading the failure swing as if it were the real level. — Afyz — week trade recap, 12:41
When there's no other key level nearby — for example at all-time highs — only use a 6H consecutive-candle high, a 4H consecutive-candle high, or a 90-minute SSMT to frame a reversal off of. Don't reach for a 1H high in that situation; it isn't a strong enough level on its own. — Erik — trade recap, 12:15
This course deliberately keeps the key-level toolkit to just highs, lows, and gaps — order blocks are excluded on purpose, a stricter and simpler stance than treating them as a valid key-level type. — Anomaly — Advanced course — Lesson 2: Key Level Filtering, 29:14
When there's no refined key level anywhere in the previous range to work with, the fallback is to demand a swing form organically within the current candle instead — treating that self-created "relevant" high/low as the level to reverse from. — Anomaly — Livestreams — Full Chart Process, July 5, 2026, 17:24
A stricter personal test for whether a candle counts as an order block at all: only count it if that specific candle is the one whose body close actually broke the previous swing high/low — a candle that merely sits near the break without being the breaking candle itself doesn't qualify. — AP — "Use Market Structure & Fibonacci Levels for Sniper Entries", 51:33
The canonical construction of a 'Unicorn' setup (traced back to ICT's own paid material and demonstrated live on ICT's channel, by this creator's account): a breaker block — the lowest-closing down-candle, or highest-closing up-candle, immediately before an impulsive stop-hunt leg — that also overlaps a fair value gap left by that same leg. Executed as a limit order at the breaker's near edge, stop on its far side, fixed 2R target, with no partials or breakeven. — Ash Trades — "The Journey To Profitability & The Unicorn Model", 1:16:53
Deliberately restricts daily-bias-building to exactly two liquidity categories — old highs/lows (external) and higher-timeframe fair value gaps (internal) — excluding every other PD-array type from the bias-forming step itself. — Jokerszn — "Daily Orderflow - ICT Daily Bias", 4:02
Proposes a fixed order in which PD arrays appear after a candle opens, keyed to where the open sits relative to the prior range: first a Turtle Soup, then an Order Block, then a Breaker, then a Fair Value Gap — which stage is 'live' depends on the new candle's opening location. (This creator labels that last stage a 'Silver Bullet Zone' — his own idiosyncratic usage, a standard-deviation retracement band, not ICT's canonical AM/PM time-window meaning.) — DexterLab — "Putting All The ICT Pieces Together - Master PO3", 35:56
A concrete mechanical definition of a breaker: the up-close candle sitting immediately before a sweep/reversal down, once price closes back over it, is marked as the breaker and used as the entry zone (mirrored for bullish). — TTrades — "ICT Silver Bullet Strategy - No Daily Bias | With Backtest!", 7:24
The 'Box setup': define an accumulation/consolidation zone using candle bodies only (ignoring wicks), treat any trade over those bodies as manipulation, use the highest swept bodies extended forward as the entry level, stop at the manipulation high, target the 2nd standard deviation. — TTrades — "AMD & Standard Deviations - ICT Concepts", 4:55
A breaker and an order block occupy the same general price area but mark it differently: a breaker is drawn as the full candle range (wicks included) of the qualifying up/down-close candles, while an order block only marks the opening price as a line — the two can sometimes diverge meaningfully as entry levels on the same setup. — TTrades — "Breaker Blocks Simplified - ICT Concepts", 7:10
A fallback key-level rule when a swing has no genuine point of interest to reference: use the 50% midpoint of the nearby opposing candle as the level, and require a close through that midpoint before treating the swing as reclaimed/protected. — TTrades — "How To Trade Consolidation - Phases Of Price", 6:47
A volume imbalance — a gap-like void left when a large candle's range doesn't overlap at all with the following candle — can be treated the same way as a fair value gap for continuation purposes, especially when it lines up with an opposing candle reaching into it. — TTrades — "Trade Continuations Using Order Blocks", 16:29
Choosing a multi-candle series (rather than one small candle within it) as the order block keeps the reference consistent across timeframes — moving up a timeframe collapses that whole series into a single candle, so the series is what the order block actually corresponds to one step up. — TTrades — "Orderblocks Simplified - ICT Concepts", 4:25
A three-way taxonomy of where a continuation-direction order block's reference level can come from: the order block at the smart-money reversal itself, a down-close candle reaching into a fair value gap, or a down-close candle that sweeps a swing low mid-trend. — TTrades — "Order Blocks Simplified - ICT Concepts", 7:03
Two discretionary substitutes for a textbook breaker when the sweep doesn't literally occur: if SMT confirms on a correlated asset even without a literal low sweep, the level can still be treated as a protected breaker; or if price instead reaches a fair value gap rather than sweeping a low, the up-close candle just before that gap can be treated as the functional breaker once price closes back over it. — TTrades — "Trade Continuations Using Breaker Blocks", 1:56
Rather than entering immediately on a retest of a Box-setup level, wait for an actual order block/CISD to form inside that retest zone first — giving two stop-placement choices, tighter (at the CISD candle's low) or wider (at the original manipulation swing). — TTrades — "Change In State Of Delivery [CISD] - Orderblock Formation - ICT Concepts", 13:09
A numeric look-back rule for candidate relevant swings: on any timeframe, only look back the current higher-timeframe candle plus two prior (three total) — explicitly fractal (e.g. 3 months on a daily chart with monthly boxes, 3 days on an hourly chart). — TTrades — "Which Swing Highs and Lows Matter in Trading", 3:01
Marks not just yesterday's high/low but the highs/lows of the prior three days as a standing confluence layer for daily bias, alongside tracking displacement (or failure to displace) over the previous day's high/low as the core bias mechanic. — TTrades — "The Easiest Draw On Liquidity (DOL) - Daily Bias - ICT", 4:36
A precise structural distinction between a mitigation block and a breaker block: a mitigation block only needs low→high→higher-low→higher-high (no opposing extreme needs to be swept — just a failure swing at the highs), while a breaker needs low→high→lower-low→higher-high (a genuinely swept/failed opposing extreme). Personal preference is to avoid trading bare mitigation blocks (often just failure swings that get run through), except when a correlated asset's SMT independently confirms the level. — TTrades — "Mitigation Blocks Simplified - ICT Concepts", 0:14
A four-tier setup-quality rubric combining two axes at once: A+ = internal-range-liquidity-pool reversal with SMT, executed on the weaker/correct correlated pair; A = the same signature executed on the stronger pair instead; B+ = external-range-liquidity-pool reversal with SMT on the weaker pair; B = the same on the stronger pair. — $niper — "This Is The Key To Trading Success", 6:12
Defines the 'B+' grade specifically: when no internal-range-liquidity pool is available (price has been range-bound with no clean HTF PD array), substitute SMT divergence at the previous day's high/low as the sole higher-timeframe reference and still trade it — graded lower than A+ precisely because the PD-array-tag leg of the checklist is missing. — $niper — "How I Made $8,520 Day Trading Today", 0:39
At quarter-end, checks whether price has taken external buy-side liquidity going back to a high from roughly ten months earlier, rather than confining the external-liquidity search to the current quarter or most recent swing — a much longer look-back horizon than a recency-biased read would use. — $niper — "Top Down Analysis Made Easy | Full Trade Breakdown & Review", 0:59
If no imbalance/FVG is left over from the swing that most recently took out a major swing point, expect price to make a single, direct, uninterrupted move to the opposing liquidity with nothing in between to react from — absence of an intermediate PD array predicts a one-shot move rather than a stepped one. — $niper — "How To Find Your Daily Bias Simplified ICT Concepts", 5:23
A second, independent grading axis alongside pair-strength: an execution off an FVG-based array rates higher (A/A+), while an execution off external-range liquidity, a plain order block, or a breaker (no FVG involved) rates lower (B/B+) — which array type supplies the entry, not just which pair, decides the grade. — $niper — "The Forever Model | B Rated Trade Breakdown | $YM", 0:10
States a volume imbalance (rather than an FVG) as the day's explicit stated draw-on-liquidity target, tied to a prior week's low sweep needing 'rebalancing.' — $niper — "ICT Concepts | ES S&P 500 LIVE TRADING w/ COMMENTARY | Top Down Analysis | 09.19.2022", 0:57
Treats the new-week opening gap (the Sunday-night/Monday gap between Friday's close and the new week's open) as its own explicit draw-on-liquidity object, distinct from a standard FVG. — $niper — "Live Day Trading Making $7,680 (Trading Continuation Setup)", 0:58
States an explicit ranked preference order for which PD-array type to trade off when more than one is available: inverse FVG first, order block second, breaker block third. — $niper — "Making 12,046 Day Trading in 55 Minutes", 3:40

Common key levels:

Daily H/L
Weekly H/L
Previous Week EQ
Daily EQ
NWOG / NDOG
HTF gap
4h+ gap
1h+ swing high / low
1h order-pairing range
Range edge
Internal relevant level
Driver-related manipulation level
Valid relevant swings

NWOG / NDOG can act as support / resistance, especially when overlapping with gaps.

If no key level exists yet within the current range, you have to engineer one: wait for price to expand, retrace, and print a fresh swing high/low — that new swing becomes the key level going forward. Filtering which swing counts as relevant isn’t purely mechanical either — the premium/discount EQ test is a helpful heuristic, not an absolute rule, and some of it comes down to judgement built from experience.

Additional rules:

Daily H/L usually forms from a key level, especially a 4h gap or 1h order-pairing range.
Swing formation also requires key-level support.
Without a key level, swing formation is less valid.
A new key level can be used as the third type of expansion timing tool, but it is weaker than SS and HTF open.
Valid key levels are relevant swings.
Exception: a failure swing with SMT can be used as a tradable reference.

5.2 Refined Key Level

Community insights

Gap selection is fractal and nests exactly one timeframe down from whatever candle you are refining: a daily gap for a weekly candle, a 4H gap for a daily candle, a 1H or 30m gap for a 4H candle, and so on. — Garrett — "GxT | Universal Sequence", 12:42
When a lower-timeframe gap is formed from a sloppy, barely-there sweep, prefer using the cleaner higher-timeframe gap instead — for example a 90-minute gap over a messy hourly one — even though the hourly is technically lower timeframe. — Erik — weekly trade recap, 17:00
Don't use the second gap in a same-timeframe sequence as your key level — if price skips past the first gap, you can't know whether it will reverse from the second gap or continue to a third. The first gap is the only one you can trust, unless the asset only has one gap to begin with. — Erik — trade recap, 2:08
Which gap timeframe to trust also depends on where price sits inside the overall reversal-to-draw range: a 15-minute gap while still in discount, stepping up to a 1H or 30-minute gap once price crosses into premium — a 15-minute gap high in the range is really just a lower-timeframe order block, not a protected swing. — Garrett — Q&A stream, 18:10
Whether a breakaway gap stays unfilled also depends on where it forms in the range: one forming in premium, early in a reversal, is much more likely to be left open than one forming in discount. — Garrett — Q&A stream, 16:37
Where the actual sensitive level of an order block sits depends on that candle's own shape: for a big, thick-bodied candle, use 50% of its full range; for a small candle or one with only a small wick, use the zone between its opening price and 50%; for a candle with a large wick, use 50% of the wick specifically rather than the whole candle. — MMXM Trader — The Sequence Ep. 4: Advanced Order Block Model, 1:04
One reason a lower-timeframe fair value gap stays unfilled: check whether a higher-timeframe fair value gap sits between current price and that lower-timeframe gap. The higher-timeframe gap takes priority — a candle closing back into range on a wick (not a body) as it reaches that HTF gap is the sign it's being respected, which leaves the lower-timeframe gap behind as an open breakaway gap. — MMXM Trader — Why Some FVGs Stay Open, 0:52
"Redelivered rebalanced" (RDRB): a fair value gap can stay unfilled when the price action sitting above it (for a bullish gap) has already alternated through both buy-side and sell-side within itself — e.g. a candle opens, trades higher, then wicks lower, then the next candle opens and offers buy-side again. That zone is already internally balanced, so there's no structural reason for price to come back down through it to reach the gap below. — MMXM Trader — Why Some FVGs Stay Open, 3:07
Pair each higher-timeframe key level with a specific lower-timeframe range and change-in-state-of-delivery timeframe, rather than picking one ad hoc: a monthly key level pairs with a daily range and daily CISD, a weekly key level with a 4H range and 4H CISD, a daily key level with a 1H range and 1H CISD. Going lower than this ladder produces too many false stop-hunt signals; going higher risks price running away before entry — this specific pairing is the middle ground that avoids both. — MMXM Trader — Order Blocks: The Complete Guide, 7:26
After a change in state of delivery gives an immediate tap-and-go entry, expect a separate, much deeper retracement back into the full engulfed candle later on — this isn't a sign the level failed, it's smart money engineering stops and shaking out anyone who moved to break-even too early. If the immediate tap wasn't offered at all, that deep retracement back into the candle is usually the entry itself. — MMXM Trader — Order Blocks: The Complete Guide, 4:47
When a chart shows a mess of cluttered, overlapping gaps, step up one timeframe — the clutter typically resolves into a single, clean gap at the higher timeframe. — Anomaly — Livestreams — Full Chart Process, July 5, 2026, 0:45
A lower-timeframe PSP can justify entering before a higher-timeframe gap even fills — e.g. a 15-minute PSP swing point ahead of an hourly gap fill, using the SMT fill plus that LTF PSP as the trigger, since waiting for the HTF gap alone risks missing the move entirely if price never re-sweeps to fill it. — Anomaly — Livestreams — Q&A, July 7, 2026, 1:01:47
"Swing grading" adds a second filter on top of requiring a gap to sit in the upper half (EQ) of the previous key candle: the gap's position must also sit above the 50% mark of the overall implied dealing range. If price has already expanded past that 50% mark without a valid gap forming, wait for a fresh higher-timeframe swing formation rather than forcing a continuation deep in premium (or discount, for shorts). — Lathyrus Masterclass — Supplementary Lessons — Afyz Mini-Lesson: Continuations, 28:01
A concrete post-mortem: a long taken off a 10am SMT/PSP fill without a fresh higher-timeframe swing formation, deep in premium of the range, was a technically "valid" lower-timeframe entry that still violated the swing-grade requirement — one more higher-timeframe candle later produced a proper two-stage crack with a gap in the upper half instead. — Lathyrus Masterclass — Supplementary Lessons — Afyz Mini-Lesson: Continuations, 34:00
The premium/discount gap-timeframe caution (favouring a higher-timeframe gap once in premium) has an exception: if the continuation candle opens immediately adjacent to the original reversal, with little room having existed yet for expansion, even a 15-minute gap is acceptable regardless of premium/discount position — proximity to the reversal overrides the usual timeframe caution. — Lathyrus Masterclass — Supplementary Lessons — Afyz Mini-Lesson: Continuations, 53:55
Before a lower-timeframe candle even prints, its likely gap location can often be inferred purely from how the parent higher-timeframe candle is already closing — a daily candle closing strongly in one direction suggests the next session's gap will sit in the upper (or lower) half of that daily range, letting you pre-identify roughly where to look. — Lathyrus Masterclass — Supplementary Lessons — Afyz Mini-Lesson: Phase of Price & Asset Synchronization, 1:04:18
Reading whether a gap is likely to get run through should combine two factors, not just one: how close the gap sits to the opposing extreme of its range (closer = more "low resistance"), and how convincing the expansion leg that originally created the gap actually was. — Lathyrus Masterclass — Supplementary Lessons — Afyz Mini-Lesson: Asset Synchronization & Phases of Price, 1:04
An unusually large single-candle wick can be treated as its own miniature balanced price range — the open-to-extreme-and-back move implies its own accumulation/manipulation/distribution — with that wick's own equilibrium used as a point of interest or stop reference, distinct from the usual two-overlapping-FVG construction. — Jokerszn — "Daily Orderflow - ICT Daily Bias", 30:17
Whatever PD array a manipulation leg is 'looking for' must sit on the same timeframe as the candle in question or higher — never a lower timeframe — even when zooming into a lower timeframe purely to visually refine/confirm that same level. — DexterLab — "Putting All The ICT Pieces Together - Master PO3", 3:42
'T-spot': a marked zone on the higher-timeframe candle representing where the confirming/opposing wick is expected to form before price continues in the anticipated direction — a forward-projected point of interest tied to the CiSD, used as a continuation-invalidation reference if price closes back through it. — TTrades — "Intro To TTrades Fractal Model [Indicator & Settings]", 11:22
Introduces 'mean threshold' — an order block's midpoint as the specific mechanic behind its reaction — and a strict trust rule: don't treat an unswept low as valid support/invalidation, even if an order block appears to respect it, until that low has actually been swept to create high-resistance liquidity; only then is the new low trustworthy as a stop reference. — TTrades — "High Resistance vs Low Resistance Liquidity - ICT Concepts", 9:33
An alternate construction for a balanced price range: two overlapping opposite-polarity fair value gaps formed at different times combine into a single 'price window' — once price fully rebalances/redelivers through that whole window, expect an aggressive move away from it. — $niper — "How to Find the Best Trade Entry ICT Concepts", 10:03
As a secondary entry when the primary low-risk-buy is missed: drag the wick-high-to-wick-low range of an up-close candle on the 'sell side of the curve' forward across to the 'buy side of the curve,' and wait for a PD array (ideally with its own SMT) to form inside that projected zone as the next entry. — $niper — "Market Maker Models Explained | Step By Step Approach | ICT Concepts", 6:59
'Flat top high' / 'flat body bottom' candles — a single-candle shape formed by repeated wick-taps without a body close through — are treated as inefficiencies that act as price magnets, distinct from the equal-highs-from-two-separate-swings magnet concept. — $niper — "Identifying Liquidity Pools | A Secret To Daily Bias", 8:33
Names a 'balanced price range' on the lower timeframe as its own standalone point-of-interest type, separate from FVG/order-block/breaker/mitigation-block vocabulary — treated as its own execution reference, expecting price to rebalance into it and reject. — $niper — "ICT Concepts | ES S&P 500 TRADE BEAKDOWN w/ COMMENTARY | Top Down Analysis | 09.27.2022", 4:06
Drags an intermediate-term PD array forward in time as a marker of where price will react later, explicitly used as a forward 'sensitivity anchor' or watch-zone rather than an execution level to trade off directly. — $niper — "How I Made $961.90 Live Trading NQ", 1:27

A refined key level:

is within daily EQ;
and is one of:
- a 4h+ gap, or
- a 1h+ swing high / low;
wait for price to hit the level with SMT.
Gap selection is fractal and nests by one timeframe step: to refine a weekly candle’s key level, look for a daily gap inside the previous week’s range; for a daily candle, a 4h gap; for a 4h candle, a 1h or 30m gap; and so on down the timeframes.
Prefer gaps in close proximity to the opening price of the higher-timeframe candle over gaps far from it — a gap far from the open, even if technically inside the range, still forces a large wick and disrespects equilibrium to reach it.
Which gap timeframe to use also depends on where price currently sits inside the overall reversal-to-draw range: use a 15m gap while still in the discount portion of that range, and switch to a 1h or 30m gap once price crosses equilibrium into premium. Trading a 15-minute gap that’s high in the range risks entering on what is really just a lower-timeframe order block, not a genuinely protected swing.
A breakaway gap’s odds of staying unfilled also depend on range position: one forming in premium of the range, early in a reversal, is much more likely to be left open than one forming in discount.

If the key level is not hit with SMT, SMT must form inside the key level before a reversal can be considered.

5.3 When a Level Stops Being a Draw

Community insights

Sweeping one failure swing inside a close-proximity cluster doesn't remove the draw — the market still has a reason to return until every relevant high/low in that cluster has been manipulated. — Garrett — "How I Confirm Narrative With Drivers", 5:47
An SMT built from sweeping a whole cluster of failure swings in an already-trending market is generally a low-probability one — there is usually still low-resistance liquidity left beyond it to draw price back out. — Garrett — "A Deeper Dive into SMT Divergence", 7:48
Don't worry about every failure swing in the market — only ones near the opening price of the current candle, roughly within EQ of the previous candle's range. A failure swing hundreds of points away is irrelevant to the current move. — Garrett — "resuming our back test from last week" stream, 49:13
Don't default to a fixed timeframe to find a relevant swing. Climb up from the intraday chart until you reach the first timeframe where that swing point stands alone, spaced apart from other swings — that's the timeframe whose swing is actually relevant. — Garrett — Q&A stream, 19:49
When scanning the daily chart for relevant swings, cap the look-back at 30 daily candles. Highs/lows further back than that are excluded regardless of how they'd otherwise qualify — the goal is staying fluid for intraday bias, not weighting price action from months or years prior the same as recent structure. — AM — "Relevant Swings", 10:36
A quick mechanical test for whether a swing is relevant: mark premium/discount across the current range. A low printed at EQ or in premium is a defined relevant swing; a low printed in deep discount isn't — treat it as a failure swing and a target instead. — Garrett — Anomaly Course: Key Level, 0:31
"Deferred delivery": at a level of genuinely high-resistance liquidity (stops already run multiple times, confirmed by an order block), price often won't continue straight to the obvious draw. It first deviates to grab the opposing side's stops instead — engineering a fresh relative-equal high/low there — which converts that opposing level from high-resistance to low-resistance, and only then does it continue toward the original draw. Trading a PD array at the high-resistance level itself, expecting immediate continuation, is how this catches people out. — MMXM Trader — Deferred Delivery, 0:33
A deeper manipulation through a low during a bearish continuation is actually a good sign — it shows the liquidity below has been fully cleared, which supports further downside rather than casting doubt on it. — Anomaly — Advanced course — Phases of Price, 17:44
When a retracement falls short of the overall draw and leaves behind close-proximity highs (failure swings), that range is itself a high-probability place to look for the gap that will fuel the next expansion-retracement-expansion leg. — Anomaly — Advanced course — Phases of Price, 17:44
If the only remaining plausible target is disqualified as too far away relative to the current wick size on a higher timeframe (e.g. the weekly), treat the day's objective as already met once an early session prints the extreme — expect the rest of the day to chop inside that established range rather than expand further. — Lathyrus Masterclass — Supplementary Lessons — Summit Exclusive: Final Q&A, 0:17
Once a swing is already protected/confirmed, stop expecting price to revisit that exact level — instead use the very candles that produced the candle-2 closure (the opposing candles) as the new point of interest going forward. — TTrades — "Candle 2 Closure: The Foundation of TTrades Fractal Model", 6:14
Once a gap has been tagged internally and already delivered price to an external liquidity pool, don't reuse that same gap as the basis for the next same-direction signal — wait for a fresh internal-range-liquidity pool to form instead. — $niper — "The Only Liquidity Trading Video You'll Ever Need.", 5:34
Identifies one small PD array as the sole remaining resistance standing between price and a reversal, and treats its violation as near-certain evidence of the reversal precisely because it's the last one left — not because of its size or type. — $niper — "How I Made $8,806 Day Trading Today", 6:29

Relevant swings are spaced-out highs / lows. Failure swings are close-proximity highs / lows within the same small area, i.e. low-resistance liquidity.
Manipulating a relevant swing creates a protected swing.
Manipulating a single failure swing inside a close-proximity cluster does not remove the draw. There is still a reason to return to that area until every relevant high / low within the cluster has been swept.
Once the whole cluster has been manipulated, there is no reason left to return, and this is where price tends to expand away.
To find the relevant swing for a given low/high, do not default to a fixed timeframe (e.g. always the hourly). Instead, climb timeframes upward from the intraday chart until the first one where that swing is spaced apart from other swings, not clustered among close-proximity lows/highs. That is the timeframe whose swing point is actually relevant.
A quick mechanical test for “spaced out enough”: mark premium/discount (EQ) across the current range. A low printed at EQ or in premium is a defined relevant swing; a low printed in deep discount is not — treat it as a failure swing instead, and prefer it as a draw/target rather than a place to expect a reversal.
Give a newly-formed relevant swing at least two higher-timeframe candles to expand away from it. If both fail, treat it as demoted rather than still relevant — it needs to be manipulated again before continuation past it can be trusted.

5.4 Future Gap Filter

Community insights

Before trusting a wick as a protected stop-loss reference, check the timeframe below it for a gap that hasn't finished forming yet. If the lower-timeframe candles bounding it haven't both closed, that "gap" doesn't exist yet — placing a stop there puts it on a level about to become unprotected once the candle completes. Wait for the candle to close and for price to actually tag the resulting gap first. — Garrett — Anomaly Course: Key Level, 4:38
A precise definition of what makes a gap a 'breakaway': a single candle's wick that spans both the buy-side and sell-side of a level in one move, leaving the range it creates 'balanced' and therefore not worth trading back into. — $niper — "The Market Maker Sell Model & Quarterly Theory | Full Trade Breakdown $OIL", 2:39

Before trusting a level as a protected stop-loss reference, check whether it is actually a gap yet, or only about to become one:

A “future gap” is a wick that has not finished forming — the two candles that will bound the gap have not both closed yet, so the gap does not technically exist until the candle in progress closes. Placing an entry or stop against that wick puts it on a level that is about to become an unprotected gap once the candle completes, not a genuinely protected swing.
The mechanical check: within the candle you’re using for entry, look at the next timeframe down (e.g. check the 30m and 15m inside an hourly C2/C3) for any gap that already exists or is about to form. If the wicks of the candles on that lower timeframe are not going to meet once they close, a gap is being created there — wait for that candle to close and for price to actually tag the resulting gap (ideally with a CiSD confirming it) before treating the level as protected.
This applies going up the timeframes too: a 30-minute gap can itself be a future 1-hour gap if the hourly candles bounding it have not both closed. Only once the higher-timeframe candle has closed and the gap is confirmed does it become a valid reference for a stop or entry on that timeframe.
Consequently, C4+ entries are usually lower quality specifically because of this: by the time you are trading a C4, the earlier candles’ wicks have often not resolved into a closed gap yet, so the “key level” backing the entry may still be a future gap rather than an already-protected one.

Personal study notes, shared as-is and in good faith. Educational material only — nothing on this site constitutes financial advice.