What Power of 3 (AMD) actually is
Power of 3 — AMD, for Accumulation, Manipulation, Distribution — is the three-phase cycle ICT traders credit to Michael Huddleston as a model for reading institutional flow from price. The model reads every timeframe, from monthly swings to one-minute charts, as the same cycle: a range where positions are read as being built (accumulate), a failed break that takes the liquidity at one edge (manipulate), then the directional move (distribute). Read this way, it is less a pattern than a sequence — and once you can name the sequence, you can build a checklist against it.
This article turns that theory into the checklist: three steps, each with an objective trigger, that convert the Power of 3 cycle into a repeatable trading routine.
Step 1 — Locate the accumulation range
Accumulation looks like boredom: a compressing range, overlapping candles, no follow-through in either direction. Classic containers include the Asian session, the pre-news drift, and the consolidation under a prior day's level. Two features matter operationally:
- The extremes. The range high and low become pools — stops are assumed to build up behind both as the range ages. These are your reference liquidity levels.
- The midpoint. The range's equilibrium will act as the pivot for judging the later reversal.
Checklist trigger: a defined range with at least two touches per side, sitting inside a known session container. No range, no framework — stand down.
Step 2 — Wait for the manipulation
The manipulation is the range break that fails — the raid of one extreme, read in this model as a large position being filled at better prices. It is the Judas swing when it happens at a session open, and it follows the anatomy of a sweep: raid, absorption, close back inside.
Checklist trigger: price takes a range extreme and then closes back inside with displacement — a fair value gap in the reversal direction, ideally launched from an order block. Until that close prints, the break is a breakout, not a manipulation; the framework forbids anticipating it.
Amateurs trade the break. The framework trades the failure of the break.
Step 3 — Trade the distribution
With the raid confirmed, the market enters its honest phase: distribution, the directional leg that the first two phases financed. Execution is now mechanical:
- Entry on the confirmation close (or its immediate retest).
- Stop beyond the manipulation extreme — the price at which the reading is invalidated.
- First target at the range equilibrium; final target at the opposite external pool — the draw on liquidity.
- Management in R: partial at a fixed multiple, break-even once price has moved a set distance in favour, statistics recorded per setup — the discipline detailed in prop-firm risk management.
The 3-step checklist
- 1. Range: defined container, two touches per side, known session.
- 2. Raid: extreme taken → close back inside + displacement. No close, no trade.
- 3. Distribution: stop beyond the raid, targets at equilibrium then the opposite pool, results in R.
What "systematic" really buys you
The framework's value is not prediction — plenty of ranges resolve without a clean raid. Its value is refusal: three gates that filter out every trade that lacks the full sequence. Refusing a setup is hard to do consistently by hand under time pressure; a rule written in code applies it the same way every time. SWEEP PROTOCOL exists for that reason: a built-in Power of 3 (AMD) filter watches the containers, refuses the unconfirmed breaks, and only arms a signal once accumulation, manipulation and the close back inside have printed — then logs each signal's outcome in R, so your own statistics build up.
Recognizing the three steps is only half the job — not every cycle that technically completes them deserves the same risk. The A+ setup is the follow-up: a 5-point checklist, read on real US100 and BTCUSD charts, after the close, for grading a sequence before you size it.
What this article does not do
- No prediction. A setup that meets every condition described here can fail.
- Weak contexts. Around session opens, scheduled releases and thin hours, levels can be traded through and fills can be worse than planned.
- No recommendation. Nothing here is an instruction to trade a specific instrument or level.