I lost $1,240 on a Thursday that taught me the most expensive lesson of my funded career. GBPUSD spent the whole Asian session crawling sideways between 1.2710 and 1.2730 — a dead 20-pip box I ignored while waiting for London. At 08:00 the pair ripped through 1.2730, printed 1.2755, and I bought the breakout like my plan told me to. Ninety minutes later I was stopped out at 1.2718, down 0.5%, and by the New York session the pair was at 1.2640 — one hundred and fifteen pips below my entry. The breakout had been a lie. The rip upward was not buyers arriving; it was the market raiding the buy stops above the Asian high, collecting liquidity, then distributing downward all day. That three-act structure has a name in ICT concepts — the Power of 3 — and once I learned to read it, my funded win rate changed completely.
In this article
- What the Power of 3 Actually Is
- The Three Phases on a Real Chart
- My 5-Rule PO3 Plan
- The Sizing Maths Under a 5 Percent Daily Loss Limit
- I Tested It: 10 PO3 Days on a 100k Demo
- When the Power of 3 Fails
- How PO3 Fits Prop Firm Drawdown and Consistency Rules
- My Daily PO3 Routine
- Power of 3 Strategy FAQ
The Power of 3 says every clean trading day has three acts: Accumulation, Manipulation, Distribution. Price coils into a tight range and builds resting orders on both sides (accumulation), spikes beyond one side to raid those stops (manipulation), then expands hard in the true direction (distribution). On a prop firm account this matters more than anywhere else, because the daily loss limit punishes exactly the two PO3 mistakes: buying the manipulation spike, and trading the accumulation chop. This guide is the full funded-account treatment: each phase on a real chart, my five-rule plan with exact entry, stop and target placement, sizing maths under a 5% daily loss limit, a ten-day worked log from my 100k demo, the five ways the model fails, and how it fits drawdown and consistency rules.
What the Power of 3 Actually Is
ICT’s Power of 3 (PO3, sometimes written AMD) is a daily market-structure model, not an indicator: sideways first to build liquidity, a false push to harvest it, then the real directional expansion. The profitable trade is always in the third act.
Each phase has a job. Accumulation is the quiet box: price ranges, volatility dies, and stop orders pile up above the range high and below the range low — on most FX majors this happens during the Asian session, roughly 00:00 to 07:00 GMT. Manipulation is the raid: usually into the London open, price spikes beyond one side of the box, triggers those resting stops, and often closes back inside the range. Distribution is the payoff: a displacement candle launches in the opposite direction of the raid, and price trends for hours. Your job is to miss the first two acts and trade only the third.
The model connects to ideas already on this site. My kill zone strategy maps the time windows where each act usually plays out. The turtle soup strategy trades the manipulation act itself. PO3 is the bigger frame containing both: it tells you which failed breakout is worth trading — the one completing the three-act cycle. And it mirrors my London breakout plan in reverse — where that plan buys the first break of the Asian range, PO3 expects that first break to be the lie and trades the reversal after.
The Three Phases on a Real Chart
Let me walk the model through a real GBPUSD day I traded in October — the actual levels:
| Phase | Time (GMT) | What price did | The tell |
|---|---|---|---|
| Accumulation | 00:00 – 07:00 | Ranged 1.2710 – 1.2730 (20 pips) | Seven hours, no candle body longer than 8 pips |
| Manipulation | 08:00 – 08:40 | Spiked to 1.2755, then closed back under 1.2730 | Wick raided the Asian high; H1 candle closed back inside the box |
| Distribution | 09:00 – 16:00 | Fell to 1.2640 (100 pips) | Displacement: a 40-pip bearish H1 candle with almost no lower wick |
Read that table like a detective. During accumulation the range was so tight it felt boring — that boredom is the liquidity building. The manipulation spike lasted forty minutes and its only achievement was printing 1.2755, a level that existed purely to execute the buy stops above 1.2730. The giveaway was the close: the hourly candle wicked to 1.2755 but closed back inside the box. Smart money had finished shopping. Then distribution did the honest work — 100 pips down, no looking back.
Three checks separate a real PO3 day from noise. First, the box must be genuinely quiet: under 30 pips wide on the majors, at least four hours, mostly inside the Asian session. Second, the raid must hit a visible pool — the Asian high/low or the previous day high/low. A random 15-pip wiggle is not manipulation. Third, distribution must announce itself with displacement: a fast, large-bodied candle in the true direction. Slow drift after the sweep is just more chop, and trading it is how accounts die.
My 5-Rule PO3 Plan
Here is the complete plan, exactly as I run it — every rule exists because I broke its earlier version.
Rule 1 — Bias first, from the dealing range. Before London, mark the previous day’s high and low. Price in the top third (premium): short the distribution leg only. Bottom third (discount): buy it only. No bias, no trade — a PO3 without bias is a coin flip with extra steps.
Rule 2 — The box must be real. Accumulation qualifies only if it is at least four hours long, no wider than 30 pips on majors (about 40 on GBPJPY and gold), and formed mostly between 00:00 and 07:00 GMT. A two-hour 45-pip drift is a slow morning, not accumulation.
Rule 3 — The raid must take a visible pool. Manipulation counts only if it spikes beyond the Asian high/low or the previous day high/low, prints a wick there, and closes back inside. The close back inside is non-negotiable — it separates a stop raid from a genuine breakout.
Rule 4 — Enter on the retest, never on the sweep. After the raid, wait for price to retest the manipulated level and show rejection. Stop beyond the manipulation extreme plus 5 pips buffer; target the opposite side of the dealing range or the previous day extreme, whichever comes first. Never enter during the spike — fading the sweep donates to the liquidity pool.
Rule 5 — Kill-zone windows and a two-trade cap. PO3 only between 07:00-10:00 GMT (London) and 12:00-15:00 GMT (New York). Flat 30 minutes either side of red-folder news. Maximum two PO3 trades per day — a third attempt is revenge trading in costume.
The Sizing Maths Under a 5 Percent Daily Loss Limit
The PO3 stop is structural — it sits beyond the manipulation extreme, so the stop distance is whatever the market gave you. That makes fixed sizing dangerous: a 25-pip stop and a 45-pip stop are different trades. I size every PO3 trade from risk: 100,000 account, 5% firm daily limit, 2,500 personal daily stop.
| PO3 position-size formula |
|---|
| Risk per trade = 500 (0.5% of the account, one-fifth of my personal daily stop) |
| Stop distance = manipulation extreme to entry, plus 5 pips buffer |
| Lots = 500 / (stop pips x 10) on FX majors at 10 per pip per lot |
| Worst day = 2 trades x -1R = -1,000, which is 40% of my 2,500 personal stop |
Walk it through with the GBPUSD day above. Entry short at 1.2738 on the retest, manipulation extreme 1.2755, plus 5 pips buffer: stop at 1.2760, 22 pips. Lots = 500 / (22 x 10) = 2.27 lots. Target at the previous day low 1.2640: 98 pips x 10 x 2.27 = 2,225, about 4.4R. The maths does something psychological: the stop sits beyond the raid, so the market has shown its hand — if price retakes the manipulation extreme, the read was wrong, and 22 pips is cheap tuition. I run every setup through the max daily drawdown calculator before clicking, because two PO3 losers is a 1,000 drawdown day and I want that confirmed against my limit, not guessed.
| Trade setup | Stop (pips) | Lots at 500 risk | Target (R) | Win value |
|---|---|---|---|---|
| GBPUSD short (the worked day) | 22 | 2.27 | 4.4R | 2,225 |
| EURUSD short, tight raid | 16 | 3.12 | 3.1R | 1,550 |
| XAUUSD long, wide raid | 38 | 1.31 | 2.6R | 1,300 |
| GBPJPY short, news-day skip | – | 0 (skipped) | – | 0 |
Notice the fourth row. Gold’s raid was 38 pips wide, which is fine, but it printed twelve minutes before a red-folder USD release — Rule 5 killed the trade. Skipped trades never appear in win rates, but they are why the account survives. On gold, confirm your broker’s contract spec before sizing — 1 lot of XAUUSD moves roughly 100 per dollar per lot on most feeds, so never trust any table’s lots column blindly, including mine.
I Tested It: 10 PO3 Days on a 100k Demo
Two weeks, one demo, only PO3 setups, 500 risk per trade, rules 1-5 enforced. Full log — recount it yourself; I did twice.
| Day | Pair | Bias | Stop (pips) | Result | R |
|---|---|---|---|---|---|
| Mon | EURUSD short | Premium | 22 | Win, +1,300 | +2.6R |
| Tue | GBPUSD short | Premium | 25 | Win, +1,950 | +3.9R |
| Wed | XAUUSD long | Discount | 30 | Win, +1,100 | +2.2R |
| Thu | EURUSD short | Premium | 20 | Loss, -500 (entered mid-distribution, no retest) | -1.0R |
| Fri | GBPUSD long | Discount | 24 | Win, +1,400 | +2.8R |
| Mon | EURUSD short | Premium | 21 | Win, +950 | +1.9R |
| Tue | XAUUSD long | Discount | 32 | Loss, -500 (CPI spike faked the raid) | -1.0R |
| Wed | GBPUSD short | Premium | 23 | Win, +1,200 | +2.4R |
| Thu | EURUSD long | Discount | 19 | Loss, -500 (box never raided; forced it) | -1.0R |
| Fri | EURUSD short | Premium | 20 | Loss, -500 (distribution fizzled at +0.4R) | -1.0R |
Totals: 6 wins, 4 losses — 60%. R sum: 2.6 + 3.9 + 2.2 + 2.8 + 1.9 + 2.4 = 15.8R wins, minus 4.0R losses = +11.8R net, 5,900 at 500 risk per trade. Average winner 2.63R. The loss column is the honest part: Thursday week one, I entered mid-distribution without the retest (Rule 4 broken). Tuesday week two, a CPI spike faked the raid — news spikes are volatility, not manipulation. Thursday week two, the box never got raided and I forced it anyway. Friday week two, a textbook setup fizzled at +0.4R. Four losses, three my fault, one the market’s — about right for this strategy, and exactly why the two-trade cap and 0.5% risk exist.
When the Power of 3 Fails
PO3 models typical days, and markets are under no obligation to be typical. Five failure modes I have paid for:
| Failure mode | What it looks like | The cost |
|---|---|---|
| Trading the accumulation | Buying the box top / selling the box bottom during the quiet hours | Chop losses; the box exists to build liquidity, not to trade |
| Fading the sweep | Shorting into the manipulation spike before it closes back inside | The spike can extend 2-3x before reversing; stops get raided twice |
| No displacement | Slow drift after the raid instead of an explosive candle | Dead trade that bleeds to breakeven or a small loss |
| News-driven raid | A red-folder release prints a spike that looks exactly like manipulation | Direction is random post-news; the model does not apply |
| Reversal / trend day | Distribution runs opposite to your bias, or the day trends from the open with no box at all | Full -1R; bias was wrong, accept it and stop for the day |
The most dangerous failure is the first, because it feels like trading. The box is hypnotic — tight, clean, tradeable-looking. But every trade inside the box fights the phase of the day: you are buying retail entries while the market is still collecting the liquidity for the real move. My rule is blunt: before 07:00 GMT, the chart is a map, not a menu.
How PO3 Fits Prop Firm Drawdown and Consistency Rules
This section decides whether PO3 is a funded-account strategy or just an interesting theory. Three properties make it fit unusually well.
First, the stop is honest: it sits beyond the manipulation extreme, so the market has already given you the invalidation level — no guessing. That keeps risk per trade exact, which is what the daily loss limit demands. Second, frequency is naturally low: real PO3 days give one, sometimes two setups. That pairs cleanly with consistency rules, which punish a single dominant day — PO3 profits arrive as a stream of 2-4R distribution trades, not one lottery ticket. Third, the model keeps you out of the market most of the day: accumulation is untradeable by design and the windows are narrow, so overtrading — the actual funded-account killer — has nowhere to hide.
The fit is not perfect. Consistency rules punish long losing streaks, and PO3 can print three failed days in a row when the market trends from the open with no box. My answer: the two-trade cap, plus a three-day rule — if PO3 fails three days running, I stand down from it for the week and trade my trend pullback plan instead. And set a personal daily stop below the firm limit — mine is 2,500 against the firm’s 5,000 — because two full PO3 losers is a 1,000 day, and I never want a third attempt decided while tilted.
My Daily PO3 Routine
The strategy runs on a timetable — pinned to my monitor:
| Time (GMT) | Job | What I write down |
|---|---|---|
| 06:45 | Mark the map | Previous day high/low, Asian high/low, premium vs discount zone |
| 07:00 – 10:00 | London window | Box width, raid level, displacement candle, entry/stop/target if triggered |
| 10:00 – 12:00 | Hands off | Nothing — midday chop is not a PO3 phase |
| 12:00 – 15:00 | New York window | Second box or continuation; max 1 more trade |
| 15:00 | Journal | Phase read accuracy, rule breaks, R multiple, screenshot of the raid |
The journal column matters more than the trade column. PO3 is a reading skill before it is an entry skill — I grade myself on phase-read accuracy, not just wins. Weeks where my reads are 80% accurate are profitable even at a 50% win rate, because distribution legs pay 2-4R. Weeks where reads drop below 60% mean the regime changed, and I halve size until they recover.
Power of 3 Strategy FAQ
Is the Power of 3 the same as the Turtle Soup strategy?
Related, not the same. Turtle soup trades one event — the failed breakout. PO3 is the full daily cycle: it needs the accumulation box first and the distribution expansion after, and only trades the third act. Every turtle soup is a manipulation; not every manipulation completes a Power of 3.
Does PO3 work on timeframes below H1?
The structure exists on M15, but the funded version should stay on H1 and above. Lower timeframes multiply false raids, and each one costs a full stop. The edge comes from institutional-scale liquidity pools — daily and session extremes — which only read cleanly on higher timeframes. For the full breakdown, see Aron Groups’ Power of 3 guide.
What if there is no accumulation box?
Then there is no PO3 trade that day. Trend-from-the-open days skip the first act entirely, and forcing the model onto them is the reversal-day failure above. Flat is a position — on a funded account, often the most profitable one.
Can I trade PO3 on gold and indices?
Gold, yes — XAUUSD builds clean Asian boxes and raids them at London open; widen the box filter to about 40 dollars and confirm your contract’s pip value before sizing. Indices, cautiously — US30’s overnight profile is thinner and the raids messier. I keep PO3 to FX majors and gold.
How does PO3 handle the consistency rule?
Better than most breakout strategies: profits arrive as repeated 2-4R distribution trades, not one outsized day. The risk is failed setups clustering — hence the two-trade daily cap and the three-day stand-down rule. Size at 0.5% and the consistency math takes care of itself.
What is the minimum account size for trading PO3?
Any size — the maths scales. On a 10k account with a 5% daily limit, 0.5% risk is 50 per trade with the same stop distances, just smaller lots. What does not scale down is the discipline: the two-trade cap and the news blackout matter more on small accounts, not less.
Risk disclaimer: trading leveraged forex and CFD products on prop firm evaluations carries a substantial risk of loss, including the loss of challenge fees. Nothing here is financial advice — I write about what I do on my own accounts, and your rules, limits and risk tolerance are your own. Never risk money you cannot afford to lose.




