Range Trading and Mean Reversion

First Officer · C06: Strategy Construction · About 30–35 minutes

Safety rule: A research idea is not proven profitable. Predefine the stop, costs and conditions that invalidate it.

Your mission today

By the end you can design a bounded range-reversion hypothesis with a stop rather than a promise that price will always return.

Before you start

Complete C05 and your Second Officer cash-risk worksheet. Have a demo or synthetic replay. Every number below is an imaginary teaching calculation.

1. The idea in plain words

Mean reversion is the hypothesis that some price deviations may move back towards a central reference. It is not a law. A trading range must have clearly defined boundaries, a midpoint and a condition that ends the range thesis. If price breaks the floor and stays below it, the old buy-the-range idea may be invalid.

What a range is and what it is not

A range is a period during which prices repeatedly interact with a bounded area rather than sustaining a directional advance or decline. It can be studied using objectively chosen upper and lower zones. For example, a candidate classification might require at least two distinct interactions with each boundary during the previous 24 completed H1 candles and no confirmed higher-timeframe break outside the zone. This is merely a sample definition. Its count, lookback and handling of overlapping tests require research before any claim that it predicts a return to the midpoint.

A consolidation drawn after the breakout is not a valid live classification. Neither is a range that a student constantly redraws so every reversal appears to occur at support or resistance. Freeze the boundaries at a known decision time.

Loading the C06.3 teaching illustration…

Visual 01 — First learn the box. The upper area is 109.5–110.0, the lower area is 100.0–100.5, and the mathematical middle is 105.0. These are invented classroom prices, not real chart history. The floor and ceiling were drawn before the example is assessed.

In plain English: Imagine a ball bouncing between the floor and ceiling of a room. The ball may bounce again, but it can also break through. That is why a range is a working observation, not a guarantee. A zone may extend over a price interval because trades do not occur at one perfect horizontal line. If the bottom zone is 100.0–100.5 and the upper zone is 109.5–110.0, a wick to 100.3 interacts with the lower zone; a candle that closes at 104.0 does not. Define whether entry requires touch, rejection, or reclaim. Those are different experiments and must not be mixed after results are observed.

2. Worked example: loss before gain

Take an imaginary range from 100 to 110 with midpoint 105. A proposed long at Ask 100.9 has stop Bid 99.6 and potential exit Bid 105.0. Downside: 1.3 points. Proposed favourable distance: 4.1 points, so gross reward/risk is 4.1÷1.3 ≈ 3.15R, before spread and fees. At $10 per point, planned loss would be −$13. Even a 3.15R target can be missed.

Original Black Box Journey diagram; not a verified live strategy.

Mean reversion is a hypothesis about distribution, not fate

Mean reversion proposes that an extreme observation is more likely than a suitable baseline to return toward some estimated centre over a stated horizon. That idea does not mean price must eventually return. A price can trend far beyond its previous range, never revisit the original entry, or do so only after exposure has become unaffordable. A student's responsibility is to define the relevant market regime, centre estimator, event horizon and invalidation. “It will come back one day” is not a trading strategy because it contains no bounded loss and no falsifiable time or price condition.

The midpoint of a 100–110 range is 105.0. It may be a convenient illustration, but it is not automatically the market's “fair value.” Other centre estimates include a rolling median, a moving average or a session VWAP calculated from available data. They can disagree substantially. A learner should choose one, state how it is computed, and avoid switching among them whenever one makes the historical trade look better.

Bounded range rotation, synthetic exampleThe archival image is not included in this student edition.View the released teaching illustrations.{width=6.55in}

Figure C06.3A. Interaction with a lower price zone can motivate a return-to-centre hypothesis; the midpoint is a proposed target, not a destination price is obliged to visit.

3. Make the rule reproducible

Strategy card: fixed range start/end, confirmation that the boundaries existed, entry near the edge, exact stop beyond invalidation, chosen exit, max time in setup, gap handling, no-add-to-loss rule and event exclusions. Never use unlimited averaging layers to avoid recording a loss. Include worst open exposure, not only winning closed baskets.

Construct a bounded paper model

A hypothetical paper plan might specify: H1 range 100.0–110.0; lower acceptance zone 100.0–100.5; long eligibility only after a candle touches that zone and a completed H1 candle closes back above 100.5; enter at the next available Ask; hard invalidation at a specified price beneath the low of the rejection sequence; initial exit hypothesis at 105.0; skip if higher-timeframe price is closing outside the range. This design needs a risk budget, instrument specifications and a spread allowance before it could be responsibly simulated. The example is intentionally simple so another student can challenge it.

Suppose the next hypothetical Ask is 100.9 and the planned stop fill is Bid 99.6. Price risk is 1.3 units per underlying unit. A midpoint exit at Bid 105.0 would offer 4.1 units of potential gross favourable movement, or about 3.15 times the initial price risk before commission, slippage and funding. That attractive quotient says nothing about the probability of success.

Loading the C06.3 teaching illustration…

Visual 02 — Watch the order of events. Price touches the floor zone at 100.3, then an H1 candle finishes above 100.5. The next illustrative Ask is 100.9. A proposed stop is at Bid 99.6 and target is Bid 105.0. The 3.15R shown is potential gross reward relative to price risk, not an observed profit or probability.

Check yourself: The 100.3 wick is a touch, but not the entire trigger. Our stated model still needs a completed close above 100.5, a valid next quote and acceptable execution costs. If the closed reclaim never occurs, there is no trade. If the model wins only rarely, the expected result may be negative. If costs consume much of a small range target, even a high raw win rate may not survive net execution.

4. Count failures and skipped opportunities

What destroys the range thesis

The most important training exercise is to recognize that a range can end. A close beneath the old floor according to a predeclared break rule can change the regime classification.

Loading the C06.3 teaching illustration…

Visual 03 — Three decisions to stay flat: (1) the range floor is broken under the frozen rule; (2) price is near the middle, not the required lower-zone entry; (3) additional grid positions would keep increasing exposure during a continuing decline. These are separate fabricated scenarios.

The principle to remember: A failed idea must remain a failed idea in your journal. Do not redraw a floor or add unlimited positions just to avoid writing down a losing hypothesis. A scheduled macro announcement can shift volatility and invalidate previously useful assumptions about spread and gap risk. A new sustained sequence of lower lows can represent the start of a downtrend rather than a generous discount at support. In every case, add the failed range observation to the research log rather than redraw the range after seeing the outcome.

Range breakdown and hard invalidation, synthetic exampleThe archival image is not included in this student edition.View the released teaching illustrations.{width=6.55in}

Figure C06.3B. Once price leaves a bounded range, adding position layers without a hard limit converts a testable model into potentially unbounded exposure.

Averaging down can disguise risk by moving an average entry closer to the market while increasing position size and exposure. A grid with unlimited layers and no defined aggregate loss cannot be compared honestly with a single-entry, hard-stop model. Its closed-trade win rate may look excellent while a large floating loss remains unresolved. A student must report peak gross exposure, margin requirement, worst unrealized loss and capital needed during the path, not just the eventual basket close. For the Academy's basic strategy lessons, bounded one-entry hypotheses are preferable because they allow clean measurement and unambiguous invalidation.

A complying loss remains in the study ledger. Never add a rule solely because one past losing trade was inconvenient.

5. Practise with the future hidden

Practice laboratory

Choose one market and one fixed H1 lookback. Find ten consecutive range candidates chronologically. Record the boundaries as they existed before any test, the exact point a close/reclaim would have qualified, the spread at the intended execution time if available, the invalidation and whether a higher-timeframe regime change made the idea ineligible. Do not discard cases that trend away. Write a comparison of midpoint exit versus nearer conservative exit using identical entry eligibility, including the trade opportunities that never triggered. Submit a one-page range policy, a ten-row annotated ledger and a brief failure analysis.

Glossary: mean reversion = a proposed tendency to return toward a defined statistic; centre estimator = the algorithm used to calculate that statistic; reclaim = a rule-defined move back inside a zone; grid layer = an additional exposure increment at another price; floating drawdown = unrealized loss on open exposure.

Common mistake

Rewriting an entry, stop or data window after seeing the outcome destroys the original test. Keep the first version and start a new, separately evaluated variant instead.

Plain-English glossary

• Range: Area of price movement between defined boundaries.
• Midpoint: Halfway between the declared edges.
• Mean reversion: Hypothesis of return towards a reference.
• R: Result divided by initial planned risk.
• Open drawdown: Loss on positions that have not yet been closed.

Try it yourself

Create a strategy card containing market, timeframes, entry, stop, exit and skip rules. Carry out the original independent exercise above in closed-bar replay. Record at least one rules-compliant loss and one rejected candidate.

What you can now do

• Design a bounded range-reversion hypothesis with a stop rather than a promise that price will always return.
• State how the idea could fail and what it would cost.
• Keep a reproducible, non-cherry-picked evidence log.

Next lesson

C06.4 · Turning an Idea into Exact Rules.