BUILD: COURSE 3 | LESSON 3
Range and mean-reversion strategy walkthrough
Learning objectives
Assemble a complete range/mean-reversion strategy: range-qualification filter, band-extreme setup, rejection trigger, and exits at the mean or opposite extreme.
Contrast its performance profile with trend-following — high win rate, small winners, rare large losers — and identify the tail-loss risk that profile hides.
Define in advance the single most important rule of the family: what makes you stop trading a range.
Mean-reversion explained
Lesson 2's strategy made money by believing moves; this one makes money by fading them.
Mean-reversion is trend-following's mirror image in almost every statistic, which is why studying both teaches you more than either alone: the same chart event — price pushing hard at a level — is a signal in opposite directions depending on regime.
As always: teaching skeleton, illustrative parameters, no claim of profitability as written. Understand the anatomy, then test your version on demo.
The hypothesis and the mirror-image profile
Hypothesis: when a market lacks a directional driver, it oscillates inside a band while inventory changes hands; pushes to the band's edges are increasingly met by profit-taking and fading, so price at an extreme of a qualified range tends to rotate back toward the middle.
The profile this implies is the inverse of Lesson 2's, and you should internalise both columns:
| Trend-pullback | Range mean-reversion | |
|---|---|---|
| Win rate | ~30–45% | ~55–70% |
| Avg win / avg loss | 2–4 : 1 | ~0.7–1.2 : 1 |
| Equity curve | long bleeds, sharp climbs | smooth grind, occasional cliffs |
| Killer risk | trader abandons it mid-streak | one unmanaged breakout loss erases weeks |
Run the honest numbers (P2.1). Suppose 62% winners at +0.9R, 38% losers at −1R: expectancy = (0.62 × 0.9) − (0.38 × 1) = 0.558 − 0.380 = +0.18R — a real but thin edge, and thin edges are where costs bite hardest: if spread and swap cost 0.08R per trade, more than 40% of the edge is gone (P2.2). The cliff risk is the deeper problem: the strategy is structurally short volatility — it wins small and often until the range breaks, and the trade that breaks it is the same setup that paid you the last nine times. Everything distinctive in this spec exists to manage that one fact.
The example spec
- Market & timeframe: one non-USD cross with ranging habits — the walkthrough uses EUR/GBP; H1 for setup and trigger, H4 for range qualification. (A cross also keeps you outside your USD correlation group if you're running Lesson 2's strategy in parallel — P2.5.)
- Range qualification (the regime filter, and the heart of the system): at least 3 touches of a horizontal resistance zone and 3 of a support zone over ≥ 48 H1 bars, zones each no thicker than 0.25 × ATR(14, H4); range height ≥ 2.5 × ATR(14, H1) (so the rotation is worth trading after costs); H4 ADX(14) below 20 (no directional pressure). All four or no trades.
- Setup: price enters the support zone (longs; mirror at resistance).
- Entry trigger: an H1 candle that trades into the zone but closes back inside the range, above the zone's upper edge — a rejection close. No touch-and-hope limit orders in the skeleton: the close-back-inside requirement is your evidence the fade is actually being bought.
- Initial stop: 1 × ATR(14, H1) beyond the far edge of the support zone. Tight stops just outside a range edge sit exactly where stop-hunting wicks go; the ATR buffer prices that in.
- Exits: first target — the range midpoint, close 50%; second target — the opposite zone edge. Move the stop to breakeven only after the midpoint target fills (a deliberate P2.4 choice: in ranges, "retrace to entry" is common, so an instant breakeven stop would shred the win rate the system depends on). Time exit: close anything still open after 24 H1 bars.
- The kill-switch (non-negotiable): if any H1 candle closes beyond the far edge of either zone, the range is dead — exit open trades at market, cancel everything, and the instrument is untradeable by this strategy until a new range qualifies from scratch. No "it might come back". This single rule is what separates surviving mean-reversion traders from stories about them.
- Filters: calendar — no entries within 4 hours of high-impact events for either currency, and no positions held through them (the kill-switch scenario is usually a news candle); session — signals only during London hours for EUR/GBP.
- Risk: 1% per trade, max 1 position per range, bench the strategy at −8R from its equity high.
Walkthrough: nine small wins and the cliff
A textbook winner. EUR/GBP qualifies: support zone 0.8560–0.8568, resistance 0.8618–0.8625, four touches each over three days, range height 55 pips ≈ 3.2 × H1 ATR (17 pips), ADX 14. Price dips to 0.8563, and the next H1 candle closes at 0.8572 — back inside, above the zone. Long 0.8572; stop 1 × ATR beyond the zone's far edge: 0.8560 − 0.0017 = 0.8543 (29 pips). $10,000 account, 1% = $100 → 100 ÷ (29 × $10) ≈ 0.34 lots. Price rotates: 50% closed at the midpoint 0.8592 (+20 pips), stop to breakeven, remainder exits at 0.8616 near the far zone (+44 pips). Blended ≈ +1.1R. Repeat something like this for as long as the range holds — often days, sometimes weeks.
The cliff, and why the rules already handled it. Two weeks later, UK CPI prints far above forecast. Price is mid-range beforehand — the calendar filter means you have no position — and the release drives an H1 close 40 pips above the resistance zone. Kill-switch: the range is dead, no fades, strategy stands down on EUR/GBP. The counterfactual trader — short at resistance "because it worked all week", holding through news, stop widened once — donates a month of grinding to a single candle, then average-downs into the new trend (P2.4's cardinal sin) because the fade "must" work eventually. Note the deep symmetry: this breakout candle is precisely the raw material of Lesson 4's strategy. Regimes don't just end; they hand over.
What we don't know, and your demo assignment
Same honesty as Lesson 2: we make no claim about this spec's expectancy on EUR/GBP this year. What the public record supports is the shape: short-volatility, high-win-rate systems with tail risk concentrated in regime breaks, and costs mattering more than they do for trend systems because the per-trade edge is thinner. Whether your qualification thresholds find real ranges on your instrument, and whether the rotation clears your costs, is an empirical question — and mean-reversion is particularly seductive on demo because the frequent small wins feel like skill quickly. Fifty trades before you believe anything (P2.1). The task below starts that count, and Lesson 5 turns it into a verdict.
Key takeaways
Mean-reversion anatomy: strict range qualification (touches, height vs ATR, low ADX) → zone-entry setup → rejection-close trigger → stop beyond the far zone edge → targets at midpoint and opposite edge.
The profile mirrors trend-following: 55–70% win rates and smooth grinds, paid for with rare large losses when the range breaks — the system is structurally short volatility.
Thin per-trade edges make costs decisive: an 0.18R raw edge can lose half its value to spread; range height ≥ 2.5 × ATR exists exactly for this.
The kill-switch (H1 close beyond a zone = range dead, flat, stand down) is the load-bearing rule; holding fades through news or "waiting for it to come back" is how this family destroys accounts.
These parameters are illustrative teaching material, not a system with claimed profitability — qualify, trade and measure your version on demo before any live consideration.