Mastering EURUSD · Stage 2 · Setup & context

Mastering EURUSD, Stage 2: the context that splits a winner from a look-alike

A setup is not a trade. The same failed bounce is a short in one regime and a buy in another. Stage 2 of mastering EURUSD is about building the filter that tells the two apart — and proving it beats a coin flip with a blind take-or-skip test.

7 July 2026·8 min read

In Stage 1 we learned EURUSD's personality: a no-drift chop pair that grinds when policy diverges and ranges the rest of the time, whose home edges are sell-the-rally in a down-grind, mirror-long the dip in an up-grind, and fade the boundary in a range. That gives you three setup families. It does not yet tell you when to pull the trigger — because each of those setups has an A+ version that pays and a look-alike version that quietly bleeds. Telling them apart is Stage 2.

The honest gate for this stage, from the mastery ladder, is precise: your "take" set has to beat your "skip" set, tested blind. Not your win rate in isolation — the spread between the trades you chose to take and the ones you chose to pass. If the setups you skipped would have done just as well as the ones you took, your filter is adding nothing, and you are back to a coin flip wearing a costume.

Educational, not adviceEverything below is study material illustrating a method. The historical price levels are examples of how EURUSD moved, not predictions, signals, or a recommendation to trade. Past behaviour does not repeat on command. See the risk disclosure before you trade anything.

The Stage 2 reframe: a setup is a question, not an answer

The mistake that keeps Stage 1 knowledge from paying is treating a setup family as a signal. "Failed bounce into the average" is not a sell signal — it is a *candidate* that becomes a sell only when the context around it qualifies. Same chart shape, different context, opposite correct action. So the deliverable of Stage 2 is not more setups. It is, for each home edge, a short take/skip filter: the two or three context facts that have to be true, and the look-alike that fails exactly one of them.

The single fact that does most of the sorting on EURUSD you already know from Stage 1: is a policy-divergence grind in force, or not? That one classification flips almost every setup between take and skip. Keep it at the front of every read below.

Setup A — Sell the rally: take vs skip

The headline directional edge. The A+ instance and the trap look nearly identical on the chart — both are a bounce that stalls and rolls over.

The take/skip filter for sell-the-rally on EURUSD. The look-alike fails one context fact.
	Take (A+)	Skip (look-alike)
Regime	Confirmed down-grind: falling MA stack, lower highs and lower lows.	Averages flat and tangled, or turning up — no grind in force.
Location	Rally fails into the falling average from below.	Bounce has already reclaimed and closed above the average.
Macro	ECB dovish against a hawkish Fed — the divergence that drives the grind.	No divergence, or the divergence runs the other way.
Correct action	Short the failed bounce, stop above the last lower high, trail far.	Stand aside — or it is actually the mirror-long.

Key ideaThe tell that separates them is which side of the average price closes on, plus the slope of the stack. A rejection *beneath* a falling average is a sell-the-rally take. The identical-looking rejection that happens *after* price reclaimed a flattening average is a skip — you are shorting into a base, not a grind.

Setup B — Mirror-long the dip: take vs skip

The inverse edge, and home to EURUSD's most expensive look-alike: the down-grind dip that dresses up as a "buy the pullback." Getting this filter wrong is how traders catch the knife.

The take/skip filter for the mirror-long. The trap is buying a dip that is really an ongoing grind.
	Take (A+)	Skip (the knife)
Regime	Matured, multi-quarter base; averages flattened and turning up.	Averages still falling — the down-grind never ended.
Location	Dip holds a rising average and reclaims the long MA.	Dip into a falling average, no reclaim.
Proof	The up-regime has proven itself first.	You are assuming an uptrend on a no-drift pair.
Correct action	Buy the dip to the rising rail, target the measured up-leg.	Do not catch it — wait for the base and the reclaim.

The discipline here is the Stage 1 rule made operational: on a no-drift pair you never assume the uptrend, you require it to prove itself. The take demands a reclaimed long average and an up-turning stack before you commit. The skip is any dip-buy that is still waiting for that proof — no matter how oversold it looks.

Setup C — Fade the boundary: take vs skip

The default posture, and the one where the look-alike is most seductive, because fading feels safe. It is not — fading a boundary that a grind is about to blow through is the single most reliable way to bleed on EURUSD.

The take/skip filter for range-fades. The trap is fading a level during an active grind.
	Take (A+)	Skip (the bleed)
Regime	No grind: averages flat and tangled, price respecting a defined policy range.	An active policy-divergence grind is running.
Location	Price arrives stretched into a boundary it has respected before.	Price is mid-range, or at an undefined level.
The fresh high	A new high made outside an up-regime is a fade candidate.	A new high with a confirmed up-grind behind it — do not fade.
Correct action	Fade the edge toward the middle: small size, tight stop, low frequency.	Stand aside, or trade with the grind.

The fade that turns into a knifeBuying the "floor" of a range during an active down-grind is the classic EURUSD bleed. In mid-2022, price near ~1.03 looked like a range floor to fade long — but the parity grind was still running, and it broke to ~0.953. The identical action a year later, buying the defined ~1.045 floor once the averages had gone flat, held and bounced. Same setup, opposite outcome, one context fact different: was the grind still in force?

Proving it: the take/skip test, scored

A filter is worthless until the take-set beats the skip-set on outcomes you did not know in advance. So the study session built matched pairs: for each setup, one instance where the context qualifies (take) and one look-alike where exactly one context fact fails (skip), read blind, then scored on what actually happened. If the filter is real, the takes win and the skips lose.

Matched take/skip pairs across the three EURUSD setup families. Levels are the historical prices at each anchor. Illustrative of method — not signals.
Setup	Take instance	Skip look-alike	Outcome spread
Sell the rally	Aug 2014 · ~1.338 — failed bounce into a falling stack, ECB-vs-Fed divergence live.	Feb 2023 · ~1.07 — a failed-looking bounce, but averages had turned up (recovery up-grind).	Take short ground to ~1.13 (win). Shorting the skip fought the regime — it rose to ~1.124 (loss). The correct Feb-2023 trade was the mirror-long.
Mirror-long the dip	May 2017 · ~1.087 — matured base, averages turning up, long MA reclaimed.	Late 2014 · ~1.28 — a "buy the pullback" dip that looked identical, but the stack was still falling.	Take long ran to ~1.20 (win). The skip dip kept grinding to ~1.13 (knife). Same shape, opposite stack slope.
Fade the boundary	2023–24 · ~1.045 floor — flat tangled averages, a defined, repeatedly-respected policy floor.	Mid-2022 · ~1.03 "floor" — same visual, but the parity down-grind was still active.	Take long held and bounced toward ~1.12 (win). The skip long broke to ~0.953 (loss). One fact differed: grind in force or not.

In all three pairs the take won and the skip lost — and in every pair the only difference was the context classification, not the visual setup. That is exactly the Stage 2 gate: the take-set beat the skip-set, blind, and it did so *because* of the filter rather than in spite of it. A filter that only ever said "take" would have taken all six and netted close to nothing. The skips are where the edge is protected.

What the pairs teach

The look-alike is the opponent, not the setup. Every losing skip was a near-perfect visual twin of a winning take. If your rule only describes the shape, you will take both and average out to nothing. The context facts are the whole edge.
One classification does most of the sorting. "Is a policy-divergence grind in force?" flipped every single pair between take and skip. Get that one read right and the rest of the filter falls into place; get it wrong and you fade grinds and buy knives.
The skip is a trade. Choosing to pass the look-alike is an active decision that protected capital in all three pairs. A filter that only tells you when to enter is half a filter — Stage 2 is finished only when your *skip* discipline is as sharp as your entry.

How much to trust this filterThe honest verdict is medium trust, and it is worth copying the reasons. First, matched pairs prove the *direction* of the edge, but three pairs is a small sample — the filter needs many more instances, and out-of-sample ones, before you size up on it. Second, this stage tests the filter in isolation; combining it with higher-timeframe timing is Stage 3, and a filter that works standalone can still misfire once real entry timing is layered on. Name what you have not yet proven; it bounds your size.

Where Cortiq fits

Stage 2 is a filing problem as much as a trading one: for every setup family, you are accumulating the context facts that split a take from a skip, and the matched-pair evidence that each fact actually earns its place. That is precisely the kind of ledger Cortiq keeps when a session is locked to EURUSD — a durable, inspectable record of each setup's take/skip criteria, scored against outcomes and re-scored over time, so the filter is promoted on evidence rather than a feeling. You can keep this notebook by hand; the agent is built to keep it honest for you.

Key takeaways

A setup family is a candidate, not a signal — each home edge has an A+ take and a look-alike skip that only context can tell apart.
The one classification that flips almost every EURUSD setup is "is a policy-divergence grind in force?" — put it first.
Sell-the-rally requires a falling stack and a rejection beneath the average; the skip is a rejection after price reclaimed a flattening one.
The mirror-long requires a matured base and a reclaimed long average; the skip is any dip-buy in an ongoing down-grind — the knife.
The Stage 2 gate is your take-set beating your skip-set, blind — and matched pairs show the skips are where the edge is protected.

Next in the series: [Stage 3 — top-down timing](/blog/mastering-eurusd-top-down-timing), where we combine EURUSD's higher-timeframe bias with lower-timeframe entry timing — direction from above, risk from below — and score full top-down reads out-of-sample.

This article is educational and is not financial advice, a recommendation, or a signals service. Trading carries a high risk of loss. Historical examples describe how a market moved in the past and are not predictions of future behaviour. Read the risk disclosure before you trade.