Power of Three (AMD) Strategy

Power of Three (often abbreviated AMD: Accumulation, Manipulation, Distribution) is a widely-taught smart-money / price-action concept that frames how a trading session or candle is said to unfold in three phases. This breakdown explains the idea neutrally, how traders define entries, exits, and risk around it, and why it is a discretionary framework rather than a mechanical signal. It is educational only — no part of it predicts price or implies profit, and trading carries risk of loss.

How it works

Power of Three (PO3), commonly written as AMD, is a way some traders mentally segment a session or a higher-timeframe candle into three sequential phases. It is popularised in the smart-money / ICT-influenced trading community as a teaching model, not a proven mechanism.

The three phases, as traders describe them:

Accumulation — a relatively range-bound, lower-volatility period (often associated with a session open or the early development of a higher-timeframe candle) where positions are said to be "built." Price chops inside a developing range.

Manipulation — a push outside that range, frequently described as a "false move," "judas swing," or stop run, that traders believe sweeps liquidity sitting beyond the accumulation high or low before reversing. This is the phase the concept hangs on: the idea that the first decisive move out of the range can be a trap rather than the real direction.

Distribution — an expansion move in the opposite direction to the manipulation, treated by traders as the session's "intended" direction, often described as ending near a session high/low or a higher-timeframe target.

Traders typically map this onto fixed reference windows: a daily candle is described in terms of its open, a misleading wick, and the body/close; or an intraday session (for example, a chosen futures or FX session open) is watched for an early range, a sweep of that range, and a subsequent expansion. The concept is descriptive and pattern-based — it labels structure after the fact and offers a lens for considering that an early breakout might reverse. It is not a formula and does not output a numeric signal. Crucially, identifying which phase you are in is only unambiguous in hindsight; in real time it is a judgement call, which is why this reduces to a discretionary framework rather than a mechanical strategy. None of these phases is guaranteed to occur, and labelling one does not forecast what price will do next.

Entry rules

Define the accumulation reference range first: traders commonly mark the high and low of an early window (e.g. the first part of a chosen session or the prior consolidation) before looking for any setup.
Wait for a manipulation move: a push beyond the accumulation high or low that some traders interpret as a liquidity sweep rather than a genuine breakout. Under this framework no position is taken on the breakout itself.
Look for a reversal back inside the range after the sweep — many traders require confirmation such as a market-structure shift, a return inside the range, or a lower-timeframe break of structure before considering a position in the opposite direction to the sweep.
Bias the trade toward the expected distribution direction (opposite the manipulation). Some traders additionally require alignment with a higher-timeframe bias or a session open price before acting.
Because phase identification is discretionary, the 'entry' is a trader judgement, not an automatic trigger — the illustrative Pine script below only marks candidate sweeps; it does not place trades or tell you when to enter.

Exit rules

Stops are commonly placed beyond the extreme of the manipulation/sweep (above the swept high for shorts, below the swept low for longs), so invalidation is defined by price continuing past the supposed trap.
Targets are often set at the opposite side of the accumulation range, at a session high/low, or at a prior higher-timeframe level traders treat as liquidity.
Some traders scale out partially at the range midpoint or first opposing liquidity pool and trail the remainder.
Time-based exits are common: if the expected distribution does not develop within the chosen session window, some traders treat the idea as invalid and close the position.
All of these are trader-defined choices, not rules with any predictive guarantee — none of them implies a positive outcome, and any stop can be exceeded with slippage.

Risk notes

This page is educational only and is not financial advice. Nothing here is predictive, and trading carries a real risk of loss — losses can exceed your initial outlay on leveraged products.
Any backtest or marker output is illustrative; past behaviour does not indicate future results.
Position sizing, stop placement, and whether to trade at all are personal risk decisions; the AMD framework does not manage risk for you.
The smart-money / ICT terminology referenced is community vocabulary; trademarks belong to their respective owners, and nothing here implies endorsement of, or affiliation with, any individual or group.

Pine v6 example

//@version=6
// ILLUSTRATIVE ONLY — this is a detection/marker script, NOT a strategy.
// Power of Three (AMD) is a discretionary framework, so it does not reduce
// to a single mechanical entry/exit. This script merely (a) marks a chosen
// session's early "accumulation" range and (b) flags candidate liquidity
// sweeps of that range. It places no orders and predicts nothing.
indicator("Power of Three (AMD) - Illustrative Marker", overlay = true)

// --- Inputs -------------------------------------------------------------
accSess   = input.session("0930-1000", "Accumulation window (exchange time)")
fullSess  = input.session("0930-1600", "Active session window")
showRange = input.bool(true, "Show accumulation range")

// --- Session helpers ----------------------------------------------------
inAcc    = not na(time(timeframe.period, accSess))
inSess   = not na(time(timeframe.period, fullSess))
accStart = inAcc and not inAcc[1]   // first bar of the accumulation window

// --- Build the accumulation range --------------------------------------
// Levels reset at the start of each accumulation window so one session's
// range is not carried into the next day.
var float accHi   = na
var float accLo   = na
var bool  sweptHi = false
var bool  sweptLo = false

if accStart
    accHi   := high
    accLo   := low
    sweptHi := false
    sweptLo := false
else if inAcc
    accHi := math.max(accHi, high)
    accLo := math.min(accLo, low)

// --- Flag candidate manipulation (sweep) after accumulation ends --------
// Uses only the current, completed bar's high/low vs. the stored range.
// No higher-timeframe lookahead and no reference to an unformed value.
postAcc = inSess and not inAcc and not na(accHi)
sweepUp = postAcc and not sweptHi and high > accHi   // swept the range high
sweepDn = postAcc and not sweptLo and low  < accLo   // swept the range low
if sweepUp
    sweptHi := true
if sweepDn
    sweptLo := true

// --- Plot ---------------------------------------------------------------
plot(showRange and not na(accHi) ? accHi : na, "Acc High", color.new(color.teal, 0), 1, plot.style_linebr)
plot(showRange and not na(accLo) ? accLo : na, "Acc Low",  color.new(color.teal, 0), 1, plot.style_linebr)

plotshape((sweepUp) and barstate.isconfirmed, "Sweep of high", shape.triangledown, location.abovebar, color.new(color.red, 0),   size = size.tiny)
plotshape((sweepDn) and barstate.isconfirmed, "Sweep of low",  shape.triangleup,   location.belowbar, color.new(color.green, 0), size = size.tiny)

// Note: a sweep marker is NOT a buy/sell signal. Confirming whether a sweep
// becomes "manipulation -> distribution" is a discretionary judgement the
// trader makes; this tool only highlights candidates for study. Past
// behaviour does not predict future price, and trading risks loss.

Pitfalls

Hindsight labelling: the three phases can look obvious on a finished chart but are ambiguous in real time. A 'manipulation' sweep can simply be the start of a genuine trend, stopping out anyone who faded it.
Not every session forms a clean AMD pattern. Forcing the model onto every candle or session can lead to over-trading and seeing structure that is not there.
Session and timezone dependence: the concept relies on chosen reference windows. Mismatched exchange/session times or daylight-saving shifts can change or invalidate the setup. The script uses the chart symbol's exchange time, so confirm it matches your intended session.
It is a framework, not a signal generator. Marking a sweep is straightforward; judging it a trap is discretionary and not testable as a single rule.
Repainting / look-ahead risk when coding it: range levels and sweep flags must use only completed information. Improper higher-timeframe requests (e.g. request.security with lookahead enabled) or referencing an unformed bar create look-ahead bias. Validate for repainting before trusting any backtest.
No predictive or profit claim: AMD describes observed structure; it does not forecast direction or imply any win-rate, return, or edge.

FAQ

What does AMD stand for in the Power of Three strategy?

AMD stands for Accumulation, Manipulation, Distribution — the three phases traders use to describe how a session or higher-timeframe candle is said to develop: a build-up range, a misleading move that is interpreted as sweeping liquidity, then an expansion in the opposite direction. It is a descriptive model, not a guarantee that price will behave this way.

Is Power of Three a mechanical strategy I can fully automate?

Not cleanly. Identifying the accumulation range can be automated, and you can flag candidate sweeps mechanically (as the illustrative script does), but deciding that a sweep is genuine 'manipulation' followed by 'distribution' is a discretionary judgement. That is why the Pine example here is a marker/detection tool, not an order-placing strategy.

Does the Power of Three concept predict which way price will go?

No. It is a lens for interpreting structure after a possible liquidity sweep, popularised in the smart-money community. It does not forecast direction, does not imply any win-rate or profit, and trading based on it still carries risk of loss. Treat it as a study framework and validate any coded version for look-ahead bias and repainting before relying on it.

Want this as a clean, validated Pine script? run it through the Validator → or convert a trading strategy to Pine Script →

Educational & software only — not financial advice, not a recommendation to trade. Backtests are illustrative; past performance does not predict future results. Trading involves substantial risk of loss.