BUILD: COURSE 3 | LESSON 4
Breakout and news strategy walkthrough
Learning objectives
Assemble a complete breakout strategy — compression setup, confirmed-break trigger, structure-based stop — and explain why false breakouts are its central cost, not a fixable bug.
Quantify how execution costs (spread widening, slippage) change around news, and why they hit breakout styles hardest.
Adapt the anatomy into a rules-based news-session variant with strict, honest constraints.
Understanding breakouts
The third strategy family trades the moment the other two hand over: mean-reversion profits while a range holds (Lesson 3), trend-following profits after a trend is established (Lesson 2) — breakout trading tries to own the transition between them.
That in-between slot explains both its appeal (catching moves at the very start, before trend systems are even allowed in) and its two structural taxes: most breakouts fail, and the moments breakouts happen are exactly when execution is at its worst. Same framing as the previous walkthroughs: teaching skeleton, illustrative parameters, zero claim of profitability as written.
Hypothesis and the false-breakout tax
Hypothesis: consolidation builds clustered orders — stops from range traders above/below the zones, pending entries from breakout anticipators, and under-positioned participants who must chase. When price escapes a compressed range with genuine participation, this order fuel produces follow-through beyond the break.
The honest part, stated up front: the majority of raw breakouts fail. Ranges spike through an edge and close back inside constantly — Lesson 3 was literally farming those failures from the other side. A naive touch-based breakout system (buy the tick through resistance) typically sees win rates in the 25–40% region, and unlike trend-following you don't automatically get big R-multiples in exchange; a failed break often reverses fast and fully. The design problem of this family is therefore a filter-and-confirmation problem: every rule below either (a) selects ranges whose breaks carry more fuel, or (b) demands evidence of follow-through before committing — and every confirmation you demand costs entry price. There is no setting where you get early entry and high confirmation; you are choosing your poison, and the spec must choose it explicitly.
The example spec (technical breakout)
- Market & timeframe: GBP/USD; H1 signals, H4 context. One instrument, longs and shorts.
- Setup — compression, not just a range: a qualified range (Lesson 3's touch rules) whose H1 ATR(14) has contracted to below 70% of its 3-month average, with range height ≤ 2 × ATR(14, H4). Volatility is cyclical: contraction loads the spring, and it's the part of the hypothesis most traders skip. Mark the zone edges.
- Entry trigger — confirmed break: an H1 candle closes beyond the zone by at least 0.25 × ATR(14, H1), with the candle's range ≥ 1.25 × ATR (an expansion bar, i.e. evidence of participation). Enter at the next candle's open. No pending stop orders straddling the range in this skeleton — the close-plus-expansion demand is the anti-false-break filter, and it deliberately costs you the first chunk of every move.
- Initial stop: the middle of the broken range. If the break was real, the market should not trade back to mid-range; if it does, the fuel argument is dead. (Not the far edge — that's a full false-break already; and not "just inside the zone" — that's inside wick territory.)
- Exits: 50% off at +1.5R; trail the remainder by 2 × ATR(14, H1) (chandelier, P2.4) to hand the survivors to trend-management. Fail-fast exit: if the first H1 candle after entry closes back inside the zone, exit immediately at market — don't wait for the stop. Cutting confirmed-failures early is where breakout systems claw back the false-break tax.
- Filters: no entries within 4 hours before high-impact news for the pair (the news variant below is a separate strategy, not an exception); session filter — breaks must occur during London or the London/NY overlap, where follow-through participation actually exists; skip if the break candle's close is already > 3 × ATR beyond the zone (you're late; chasing extended breaks is buying the top of the impulse).
- Risk: 1% per trade, max 1 position, bench at −8R. Worked sizing: range 1.2740–1.2790, H1 ATR 16 pips; break candle closes at 1.2799 (9 pips beyond, > 0.25 × ATR ✓; candle range 22 pips ≥ 1.25 × ATR ✓). Enter next open 1.2801; stop at mid-range 1.2765 = 36 pips. $10,000 × 1% = $100 → 100 ÷ (36 × $10) = 0.27 lots. First target 1.2855 (+54 pips = 1.5R).
Expectancy sketch, honestly framed: suppose confirmation gets you to 42% winners, winners averaging +1.9R after the partial-plus-trail blend, losers −0.85R average (the fail-fast exit cuts some losses short of −1R). E = (0.42 × 1.9) − (0.58 × 0.85) = 0.798 − 0.493 = +0.31R before costs — plausible-shaped numbers whose only purpose is to show you which lever matters: move the win rate to 34% and expectancy is ≈ +0.09R; the entire game is the quality of the compression filter.
The news variant — and the execution tax nobody shows in backtests
Scheduled releases (CPI, central bank decisions, NFP) are engineered breakout conditions: guaranteed timing, guaranteed fuel. A rules-based variant: qualify only events with genuine surprise potential; require a post-release M5 candle closing beyond the pre-news 2-hour range with expansion (never position before the release — that is betting on the number, a different and worse game); stop at the pre-news range midpoint; first target 1.5R, trail the rest; one trade per event, flat within the session.
Now the part that kills the idea in most hands. Around major releases, execution degrades in three specific ways, and each subtracts directly from R (P2.2's cost logic at maximum volume). Spreads widen — a pair quoted at 1 pip can quote 5–15+ pips for seconds to minutes around the print. Slippage — price gaps between ticks, and market orders fill where liquidity actually exists, not where the chart drew a line; stops fill with the same slippage, so your −1R can execute at −1.3R. Requotes/latency under load. Do the arithmetic once and it becomes unforgettable: a news trade with a 25-pip stop, entered with 4 pips of spread-plus-slippage on entry and 4 on a stopped exit, pays 8/25 = 32% of R in execution costs — before the strategy's own maths starts. Backtests on mid-price candles are silent about all of this, which is why news systems that "test" beautifully die live. Rules that survive contact: halve position size on news trades (execution variance is doubled-or-worse), demand ≥ 2R targets so the tax is amortised, and log actual vs intended fills on every demo trade — that log is the data that tells you whether the variant is viable at all.
Where this leaves you
Three walkthroughs, one repeated shape: hypothesis → regime/qualification filter → precise trigger → invalidation-based stop → exits that fit the family's profile → the honest tax each family pays (streaks; range-breaks; false-breaks and execution). None presented as profitable to copy; all presented as testable. Lesson 5 is where testing stops being a slogan: backtesting without self-deception, forward-testing on demo, and deciding — with numbers — whether a strategy earns real capital.
Key takeaways
Breakout anatomy: compression-qualified range → close-beyond-zone-with-expansion trigger → stop at mid-range → fail-fast exit on re-entry → partial at 1.5R with an ATR trail on the rest.
False breakouts are the family's structural cost: confirmation rules reduce their frequency only by giving up entry price — a chosen trade-off, not a solved problem.
Volatility contraction (ATR below ~70% of its norm) is the most-skipped and most load-bearing part of the setup: springs must be loaded before they can release.
News trading is breakout trading with the execution tax maximised: widened spreads and two-sided slippage can consume 30%+ of R on tight stops — size down, demand bigger targets, and never position before the print.
These are illustrative teaching rules with no profitability claim; the compression filter and your actual fill quality are empirical questions only demo testing can answer.