Accumulation, manipulation, distribution

The sequence, and why it appeals

The idea cuts a session into three acts. First a quiet range where price goes nowhere. Then a brief move that leaves that range on one side and triggers the stops sitting there. Finally the real move, in the other direction, the one that goes somewhere.

Accumulation manipulation distribution is the full name, shortened to AMD almost everywhere, and the methods built on it come down to recognising which act is running so you can position before the third one.

What makes the reading appealing is that it tells a complete story with a beginning, a middle and an end. Brains keep narratives better than statistics, and that is exactly why this one deserves more suspicion, not less.

The skeleton rests on a verifiable fact, the one from the lesson on large orders and how they execute: a big execution needs counterparty, and counterparty piles up where stops pile up. So far so good.

What is laid on top, however, is a prediction: that this counterparty will be sought at a particular moment of the session, and in a particular order. That is the part that needs proving, and it is the part nobody proves.

Accumulation: the quiet range

The first act is a narrow band, often several hours long, where price oscillates without direction. The reading holds that a participant of size builds their position there without moving price, which is exactly the constraint described elsewhere in this course.

On the chart it looks like deep boredom: short candles, wicks on both sides, average or low volume. Plenty of traders lose money precisely during this act, hunting movement where there is none.

What you can check objectively is the range: its bounds are two prices, they can be read, and two people find the same ones. What you cannot check is the intent attributed to anyone inside it.

The distinction matters because it separates what is measurable from what is narrated. The range is measurable. The reason for the range is not, and nothing on your screen will tell you who was buying.

⚠️ A quiet range announces nothing on its own. Most quiet ranges are followed by more quiet ranges. That is not what the video you watched showed, because it only showed the ranges that were followed by a move.

Manipulation: the false move

The second act is the one that gives the whole family its bad name. Price leaves the range on one side, takes the stops that were sitting there, then comes back. The word suggests hostile intent, and it is better heard another way.

What happens mechanically is simpler: triggered stops supply counterparty in quantity, and an execution looking for it goes where it is. Nobody is aiming at your stop in particular, they take what is available, and what is available is always in the same place.

That is what the lesson on liquidity and smart money covers in full. A sweep is not malice, it is a place on the chart where counterparty had piled up.

The distinguishing sign of this act, when it exists, is the speed of the return. A break of the range that does not hold for more than a few candles before slipping back inside looks like a liquidity grab. A break that holds is simply a break.

⚠️ And that is where the reasoning eats its own tail: you only know whether the break held afterwards. A break that held was not manipulation, it was the move. The same drawing therefore carries two names depending on what follows.

Distribution: the real move

The third act is the decisive move, in the direction opposite to the false one. It is the one the reading is trying to catch, and it is also the one you only see in the examples because it happened.

What the move leaves behind, on the other hand, is perfectly readable: areas crossed without counterparty, which is to say imbalances. The lesson on the inverse fair value gap explains what is done with them next.

The joint between the two is the only place in the corpus where the concepts genuinely answer each other: the sequence says where to look, the imbalance says at what price. Without the second, the sequence yields no usable level at all.

In practice, most sessions do not produce all three acts. They produce a range and nothing else, or a move with no prior range, or two false moves in a row. The model describes a particular case, not the general rule.

That does not disqualify it. It means you have to count how often it actually occurs on your market and at your hour before making it the skeleton of your day.

The box setup, and session timing

Drawing all of this on a chart is what people call a box setup AMD ICT trading pattern. You box the accumulation range, wait for one side to be swept, and take the return back through the box as your signal. It is the most widely taught form, and the easiest to draw.

The literature almost always pairs the sequence with session times. London is treated as the false-move window, New York as the real one. That pairing is where AMD forex gets its following, since currency sessions are clean and their hours stable.

That timing claim is the most testable part of the whole model, and that is good news. An hour is objective data: you can count, across a hundred sessions, how often the low of the day fell inside the stated window.

Do that count before believing anyone on the subject, this page included. The result depends on the instrument, and it does not travel: a US index and an Asian pair have no reason to share an internal calendar. Most AMD model trading material skips this step entirely.

What you get is a percentage, not a certainty. A pattern that shows up in four sessions out of ten stays usable if the reward-to-risk covers it, and unusable if it does not. That is arithmetic, not belief.

The flaw, and it decides everything

The flaw in any AMD trading strategy shows up the second you try to run it live: you only know which act you were in once the session has closed.

A range that gets swept and then keeps going was never accumulation, it was the move itself. Marked in hindsight, accumulation manipulation distribution explains almost every session. Marked in advance, it is a claim you have to demonstrate.

It is the same bias described in the lesson on imbalances, and it bites harder here: a three-act sequence offers three chances to redefine what you were looking at, and an observer acting in good faith will take them without noticing.

The guard is easy to state and tedious to keep: decide in writing, before the open, which bounds you are using and what would invalidate your reading. A prediction written beforehand cannot be rewritten afterwards.

⚠️ An explanation that stays true whatever happens teaches you nothing. That test runs through this entire course, and an AMD strategy is more exposed to it than any other concept in the corpus.

Practising: fifty sessions marked live

Pick one instrument and one session. Every day, before the window you watch opens, write down three things: the bounds of the range you are using, the side you expect to be swept, and the level that would tell you that you are wrong.

Then watch, and record what happened. Not what you would have done, what happened. Three columns are enough: swept on the expected side, swept on the other side, no sweep at all.

Do it fifty times. That is two and a half months of sessions, and it is the price of an answer nobody else can give you, because it depends on your market, your hour and the way you draw a range.

Then compare the three columns. If the expected side does not come out more often than the other, your reading adds nothing to a coin toss, and you have just saved yourself years. That is what an honest AMD trading strategy test looks like.

If it does come out more often, you have something, and what remains is checking that the edge covers your costs. A trading journal that sorts your trades by setup does that arithmetic for you.