ICT setups: turtle soup, silver bullet

Stage names for known chains

No named setup adds a new object. A sweep is still a liquidity sweep, an impulse is still a displacement, a zone is still a zone. What the name adds is an order: which bricks, in which sequence, at what hour.

That is useful, and it is exactly what to ask of them. A named sequence is a written sequence, hence countable in a journal. The danger lies elsewhere: collecting names like secret techniques, when whoever knows the bricks can rebuild every setup in one sentence.

The skeleton shared by all five setups A sweep, a reversal, an entry into the zone left behind: this skeleton is the one all five named setups share. None of them adds an object to the chart, each adds a condition, and it is that condition, never the name, that has to be written down before you open the spreadsheet.

Turtle soup: the failed breakout, reversed

The oldest of the five is not ICT's. Turtle soup trading comes from Linda Raschke, in the 1990s, and the name is a joke: the Turtles, Richard Dennis's students, bought breakouts of twenty-day highs. Raschke made soup with them: take the opposite trade when the breakout fails.

The ICT turtle soup is the same idea reread with the corpus vocabulary: a sweep of an obvious extreme, then a reversal, played as an entry. It is a liquidity sweep traded as a reversal, nothing more, and the sweep lesson already contains everything needed to evaluate it, including the written criterion that separates a sweep from a break that holds.

Turtle soup: the criterion is the close Breakout buyers enter above the high, and their stops sit just under it: that pool of orders is what the move back comes for. What separates the soup from a break that holds is not the look of the move, it is a close back under the level. Without that written criterion, the same chart reads either way once the day is over.

The judas swing: the session's false start

The judas swing is a sweep with a time constraint: the fake move of a session's first hours, the one that goes one way, fills the liquidity sitting there, then lets the real direction unfold the other way. The name says the betrayal well: the day's first move is the one not to follow.

In bricks: the manipulation from the AMD sequence, dated by the killzones. What gets tested is precise: over a hundred sessions, how often is the first move of the chosen window undone before the morning ends.

The judas swing: the first move is the wrong one The session's first move goes after the liquidity sitting above the night range, then the day sets off the other way. This is not a prediction, it is a count, and it applies to ONE time window chosen before the open. Without that hour written in advance, a judas swing only becomes distinguishable from an ordinary sweep once the session is over.

The silver bullet: a one-hour window

The silver bullet ICT setup is the most mechanical of the five: a one-hour window, the most cited running from 10 to 11 am New York time, in which one waits for a sweep then a fair value gap the opposite way, and takes the entry inside that gap. It also exists in a London version and an afternoon version.

That is the whole ICT silver bullet strategy: a time macro, a sweep, an FVG. Its virtue is not magic, it is methodological: everything is written in advance, the hour, the trigger, the entry level, which makes it the simplest setup to count honestly over fifty cases.

The silver bullet: four conditions written in advance A fixed hour, a sweep inside it, an impulse the other way, an entry into the gap it leaves. Its value is not in the hit rate, it is in the fact that all four conditions are written before the open: a setup you recognise in the evening cannot be counted, this one can, over fifty cases.

Unicorn: the breaker that overlaps a gap

The unicorn model ICT is an overlap: a breaker block and a fair value gap covering the same place. The zone common to both serves as the entry level, the idea being that two independent reasons to come back to the same price beat one.

The figure below shows the overlap. It is the most recent setup in the corpus and the least documented; treating it as a curiosity to count, rather than a demonstrated improvement, is the honest position.

The unicorn: two zones, one place A breaker block, the order block that failed, and a fair value gap left by the crossing partially overlap. The shared zone is the level the setup keeps: two independent reasons to come back to the same price.

MMXM: the market maker models

The acronym MMXM, market maker models, names two symmetrical curve profiles. The market maker buy model: a stepped decline, a low zone where the move turns, then a climb back through the same levels. The market maker sell model is the mirror, a top instead of a bottom.

It is the AMD sequence drawn as a full curve, with one more piece of vocabulary: the decline is called the sell side of the curve, the climb the buy side, and the central turn often runs through a sweep, what other schools call a smart money reversal.

⚠️ It is the most retrospective of the five: a complete curve is only recognised once closed, and failed half-curves leave no name behind. Counting it therefore requires a precise entry definition, not a silhouette.

The market maker buy model, and what it hides A stepped decline, a low that sweeps the previous one, then a climb back through the levels of the decline. The sell model is the exact mirror. Its weakness is in the drawing itself: a complete curve is only recognisable once it is finished, and the halves that never turned leave no trace in the examples you get shown.

The market maker sell model, the exact mirror

A market maker sell model is the same object flipped: a climb in steps, a top that sweeps the liquidity above an obvious high, a decline that runs back through the steps of the way up. Everything that holds for the buying side holds here with the direction reversed, the retouched-steps criterion included. That is also why the two get treated together: one set of rules, two orientations.

You often read that the climb is slow and the fall is fast. That may be true on your market, it is not a law. The only way to know is to count the candles of each half on your own records, then compare the two medians. Until you have done that, treat both halves as symmetrical: it is the least costly assumption when you turn out to be wrong.

The trap in this model is a selection bias, and it is a violent one. A curve that never turned gets no name: it stays a trend, nobody publishes it, nobody counts it. Only complete curves end up in the examples, which makes the model splendid in hindsight and a lot duller live. To measure it honestly you have to count the failures, so record every sweep in your window, not only the ones that produced a nice curve.

The smart money reversal: what the term actually names

The smart money reversal names the tipping point of an MMXM curve, the place where the sell side becomes the buy side. The term circulates well beyond the ICT corpus, and it is most often used without a definition, which makes it impossible to count. Reduced to what can be observed, it comes down to three dated facts in this order: a sweep of an extreme, a displacement the opposite way, then a change of structure confirmed on your working timeframe.

A bounce without a sweep is not a reversal, it is a breather. A green candle after six red ones is no more one. What separates the two is a time limit: fix the maximum number of candles between the sweep and the change of structure, five for instance, and write it down before you start counting. Without that limit, any decline followed by a rally ends up filed as a reversal, and the hit rate you measure stops measuring anything at all.

A smart money reversal stays a state of the structure, never a buy order. It says the context has changed, not where to enter or where to put your stop. The entry is taken afterwards, in a zone defined separately and in advance, and it is reading the liquidity that will have told you which extremes were worth sweeping. Confusing the two is the fastest way to enter early and get taken out on the sweep itself.

Count them, do not collect them

Five names, one discipline: pick one setup, write its definition in verifiable conditions, and count it over fifty cases in a journal before adding a second. Setup collecting is the classic symptom of the trader looking for the recipe instead of measuring their own.

The recommended learning order follows from the lessons: the silver bullet first because everything in it is written in advance, the turtle soup next because it only requires a sweep criterion, and the MMXM curves last, because they demand the most live judgement.

Key takeaways

No named setup adds a new object: each is an order imposed on bricks this course has already explained.
Turtle soup: a liquidity sweep played as a reversal. The name is Linda Raschke's, against the Turtles' breakouts.
Judas swing: the session's false start, the AMD manipulation dated by the killzones.
Silver bullet: a one-hour window, a sweep, an FVG. The simplest to count because everything is written in advance.
Unicorn: a breaker and an FVG overlapping, two independent reasons to come back to the same price.
MMXM: the AMD sequence as a full curve, the most retrospective of the five. A setup gets counted, a collection of setups gets paid for.

Going further

These blog articles dig into this lesson's ideas, one subject per article.

Supply and demand zones explained The patterns in your trading: spotting what makes you win or lose Trading rules: how to define them and stick to them Trading journal: the complete guide to improving (2026)