This pattern measures well, and most of why it measures well is arithmetic rather than manipulation. Separating those two is the whole lesson — and what is left over is still worth knowing.
The power of three describes a trading day as a quiet early range, a probe out of it that takes stops, and then a one-way move for the rest of the session.
The idea splits a trading day into three phases. First accumulation, a quiet range near the open. Then manipulation, a push out of that range that takes stops and looks like a breakout. Then distribution, the real move, running the other way for the rest of the day.
Shortened to AMD, it is one of the most repeated ideas in smart-money teaching. Its testable core is simple: on a day that closes up, the low should form early.
A shopkeeper wants to buy stock cheaply. He spends the morning quietly buying whatever is offered, briefly drops his bid to shake out nervous sellers, and once he has enough he stops holding the price down and lets it rise all afternoon.
That is the story AMD tells about a trading day. It is a good story. Whether the chart shows a deliberate sequence or simply what any rising day looks like is exactly what has to be checked.
Every UTC day split at hour eight. For days that closed above their open, the question is whether the day's low formed in that first third.
| Instrument | All days: low early | Up days: low early | All days: high early | Down days: high early | Days |
|---|---|---|---|---|---|
| XAUUSD | 37.4% | 63.2% | 32.9% | 61.1% | 1,445 |
| GBPUSD | 31.5% | 59.6% | 32.4% | 57.6% | 1,411 |
| EURUSD | 31.7% | 60.8% | 32.6% | 58.4% | 1,420 |
| US30 | 24.9% | 44.7% | 19.3% | 39.3% | 1,040 |
| NAS100 | 24.6% | 41.1% | 18.5% | 36.8% | 1,059 |
What was counted: Each UTC day is split at hour eight. For days that closed above their open, the question is whether the day's low formed in that first third; for days that closed below, whether the high did. The chance baseline is 33.3%, because eight hours is a third of the day.
The "all days" columns are the honest baseline — they show how often the extreme lands early regardless of which way the day went.
On gold, 63.2% of up days had their low in the first third of the day, against 37.4% of days overall. Down days mirror it. Every instrument shows the same direction, though the indices are weaker — the Dow manages 44.7% against a 24.9% baseline.
A large part of that gap is guaranteed by arithmetic, not by any manipulation. A day that closes up must, by definition, have travelled upward on balance — and a day that travels upward tends to make its low near the start. You would find the same pattern in a market with no participants at all, just a rising line.
So the finding is not "someone engineered a shakeout". The finding is narrower and still worth having: on days that end up, the low is usually already in before 08:00 UTC. That is a fact about timing you can act on. The causal story about deliberate manipulation is not something this measurement can support, and no measurement of price alone could.
Ana wants to buy gold today but thinks it is too expensive. It is 15:00 UTC, gold is up eighteen dollars from the open, and she decides to wait for a pullback to the morning lows.
What the data says about that plan. If today closes up — which it currently looks like doing — there is roughly a 63.2% chance the low is already in and it happened before 08:00. She is waiting for something that, on most up days, does not come back.
The mistake underneath. She is treating the morning low as a level that will be revisited. On up days it usually is not. Waiting for it means either missing the day or eventually buying much higher out of frustration, which is worse than either alternative.
The better framing. Either she finds a setup at current prices with an invalidation close by — a demand zone or an order block formed during the move up — or she accepts she has missed this one and waits for tomorrow's early range.
The part to be careful about. None of this tells her the day will close up. It says that if it does, the low is behind her. Reading a conditional statement as a forecast is the most common way this concept gets misused.
Gold showed a strong pattern; the Dow a much weaker one. Replay a few months, record the hour of each day's high and low along with whether the day closed up, and build the table yourself. The gap between your two columns is the only version of this that applies to you.
Test the daily shape in the backtester →Run the drill in the free backtester → Free, no sign-in to begin.
What is the power of three in trading?
A model that splits the day into three phases: accumulation, a quiet early range; manipulation, a push out of it that takes stops; and distribution, the real move that runs for the rest of the session. It is often shortened to AMD.
Does the power of three actually happen?
The measurable part does. On gold, 63.2% of days that closed up had their low in the first third of the day, against 37.4% of days overall. Every instrument points the same way, though the Dow is much weaker at 44.7% against a 24.9% baseline.
Is that not just what an up day looks like?
Largely, yes — and this is the honest caveat. A day that closes up has, by definition, travelled upward on balance, and a rising day tends to make its low early. Part of the gap is guaranteed by arithmetic rather than by anyone manipulating anything. What survives is a fact about timing, not a claim about intent.
Can I trade the three phases?
Not directly, because you cannot label a phase until it has finished. The usable version is narrower: mark the first eight hours as a range, watch which side breaks, and remember that on a day which ends up, the low is usually already behind you by 08:00 UTC.
Does it prove the market is manipulated?
No, and no measurement of price alone could. The pattern is consistent with deliberate accumulation and equally consistent with ordinary trending price. This page reports what the candles did and stops there.
What timezone should I use?
UTC, which is what these measurements use. The day boundary is the entire basis of the pattern, so changing it changes the answer — and a chart set to your local time has a different boundary from the one measured here.
Record the hour of each day's high and low along with whether the day closed up, then compare the conditional figure against the all-days baseline. The gap between those two columns is the only part of this that is a finding.
Open the free backtester →Beginner exploration
Open each answer for a plain-language way to read Power of Three, And What It Really Shows, test it carefully and decide what to explore next.
This page focuses on “Power of Three, And What It Really Shows”.Accumulation, manipulation, distribution — tested across 6,375 days. On gold, 63.2% of up days had their low in the first third against a 37.4% baseline. Real, and less impressive than it looks.For “Power of Three, And What It Really Shows”, a beginner should identify what the learning guide measures, assumes or teaches before acting on its conclusion.Treat this page's account of “Power of Three, And What It Really Shows” as a learning reference rather than a prediction, signal or promise of future performance.
For “Power of Three, And What It Really Shows”, translate the idea into a definition you could apply the same way on two different charts.While exploring “Power of Three, And What It Really Shows”, work through one example slowly and record which inputs or observations determined the result.Keep your “Power of Three, And What It Really Shows” record honest: list the limitation or counterexample before using the concept in a trading plan.Before leaving “Power of Three, And What It Really Shows”, practise the definition on unseen history and review consistency before judging performance.
Turn one idea from “Power of Three, And What It Really Shows” into a rule with explicit inputs, dates, costs and pass-or-fail conditions.Ask AI to expose missing assumptions in that “Power of Three, And What It Really Shows” test, not to guess the next market move.Use the FXAbsolute AI Backtesting Lab to inspect calculations connected to “Power of Three, And What It Really Shows” and the assumptions behind them.Reproduce any important “Power of Three, And What It Really Shows” result and reserve unseen data before deciding that an apparent pattern is useful.
Continue your exploration of Power of Three, And What It Really Shows with the beginner AI prompt guide, or inspect public calculations in the AI Backtesting Lab.