Technical Framework

Why structure comes first

Every ICT concept you'll learn in this module, from liquidity to fair value gaps to order blocks, is read in the context of structure. A fair value gap in an uptrend is a possible place to buy. The exact same gap in a downtrend is a possible place for price to bounce before it falls further. Same pattern, opposite meaning.

Market structure answers two questions:

1. Who is in control right now? Buyers (uptrend), sellers (downtrend), or nobody (range)?
2. What price would prove that wrong? Which swing, if broken, means the current story is over?

If you can answer both on any chart in under a minute, you're ready for the rest of this module.

Swing points: the three-candle definition

In Foundations you met swing highs and lows informally. Here is the precise definition EdgeFlow uses, which is the one most commonly taught in ICT material:

• Swing high: a candle whose high is higher than the high of the candle immediately before it and the candle immediately after it.
• Swing low: a candle whose low is lower than the low of the candle immediately before it and the candle immediately after it.

That's a three-candle pattern: left candle, middle candle, right candle. The middle candle is the swing point. Colour doesn't matter, and bodies don't matter; only the highs (for swing highs) and lows (for swing lows).

Illustrative NQ prices.

Two practical points:

• A swing is only confirmed once the right-hand candle closes. Until candle 3 closes, you don't know whether candle 2 is a swing high, because candle 3 could still trade above it. This is why structure always lags a little.
• Some educators use five candles (two on each side) to filter out noise. That produces fewer, more significant swings. EdgeFlow uses three candles for labelling, then filters by significance using the external/internal idea later in this chapter.

Higher highs, higher lows, lower highs, lower lows

Once you've marked swings, you compare each one with the previous swing of the same type:

• HH (higher high): a swing high above the previous swing high.
• HL (higher low): a swing low above the previous swing low.
• LH (lower high): a swing high below the previous swing high.
• LL (lower low): a swing low below the previous swing low.

Not every three-candle swing deserves a label. A tiny two-tick wobble inside a strong leg technically counts, but labelling it adds noise. For labelling trend, use the swings that bound each leg: the high where an up-leg ended and the low where a down-leg ended. The structure lab does this for you so you can see the difference.

Trend definitions

The swing you watch has a name in many SMC courses: the protected low in an uptrend (the HL that produced the latest HH) and the protected high in a downtrend. While that level holds, the trend is intact, and a pullback is just a pullback.

BOS, CHoCH and MSS

These three labels describe what happens when price trades beyond a swing point.

Break of Structure (BOS) is a break in the direction of the existing trend. In an uptrend, price closing above the previous swing high (making a new HH) is a bullish BOS. In a downtrend, closing below the previous swing low is a bearish BOS. A BOS says: the trend continues.

Change of Character (CHoCH) is the first break against the existing trend. In an uptrend, the first close below the protected HL is a bearish CHoCH. In a downtrend, the first close above the protected LH is a bullish CHoCH. A CHoCH says: the trend may be ending. It's a warning, not a confirmed reversal. Price can CHoCH, then resume the original trend.

Market Structure Shift (MSS) is ICT's own term. It's often used interchangeably with CHoCH, but in ICT teaching it usually carries two extra conditions: price took liquidity first (swept a high or low), and the break happened with displacement (a strong, fast move, covered in Chapter t3). EdgeFlow uses the terms like this:

So every MSS is a CHoCH, but not every CHoCH is an MSS. The MSS is the higher-quality version.

Body close vs wick: when is a level "broken"?

This is one of the most argued points in SMC. Imagine the swing high is 20,080 and a candle trades to 20,086 but closes at 20,074. Is structure broken?

• Wick-break camp: yes, price traded above the level, so it's broken.
• Body-close camp: no, the candle closed back below. That's a probe, possibly a liquidity sweep (Chapter t2), not a break.

EdgeFlow rule: a break of structure is confirmed only by a candle-body close beyond the swing point on the timeframe you're using to read structure. A wick through the level with a close back inside is treated as a sweep or raid, not a BOS.

Why we choose this:

1. It separates sweeps from breaks. Most of the ICT framework depends on telling "price grabbed liquidity and rejected" apart from "price broke and accepted". The close is the cleanest objective test.
2. It's testable. You can backtest "close beyond" without arguing about how many ticks count.
3. It reduces false signals. Wick-only breaks are often exactly the trap that the ICT approach teaches you to fade.

The cost is that body-close confirmation is slower. On a 15-minute chart you wait for the candle to finish. That's a trade-off we accept.

Internal vs external structure

Look at any chart and you'll find swings inside swings. SMC splits them into two layers:

• External structure (swing structure): the major swing high and swing low that define the current dealing range on your chosen timeframe. These are the levels that, when broken, change the trend on that timeframe.
• Internal structure: the smaller swings that form inside a leg or inside a pullback between the external points.

Example: On the 1-hour chart NQ made a swing low at 20,000 and a swing high at 20,200. Those two points are external structure. During the pullback from 20,200, price made a series of small LH and LL down to 20,080. That's internal bearish structure inside an external uptrend.

Why it matters:

• An internal CHoCH is not an external CHoCH. If price breaks one of those small internal lower highs during the pullback, it's an internal bullish CHoCH: often the signal that the pullback is finished and the external uptrend is resuming. It's a possible entry trigger, not a trend change.
• External breaks change the bias on that timeframe. Only a body close below the external swing low of 20,000 would turn 1-hour structure bearish.

Timeframe changes structure

The same price action can be bullish on one timeframe and bearish on another, and both readings are correct.

All three are true at the same time. The daily trend is up. The 1-hour is in a pullback. The 5-minute may be signalling that the pullback is ending. This is the core of the multi-timeframe process in Chapter t7: the higher timeframe gives direction, the lower timeframe gives timing.

Rules of thumb:

• One candle on a higher timeframe contains a whole structure on a lower one. A single daily candle can contain an entire 5-minute uptrend and downtrend.
• Higher-timeframe swings carry more weight. A daily swing high has more resting orders and more significance than a 1-minute swing high.
• Always say which timeframe you mean. "NQ is bearish" is meaningless. "NQ is bearish on the 1-hour inside a bullish daily" is a statement you can act on.

A five-minute structure routine

Use this every session before you look for setups:

1. Pick your structure timeframe (for intraday NQ, usually 1-hour or 15-minute).
2. Mark the last 6–10 significant swing highs and lows using the three-candle rule.
3. Label each HH, HL, LH, LL.
4. Identify the external range and the protected swing.
5. Note the last BOS or CHoCH and whether it was a body close.
6. Write one sentence: "1H bullish, protected low 20,040, last BOS at 20,080."

That sentence becomes the first line of your trade plan in Chapter t8.

What "liquidity" means here

In Foundations, liquidity meant how easily you can trade size. ICT uses the word in a narrower, more practical sense: liquidity is clusters of resting orders sitting at predictable prices, mostly stop-loss orders and pending entry orders.

Why does it matter? A large participant who wants to buy a lot needs a lot of sellers to trade against. A big pool of sell stops (stop-losses on long positions, plus sell-stop breakout entries) is exactly that: a block of sell orders that will fire at a known price. If price trades into that pool, the large buyer can fill size without pushing price up against themselves.

This is why ICT says price is drawn toward liquidity. It isn't because anyone is targeting your stop. It's because thousands of stops at the same obvious level form a useful pool of orders.

Buy-side and sell-side liquidity

The naming confuses people at first. "Buy-side" describes the type of orders resting (buy orders), not the direction you should trade. A run on buy-side liquidity is a move up into buy stops, and it often provides the fuel for a move down afterwards, because big sellers have now filled.

Where liquidity pools form

Some highs and lows attract far more orders than others. These are the levels EdgeFlow marks every day.

Equal highs and equal lows. Two or more swing highs at roughly the same price (within a few ticks) look like "resistance" to most traders, who put their stops just above. Equal highs are sometimes called relatively equal highs to make clear they don't need to match to the tick. Two touches create a pool; three create a bigger one. ICT calls these "engineered liquidity" when they appear to form deliberately before a run.

Previous day high and low (PDH / PDL). Yesterday's high and low are watched by almost every day trader. For futures, EdgeFlow defines the trading day as the CME session (18:00 ET to 17:00 ET), so PDH/PDL are the high and low of that full session, not just cash hours. Some traders also mark the regular-hours (RTH) high/low separately for NQ and ES; if you do, label them clearly.

Previous week high and low (PWH / PWL). The same idea on a weekly scale. Bigger pool, bigger reaction, slower to reach.

Session highs and lows. - Asia range: the high and low formed during the Asian session (roughly 20:00–00:00 ET, which is 04:00–08:00 GST in US summer / 05:00–09:00 GST in US winter). Very often London takes one side of it. - London range: the high and low formed in London (roughly 02:00–05:00 ET for the killzone, 10:00–13:00 / 11:00–14:00 GST). Very often New York takes one side of it.

Trendline liquidity. When price respects a rising trendline three or more times, traders buy each touch with stops just below the line. That stacks sell-side liquidity along the whole line. A clean trendline in ICT eyes isn't support; it's a queue of stops waiting to be run.

Round numbers and old highs/lows. 20,000 on NQ, 3,000 on gold, 100,000 on BTC: these psychological levels collect stops and limit orders. Old untouched highs and lows from weeks ago can also still hold orders.

Inducement

Inducement is a small, obvious pool of liquidity that forms before price reaches the level it's really heading for. It "induces" traders to enter early.

Typical example: price is in an uptrend and pulling back toward a bullish fair value gap (Chapter t3). Before reaching it, it forms a small internal swing low. Eager traders go long at that minor low, with stops just below it. Price then drops through that minor low (taking their stops), tags the real zone, and rallies. The early buyers were the liquidity.

How to use the idea:

• Expect the first obvious minor low inside a pullback to be taken before the real reaction comes.
• Prefer entries after inducement has been cleared. If there's a clean minor swing right in front of your zone, assume it's bait.
• Don't overuse it. "That was just inducement" can explain away any losing trade. Like every concept here, write a rule before you test it.

Sweep vs genuine break

A liquidity sweep (also called a raid, stop hunt or run) is when price trades through a high or low, triggers the resting orders, and then fails to hold beyond the level. A genuine break is when price trades through and accepts beyond it, continuing in the same direction.

On the chart, both start out looking identical: price pokes through an obvious level. The difference shows only afterwards. EdgeFlow uses two tests.

Test 1: the close. On your execution timeframe (usually 5-minute or 15-minute): - Sweep: price trades beyond the level, but the candle closes back inside (above the low it swept, or below the high it swept). - Break: the candle closes beyond the level, and subsequent candles hold there.

Test 2: displacement afterwards. - Sweep: shortly after taking the level, price moves strongly the other way, ideally breaking a nearby swing (an MSS) and leaving a fair value gap. - Break: after taking the level, price continues, retests the broken level from the other side and holds, and the next displacement is in the breakout direction.

Neither test is perfect. A sweep can close back inside and still fail. Treat them as evidence that shifts probabilities, then let your setup rules (Chapter t8) decide.

Draw on liquidity (DOL)

The draw on liquidity is the level price is most likely heading toward next. It's the target of the current move and the basis of daily bias (Chapter t7).

How to pick the DOL:

1. List the nearest untaken pools above and below (PDH/PDL, session highs/lows, equal highs/lows, PWH/PWL, and unfilled fair value gaps from Chapter t3, which also act as draws).
2. Check higher-timeframe structure. In a bullish daily structure, the obvious draw is buy-side above. In a bearish one, sell-side below.
3. Check what's already been taken. If sell-side below (say, PDL) was swept this morning and price displaced up, the next draw is more likely buy-side above.
4. Pick one primary DOL and one alternative. Write them down before the session.

Putting liquidity on your chart

Daily routine, ten minutes before your session:

1. Mark PWH/PWL and PDH/PDL.
2. Mark the Asia high/low (and London high/low before New York).
3. Circle any equal highs or lows on the 15-minute and 1-hour chart.
4. Note which pools have already been taken today.
5. Write the primary DOL and alternative.

Keep it clean. Six to eight levels is plenty. If your chart has 25 lines on it, every price is "near liquidity" and the concept stops helping.

Displacement

Displacement is a strong, fast, one-directional move: one or several large-bodied candles moving in the same direction, with small wicks against the move, travelling far relative to recent candles.

It's the chart's way of showing that one side has overwhelmed the book. In Foundations you saw how a large market order eats through several price levels. Displacement is that process at scale: aggressive buyers (or sellers) consuming every offer (or bid) in their path faster than new orders can rest.

How to recognise it:

A practical EdgeFlow filter: on your execution timeframe, a displacement candle typically has a body at least about 1.5–2 times the average body of the previous 10–20 candles. This isn't an official ICT number; it's a consistent rule of thumb for backtesting. Pick a multiple, write it down and test it.

The fair value gap: exact definition

A fair value gap (FVG) is a three-candle pattern where the move was so one-sided that the wicks of the first and third candles do not overlap. The space between them is a price range that traded only in one direction, during the middle candle.

• Bullish FVG: the low of candle 3 is above the high of candle 1. The gap runs from candle 1's high (bottom of the gap) to candle 3's low (top of the gap).
• Bearish FVG: the high of candle 3 is below the low of candle 1. The gap runs from candle 3's high (bottom) to candle 1's low (top).

Candle 2 is the displacement candle. Its body passes straight through the gap.

Notice the definition uses wicks (highs and lows), not bodies. If candle 3's low touches or dips below candle 1's high, even by a tick, there's no FVG.

BISI and SIBI

ICT gives the two directions their own names, describing the order flow imbalance:

• BISI — Buy-side Imbalance, Sell-side Inefficiency. A bullish FVG. Buyers were in control; the range was offered only to buyers, and sellers didn't get a fair chance to trade there.
• SIBI — Sell-side Imbalance, Buy-side Inefficiency. A bearish FVG. Sellers were in control; buyers didn't get to trade in the gap.

Most traders simply say "bullish FVG" and "bearish FVG". Know the terms so you can follow ICT material, but use whichever you prefer in your journal.

Why price returns to gaps

The ICT explanation: an FVG is a range where price was delivered inefficiently, only one side traded. Markets tend to return to rebalance those areas, and large participants may use the return to add to positions at better prices.

A more neutral explanation: after a fast move, early buyers take profit and late buyers wait for a pullback. The pullback often stops where the move began accelerating, which is often the gap.

Either way, the practical observation is the same: in a trending market, pullbacks often reach into a recent FVG and react there. That makes the gap a candidate entry zone. It doesn't make it a guaranteed one.

Consequent encroachment, partial and full fills

Consequent encroachment (CE) is the 50% midpoint of the gap:

CE = (gap top + gap bottom) ÷ 2

ICT treats CE as a key reference inside the gap. Commonly taught behaviour:

• In a bullish FVG, a strong trend often respects CE: candles may wick into the lower half but close above CE. A body close below CE weakens the gap.
• In a bearish FVG, mirror image: price may wick into it but closes below CE.

Fill status:

Inversion FVG (IFVG)

When price closes through an FVG, the gap has failed. ICT teaching says that failed gap can flip roles:

• A bullish FVG that price closes below can later act as resistance (a bearish inversion FVG).
• A bearish FVG that price closes above can later act as support (a bullish inversion FVG).

The IFVG is especially useful as a sign of a shift. When the most recent gap in one direction fails and flips, it's evidence order flow has changed.

Volume imbalance and liquidity void

Two related ideas you'll see in ICT material:

• Volume imbalance: the gap between the bodies of two consecutive candles where the wicks still overlap. Example: candle A closes at 20,040, candle B opens at 20,043; the bodies don't touch, but wicks fill the space. It's a smaller imbalance than an FVG, often seen at session opens on NQ and ES, and price often returns to "close" it.
• Liquidity void: a large, fast price range covered by one or a few long candles, with very little two-sided trading. Think of a 100-point NQ drop in two minutes after CPI. Several FVGs usually sit inside it. Price often returns into voids over time, though not necessarily soon.

Which FVGs matter

On a 1-minute NQ chart you can find dozens of FVGs per hour. Most mean nothing. EdgeFlow ranks an FVG by these filters:

1. Created by displacement. The middle candle must be a genuine displacement candle, not a slightly larger candle in a chop.
2. Created after a liquidity sweep. The best gaps form in the move away from a sweep: price takes the Asia low, then displaces up, leaving a bullish FVG. That gap is the footprint of the reversal.
3. Breaks structure. The displacement that created the gap also produced a BOS or MSS (Chapter t1).
4. Aligned with bias. Bullish gaps in bullish higher-timeframe conditions; bearish gaps in bearish ones.
5. In the right zone. Bullish gaps in discount, bearish gaps in premium (Chapter t5).
6. In a killzone. Gaps formed during the London or NY AM killzones (Chapter t6) are given more weight than ones formed during lunch or late Asia.

A gap that meets 1–4 is a qualifying FVG in EdgeFlow's model. One that also meets 5 and 6 is high quality.

The idea behind order blocks

Large participants can't fill their full size in one go. ICT's view is that a big buyer builds a position partly during a down move, buying from sellers, and then the buying pressure tips over and price displaces up. The last down candle before that displacement marks the area where the final buying happened before price "left".

When price later returns to that area, the theory goes, the same participants may defend it, or add to their position, so it often acts as support. Whether or not you accept the "institutional footprint" story, the practical observation is useful: the origin of a strong move is a logical place for price to react on its first return.

Order blocks: exact definition

• Bullish order block (bullish OB): the last down-close candle (close below open) before a displacement move up that breaks structure (a BOS or MSS to the upside).
• Bearish order block (bearish OB): the last up-close candle (close above open) before a displacement move down that breaks structure to the downside.

If there are several consecutive down candles right before the displacement, ICT material commonly treats the series as the order block, with the zone taken from the whole cluster. EdgeFlow's default is simpler: use the last single opposing candle unless the cluster is tight (all bodies within about the same range), in which case you may merge them.

The OB zone. Educators differ:

Mean threshold (MT) is the 50% level of the order block. In ICT teaching it's commonly measured on the body (halfway between open and close). EdgeFlow uses the body midpoint as the MT. ICT teaches that in a strong move price should not close beyond the mean threshold on its return; a body close past the MT weakens the block.

Validity rules: not every down candle is an order block

Every move up starts after some down candle. If you called them all order blocks, the concept would be useless. EdgeFlow requires:

1. Displacement away from it. The move after the candle must be a genuine displacement (Chapter t3), not a slow drift.
2. A break of structure. The displacement must produce a BOS or MSS on the timeframe you're reading. No structure break, no order block.
3. Ideally, a fair value gap. The best OBs have an FVG in the displacement immediately after them. The FVG proves the move was one-sided.
4. Ideally, a liquidity sweep. The strongest bullish OBs form after price took sell-side liquidity (the OB candle itself, or the one before it, often swept a low). The strongest bearish OBs form after taking buy-side liquidity.
5. Unmitigated. It must be the first return. An OB that has already been revisited and reacted is used up (see mitigation below).
6. Aligned with bias and zone. Bullish OBs in discount within bullish conditions; bearish OBs in premium within bearish conditions (Chapter t5).

Rules 1 and 2 are required. Rules 3–6 raise the quality.

Mitigation

Mitigation is ICT's word for price returning to an order block (or other PD array) and trading into it. The term comes from the idea that participants who were caught on the wrong side, or who didn't finish filling, use the return to "mitigate" (reduce) losses or complete their orders.

Practical rules:

• First touch matters most. The first return into an unmitigated OB is the highest-probability reaction. Each later return is weaker.
• Once mitigated, it's spent. After a clean reaction, EdgeFlow no longer treats that OB as an entry zone. (It can still be a reference for stops.)
• Depth. A strong trend often reacts at the top of the zone. A deep return to the mean threshold is normal. A body close beyond the far edge means the block has failed.

Breaker blocks

A breaker block is an order block that failed, and now acts with the opposite role. The classic ICT sequence for a bullish breaker:

1. Price makes a swing high, then a lower low that sweeps sell-side liquidity (taking a prior low).
2. The last up-close candle before that drop is a potential bearish OB.
3. Price then displaces up through that bearish OB, closing above it and breaking the swing high (MSS up).
4. The failed bearish OB is now a bullish breaker. When price returns to it from above, it's expected to act as support.

The bearish breaker is the mirror: a swing low, a higher high that sweeps buy-side liquidity, a failed bullish OB that price displaces down through, and that zone then acting as resistance.

Why it can work: traders who bought at the failed bullish OB are trapped long. When price rallies back to their entry, many exit at breakeven, adding selling pressure exactly at the breaker zone.

Rejection blocks (briefly)

A rejection block forms at a swing high or low where price left long wicks: repeated tests that were rejected. ICT uses the area between the bodies and the wick extremes as the zone.

• Bearish rejection block: at a swing high with long upper wicks. The zone is from the highest body (open or close) up to the wick high.
• Bullish rejection block: at a swing low with long lower wicks. The zone is from the lowest body down to the wick low.

The idea is that price often returns to the body level once more and fails again, or runs through the wicks to take the liquidity beyond them. Use it as a stop-placement reference more than a stand-alone entry. Further block types, such as propulsion blocks and vacuum blocks, are covered in Execution.

PD arrays: the toolkit in one table

ICT groups all these reference zones under the name PD arrays (premium/discount arrays): price levels and zones that algorithms and traders reference for entries and targets. You've now met the main ones.

A mitigation block is closely related to the breaker: the block fails, but price didn't first make a new extreme (no sweep). Many ICT traders treat it as a weaker version of the breaker.

Choosing between an OB and an FVG

Often the order block and the FVG sit right next to each other: the down candle, then the displacement with a gap above it. Which do you use?

• FVG first. EdgeFlow's default entry is the qualifying FVG's CE (Chapter t8), because the gap is closer to the current price and fills more often.
• OB as the deeper level. If price fills the FVG completely, the OB below it is the next zone, and the stop goes below the OB's low or the sweep low, whichever is further.
• Overlap is the best case. When the FVG, the OB and the OTE zone (Chapter t5) overlap, you have strong confluence in one tight area.

The dealing range

A dealing range is the range between a significant swing low and swing high that price is currently trading within. In practice, it's the most recent external leg on your chosen timeframe (Chapter t1): the low and high that bound the move you're trading a pullback of.

• Bullish dealing range: price displaced from a swing low up to a swing high, and is now pulling back. You anchor from the low (start of the leg) to the high (end of the leg).
• Bearish dealing range: price displaced from a swing high down to a swing low, and is pulling back up. You anchor from the high to the low.

The leg should be a clean impulse: ideally the one that created a BOS or MSS and left a fair value gap. A choppy, overlapping leg makes a poor dealing range.

Equilibrium, premium and discount

Equilibrium (EQ) is the 50% level of the dealing range: EQ = (high + low) ÷ 2.

• Premium: the part of the range above EQ. Price is "expensive" relative to the range.
• Discount: the part of the range below EQ. Price is "cheap" relative to the range.

The ICT rule is simple and very useful:

Think of it like buying stock from a wholesaler. If you believe the price is heading higher, you want to buy in the lower half of the range, not at the top after it's already run.

Optimal Trade Entry (OTE)

The Optimal Trade Entry (OTE) is ICT's name for a specific retracement band inside the discount (for longs) or premium (for shorts):

• 62% retracement: the start of the OTE band.
• 70.5% retracement: the "sweet spot", the midpoint of the band.
• 79% retracement: the deep end of the band.

ICT uses 62% and 79% where classic Fibonacci tools use 61.8% and 78.6%. The difference is tiny; EdgeFlow uses ICT's round versions.

The logic: a healthy pullback in a strong trend often retraces past the 50% mark (into discount) but not all the way. The 62–79% band is deep enough to give a good price and a tight stop, but shallow enough that the original impulse is still intact.

Retracement level formulas (r = retracement ratio, e.g. 0.62):

• Bullish range: level = high − r × (high − low)
• Bearish range: level = low + r × (high − low)

Extension target formulas (ext = extension ratio, e.g. 0.27):

• Bullish range: target = high + ext × (high − low)
• Bearish range: target = low − ext × (high − low)

Fibonacci tools display extensions as negative levels (−0.27, −0.62, −1.0) because they sit beyond 0% on the far side of the range. The numbers in the formulas above are the size of that extension.

Anchoring the fib: wick or body?

Another area of disagreement:

• Wick to wick: anchor at the true low (lowest wick) and true high (highest wick) of the leg.
• Body to body: anchor at the lowest and highest candle bodies (opens/closes), ignoring wicks.

EdgeFlow convention: wick to wick. The wick extremes are where price actually traded, and they're usually where the liquidity sweep happened, which is where your stop goes. Anchoring to the same points as your stop keeps the risk maths consistent.

For a bullish range, put the 1.0 (100%) at the swing low and 0 at the swing high. For a bearish range, put 1.0 at the swing high and 0 at the swing low. On most charting platforms that means dragging from the start of the leg to the end of the leg. Check your tool's settings show 0.5, 0.62, 0.705, 0.79, −0.27 and −0.62.

Worked NQ example: every level

Bearish example: gold

Anchoring mistakes and what they cost

Small anchoring choices move every level. Using the same NQ leg:

Rules that prevent this:

• Anchor the leg that broke structure. It's the impulse that produced the BOS or MSS, usually starting at a liquidity sweep.
• Re-anchor when a new extreme prints. If price makes a new high at 20,230 before pulling back, the dealing range is now 20,000 → 20,230 and every level moves.
• Don't anchor mid-leg. If the impulse hasn't finished (no pullback yet, no swing high confirmed by the three-candle rule), you don't have a range to measure.

Premium and discount on more than one timeframe

Premium and discount are always relative to a specific range, so they nest the same way structure does (Chapter t1). A 5-minute pullback can be in deep discount of its own small range while the whole session sits in premium of the daily range.

EdgeFlow checks two layers:

The best longs are in discount on both: for example, the daily range is 19,800 – 20,400 (EQ 20,100), price is at 20,050, and your 5-minute entry is at the 70.5% of the morning's displacement leg. A long that's in LTF discount but in deep HTF premium, right under a daily bearish FVG, gets a lower grade or a smaller size.

Using OTE with the other PD arrays

On its own, a fib level is just a number. It becomes useful when it overlaps with a PD array:

• A bullish FVG whose CE sits inside the OTE band.
• A bullish OB whose zone sits at or near 70.5%.
• Sell-side liquidity just below the 79% level that might be swept before the reaction.

EdgeFlow treats OTE as a confluence filter, not a trigger: "Is my FVG or OB entry in discount, ideally in the OTE band?" If yes, quality goes up. If the only PD array is in premium, the long is downgraded or skipped.

Targets and extensions

Where to take profit? EdgeFlow's priority order:

1. Opposing liquidity (Chapter t2): the nearest meaningful pool on the other side, such as the swing high of the range, equal highs, PDH or the session high.
2. The swing high itself (the 0 level): a natural first partial.
3. Extensions: −0.27 and −0.62 are the most commonly used ICT targets beyond the range. Some traders also use −1.0 (a full measured move).
4. Standard deviation projections: ICT also projects targets using multiples of the manipulation leg, called standard deviations. These are covered in Execution.

Use liquidity first and extensions second. If an extension lines up with a pool of buy-side liquidity, it's a strong target.

Why time matters

ICT's view is that price is delivered in a time-based rhythm. Liquidity is thinnest overnight, builds through Asia, is taken during London, and is taken again (often the other side) during New York. The largest participants trade when liquidity is deepest, which concentrates meaningful moves into specific windows.

You don't need to accept any theory about algorithms to use this. The observable facts are enough: volume and volatility in NQ, ES, gold and oil are heavily concentrated around the London open, the 08:30 ET data releases and the 09:30 ET cash open. A fair value gap that forms at 09:45 ET is formed by far more participants than one that forms at 12:30 ET.

EdgeFlow uses time in three ways:

1. Reference prices: the midnight open and session opens.
2. Windows to trade: killzones and Silver Bullet hours, where we look for setups.
3. Windows to avoid: lunch, late afternoon and the minutes around high-impact news.

A Dubai trader's clock

All ICT times are anchored to New York time. Dubai doesn't change its clocks, so every ICT window moves by one hour in Dubai when the US changes:

• US summer time (EDT): GST = ET + 8 hours. In 2026 this runs until Sunday 1 November 2026.
• US winter time (EST): GST = ET + 9 hours. From 1 November 2026 until Sunday 14 March 2027.

The New York midnight open

The New York midnight open is the price at 00:00 ET (08:00 GST summer / 09:00 GST winter). ICT uses it as the opening price of the "true day" for intraday analysis, rather than the 18:00 ET Globex open.

Why midnight? It sits between the quiet Asian session and the start of London. In ICT's framing, the hours before midnight are about building a range, and the true daily move starts afterwards.

How EdgeFlow uses it:

• Draw a horizontal line at the midnight open price each day.
• On a bullish day, the best long entries are usually below the midnight open: buying in the day's discount after a manipulation lower.
• On a bearish day, the best short entries are usually above the midnight open.
• As a bias check: if your daily bias is bullish and price spends the NY AM session well below the midnight open with no sign of a reversal, your bias may be wrong.

Some ICT traders also mark the 08:30 ET open and the 09:30 ET open for the same reason on a shorter scale.

Power of 3: accumulation, manipulation, distribution

The Power of 3 (PO3), also called AMD, is ICT's template for how a trading period often unfolds around its open:

1. Accumulation: price consolidates near the open. Positions build; a range forms.
2. Manipulation: price makes a false move away from the true direction, beyond the open, taking the liquidity on that side (stops and breakout entries).
3. Distribution: the real move. Price reverses through the open and expands in the true direction, often closing near the opposite end of the period's range.

On a bullish day the daily candle tends to look like: open → move down first (manipulation, forming the low and the lower wick) → rally through the open → close near the high. That's why ICT says the low of a bullish day often forms early, in London or the NY open.

On a bearish day: open → move up first (forming the high) → reversal through the open → close near the low.

PO3 works at several scales: weekly (Monday–Tuesday often forms the week's manipulation), daily (midnight open), and session (e.g. the first 30 minutes after 09:30 ET). It's descriptive, not predictive: you only know it was manipulation once distribution has started.

The Judas swing

The Judas swing is the manipulation leg viewed on its own: an early false move, usually after midnight or at the London open, in the direction opposite to the day's real move. It typically:

• moves away from the midnight open,
• runs an obvious liquidity pool (Asia high/low, PDH/PDL),
• then reverses with displacement back through the midnight open.

The trap: breakout traders see the early move and join it. Their stops then fuel the real move. You can't confirm a Judas swing in real time; you can only see that price ran liquidity, rejected, and displaced back through the open. That's why EdgeFlow's model waits for the MSS after the sweep rather than guessing the low.

Killzones: recap

You met the killzones in Foundations. These are the commonly cited ICT windows (boundaries vary slightly between sources). All are anchored to New York time.

Some educators use 08:30–11:00 ET for the New York AM killzone. The NY lunch (roughly 12:00–13:00 ET, 20:00–21:00 GST summer / 21:00–22:00 GST winter) is widely treated as a low-quality period to avoid.

The Silver Bullet windows

The Silver Bullet is an ICT time-based model built around three one-hour windows. The commonly cited windows:

The commonly taught process inside a window:

1. Before the window: mark the draw on liquidity (session highs/lows, PDH/PDL, equal highs/lows).
2. In the window: wait for price to sweep a nearby pool or shift structure, establishing direction.
3. Entry: on a fair value gap created inside the window by displacement, in the direction of the draw.
4. Stop and target: stop beyond the manipulation swing; target the opposing liquidity.
5. If no valid FVG forms before the window closes, there's no Silver Bullet trade that session.

EdgeFlow's setup model (Chapter t8) is compatible with the Silver Bullet: a valid Silver Bullet is simply our model executed inside a one-hour window.

Macro times (briefly)

ICT also teaches short "macro" windows, roughly 20 minutes long, where it expects the price-delivery algorithm to seek liquidity or rebalance a gap. Exact minutes vary between sources. The most-quoted pair is:

Fuller lists include windows around 08:50–09:10, 11:50–12:10, 13:10–13:40 and 15:15–15:45 ET, plus London windows. EdgeFlow doesn't require macros. Treat them as a note in your journal: "Did my entry happen in a macro?" Over 100+ trades, you'll learn whether it matters for you.

News days

High-impact releases change everything about time-based trading:

EdgeFlow news rules:

• No new entries from 5 minutes before to 5 minutes after a red-folder (high-impact) release. Spreads widen and stops slip.
• On 08:30 data days, the release often acts as the manipulation itself: a fast spike into one side's liquidity, then the real move. Wait for the MSS after the spike.
• On FOMC days, many ICT traders avoid the morning entirely (price often ranges) and either trade after 14:30 ET or sit out.
• Check the calendar before every session. It's in your pre-session checklist in Chapter t8.

The top-down process

Every timeframe tells the truth about its own scale. The skill is giving each one the right job:

You always work top to bottom. The LTF is allowed to give you an entry only in the direction the HTF has already approved.

Timeframe pairings

Each layer is usually 3–6 times the one below it. Use one of these standard pairings and stick to it so your journal compares like with like.

Most EdgeFlow members trade the intraday pairing on NQ and gold: daily and 4H for bias, 1H and 15m for structure and zones, 5m (sometimes 1m) for the entry.

Defining daily bias

Daily bias is your answer to: "Today, is price more likely to reach the buy-side above or the sell-side below first?" It's not a prediction of the close; it's a call on the draw.

EdgeFlow builds daily bias from three inputs, checked in order.

1. HTF draw on liquidity. On the daily and 4H chart: - Where are the nearest untaken pools? (PDH/PDL, PWH/PWL, equal highs/lows, old highs/lows.) - Are there unfilled HTF FVGs acting as draws? - What does daily structure say? (HH/HL = bullish draw toward highs.)

2. Previous day's behaviour. A few commonly taught patterns (tendencies, not rules):

3. HTF PD arrays and premium/discount. On the daily dealing range: - Is price in discount of a bullish daily range? Longs are favoured. - Is price in premium of a bullish daily range, right under a daily bearish FVG? Continuation longs are riskier; the draw may be a pullback first.

Write one line before the session:

"Bias: bullish. DOL: PDH 20,250. Invalidation: 4H close below 20,000. Alternative: if PDL 19,950 is taken with displacement, flat for the day."

Mapping the ITF

With bias set, move to the 1H and 15m:

1. Mark the ITF dealing range (last clean impulse in the bias direction).
2. Mark EQ and the OTE band (Chapter t5).
3. Mark qualifying PD arrays in the correct half: bullish FVGs/OBs in discount for a bullish bias; bearish ones in premium for a bearish bias.
4. Mark nearby liquidity that price might sweep first (Asia/London highs and lows, equal highs/lows).

Your ITF map is now a short list of "zones where I'll look for an LTF trigger, after liquidity is taken".

The LTF trigger

On the 5m (or 1m), you wait for the sequence that turns a zone into a trade:

1. Price reaches the ITF zone or sweeps nearby liquidity.
2. Displacement away, producing an MSS with a body close (Chapter t1).
3. A qualifying FVG (or OB) forms in that displacement.
4. You enter on the return to it, stop beyond the sweep, target opposing liquidity.

That's the EdgeFlow setup model in miniature. Chapter t8 turns it into a checklist with grades.

When timeframes disagree

Conflicts are normal. The daily is bullish but the 1H is in a strong pullback; the 5m just printed a bearish MSS. What now?

The rule of thumb: conflict means wait or reduce size, never increase it.

Worked comparison: aligned vs conflicted

The difference isn't the trigger, which is identical in both. It's the context the trigger appears in.

SMT divergence: NQ vs ES

SMT divergence ("Smart Money Technique" in ICT terminology) compares two closely correlated markets. When one makes a new high or low and the other fails to, ICT reads it as a sign of weakness in that move, and a possible reversal.

NQ and ES are the classic pair because they're highly correlated: both track large US stocks.

• Bullish SMT: NQ makes a lower low (sweeping sell-side), but ES makes a higher low (fails to sweep). The selling isn't confirmed across the market. Supports a long.
• Bearish SMT: NQ makes a higher high, but ES makes a lower high. The buying isn't confirmed. Supports a short.

How EdgeFlow uses SMT:

• Confirmation only. SMT never creates a trade by itself. It adds a point in the grading checklist (Chapter t8).
• Same timeframe, same moment. Compare swings on the same chart timeframe, formed at about the same time.
• At key levels only. An SMT at a sweep of PDL during a killzone matters. An SMT in the middle of lunch chop doesn't.

Other pairs: gold vs silver, EURUSD vs GBPUSD, BTC vs ETH. Using SMT with the US Dollar Index (inverse correlation) and other advanced variations is covered in Execution.

A ten-minute top-down routine

1. Daily (2 min): structure, DOL, premium or discount.
2. 4H (2 min): structure, nearest HTF PD arrays, confirm DOL.
3. 1H / 15m (3 min): dealing range, EQ, OTE, qualifying zones, nearby liquidity.
4. Write the bias line (1 min).
5. 5m (2 min): current structure; set alerts at the levels where you expect a sweep.

Then close the HTF charts and wait for your killzone.

Why a model

Every concept in Chapters t1–t7 is discretionary. Two traders looking at the same chart can see different FVGs, different swings and different draws. A model narrows that down to a fixed sequence of questions with yes/no answers. That gives you three things:

1. Consistency: you take the same kind of trade every time, so your statistics describe one thing.
2. Testability: you can backtest it (Chapter t9).
3. Emotional control: the checklist decides, not the last three trades.

The EdgeFlow model is a version of a sequence widely taught in ICT material: bias → time → liquidity sweep → displacement and shift → entry at a PD array → stop beyond the sweep → target the opposing liquidity.

The seven steps

Step 1. HTF bias and draw on liquidity. Before the session, write the bias line (Chapter t7): direction, DOL, invalidation and alternative. No bias line, no trading.

Step 2. Wait for the killzone. Trade only inside your chosen window: London (02:00–05:00 ET; 10:00–13:00 GST summer / 11:00–14:00 GST winter) or NY AM (07:00–10:00 ET; 15:00–18:00 / 16:00–19:00 GST), including the Silver Bullet hour that follows it (03:00–04:00 or 10:00–11:00 ET). Setups that complete outside your window are journalled, not traded.

Step 3. Liquidity sweep. Price must take a meaningful pool against your bias first: for a long, sell-side below (Asia low, London low, PDL, equal lows); for a short, buy-side above. This is the manipulation leg (Chapter t6).

Step 4. Displacement and MSS on the LTF. After the sweep, price must displace in your bias direction and close a candle body beyond the most recent LTF swing (5m by default). That's the MSS (Chapter t1).

Step 5. Entry at a qualifying FVG or OB in discount/premium. Mark the FVG created by the displacement (or the OB at its origin). It must be in discount for a long or premium for a short, measured on the dealing range from the sweep extreme to the displacement extreme. Default entry: limit order at the FVG's CE. OTE (62–79%) overlap is a bonus.

Step 6. Stop beyond the sweep. Place the stop beyond the sweep extreme (or the swing that created the MSS, if that's further), plus a small buffer: about 2–5 points on NQ, 0.5–1.0 on gold. If that stop is too wide for your risk budget, reduce contracts or pass. Never tighten the stop to make the size fit.

Step 7. Targets at opposing liquidity, minimum 2R, partials. The first target is the nearest meaningful opposing pool, and it must be at least 2R from entry. Default management: take 50% off at target 1, move the stop to breakeven, and hold the rest to target 2 (the DOL or a −0.27/−0.62 extension). If target 1 is less than 2R away, there's no trade.

Grading: gates, score and size

Hard gates. Every one must be "yes". A single "no" is a no-trade, whatever the score.

Score (0–10). Once the gates pass, score the quality:

Grade and size.

The R:R calculator

For every trade, before you enter:

• Risk (points) = |entry − stop|
• R for a target = |target − entry| ÷ risk
• Contracts = floor(risk budget in $ ÷ (risk points × point value))
• Blended R with partials = sum of (fraction closed × R at that exit)

Point values from Foundations: NQ $20, MNQ $2, ES $50, MES $5, GC $100, MGC $10, CL $1,000, MCL $100.

Worked trade 1: NQ long, NY AM

Worked trade 2: gold short, London

Trade 2 is a useful reminder: a perfectly executed A+ setup doesn't always reach the final target. The partial at T1 is what turned it into a winner instead of a scratch.

Worked example 3: the no-trade

Managing the trade

• Before entry: limit order at CE, stop and both targets placed as a bracket (Foundations f6). If price runs to T1 without filling you, let it go. Don't chase at market.
• Cancel the limit if price closes beyond the opposite edge of the FVG before filling, or if the killzone ends.
• After T1: move the stop to breakeven. Don't move it closer until T2 or a new structure break justifies it.
• Never widen the stop. If the stop is hit, the idea was wrong. Journal and wait for the next setup.
• Maximum two trades per session; stop for the day after two losses or at your daily loss limit, whichever comes first.

Pre-session checklist

1. Economic calendar checked; news times noted in ET and GST.
2. Daily and 4H structure and DOL marked.
3. Bias line written (direction, DOL, invalidation, alternative).
4. PDH/PDL, Asia and London highs/lows, midnight open marked.
5. ITF dealing range, EQ and qualifying zones marked.
6. Killzone set; alerts placed at expected sweep levels.
7. Risk budget set for the day; daily loss limit confirmed.

Why backtest a discretionary model

ICT concepts are discretionary: they need judgement. That's exactly why they need testing. Without numbers, you'll remember the beautiful A+ winners and forget the ten B setups that stopped out. A backtest replaces memory with a record.

A backtest won't tell you the future. It tells you whether your version of the model, applied the way you actually apply it, had a positive expectancy on past data, and roughly how bumpy that was. That's enough to decide whether to risk real money and how much.

How to backtest: the process

1. Write the rules first. Use the Chapter t8 model and write down every definition you'll use: swing rule (three candles), break rule (body close), displacement rule (body ≥ about 1.5–2× average), FVG filter, entry (CE limit), stop (beyond sweep + buffer), targets (T1 ≥ 2R at liquidity, 50% partial, BE, T2 at DOL), killzone, news rule. If a rule isn't written, you'll bend it.

2. Pick the market, timeframes and period. One instrument (e.g. NQ), one pairing (daily/4H → 1H/15m → 5m), one killzone. Choose a period that covers different conditions: trending months, ranging months, high-volatility news months.

3. Replay bar by bar. Use your platform's replay mode so you can't see the future. Mark the bias before the session starts in replay, exactly as you would live.

4. Record every setup, including no-trades. For each one, log the fields in the table below.

5. Collect at least 100 setups. Fewer than that and the numbers are mostly noise (see the sample-size note below).

6. Analyse. Expectancy overall, by grade, by killzone, by day of week. Then change one rule at a time and re-test.

Sample size: why 100 is a minimum

Suppose your true win rate is 40%. After 100 trades, the standard error of your measured win rate is √(0.40 × 0.60 ÷ 100) ≈ 0.049, about 5 percentage points. A rough 95% range is ±1.96 × 4.9% ≈ ±9.6 percentage points. Your backtest could show anything from about 30% to 50% purely by chance.

At 20 trades, that range is so wide the result is close to meaningless. That's why EdgeFlow treats 100 setups as the minimum before trusting any statistic, and 200+ before splitting results by grade or killzone.

Expectancy

Expectancy is the average result per trade, in R:

Expectancy (R) = win rate × average win (R) − loss rate × average loss (R)

Where loss rate = 1 − win rate (treat breakeven trades as a separate group, or as wins/losses of 0R, but be consistent).

Breakeven win rate. For a given average win (in R, with average loss = 1R):

Breakeven win rate = 1 ÷ (1 + average win R)

At an average win of 2.5R, breakeven = 1 ÷ 3.5 = 28.6%. At 2R, it's 33.3%. At 3R, it's 25%. This is why the 2R minimum in the EdgeFlow model matters: it lets you be wrong most of the time and still have positive expectancy, if the win rate holds.

Losing streaks and drawdown

Even a positive model has long losing streaks. You need to know how long before they happen, so you don't abandon a working model at the worst moment, or keep trading a broken one.

Probability of the next N trades all losing = loss rate^N. At a 40% win rate (60% loss rate):

Computed exactly; the 100-trade column uses a run-length calculation, not 0.6^N.

Read that again: at a 40% win rate, a streak of 8 losses in a row has roughly a coin-flip chance of appearing somewhere in 100 trades. In simulation, the typical longest losing streak over 100 trades at 40% is about 8. At a 50% win rate, a streak of 6+ still happens about 55% of the time.

Drawdown from a streak:

Streaks also come from regime change: the market moves from trending to ranging and your model stops working for a while. The Monte Carlo simulation in the backtest lab shows a spread of equity curves from the same statistics: some smooth, some with painful drawdowns. All of them are "the same model".

From backtest to forward test

A backtest has known biases: you know roughly what happened next, replay fills are perfect, and you're relaxed. Forward testing removes those:

1. Demo or sim, live market, real time. Same rules, same killzone, same journal. Minimum 30–50 trades, ideally the full 100.
2. Compare to the backtest. Is the win rate within the expected range? Is average R similar? Are you following the rules (rule-followed rate above 90%)?
3. Small live step. If forward results are in line, trade the smallest real size (one MNQ or one MGC) for another 30–50 trades. Real money changes behaviour, and that's what you're testing now.
4. Scale gradually. Increase risk only after each stage confirms the last.

If forward results are much worse than the backtest, the most common reasons are: hindsight in the backtest, unrealistic fills, or not following the rules live. Check the journal's "rule followed?" column first.

Common ICT/SMC traps

A 30-day practice plan

At the end of 30 days, you should have: a written model, 100+ backtested setups, 10–20 forward-tested trades and an expectancy number you trust more than your feelings. Bring your journal to the EdgeFlow review channel; that's where Execution begins.

Review the moduleGlossary flashcards →