Range-Trading Strategy: Boundaries, Costs and Invalidation

Define the range before the return to it.

This is an original, unvalidated study specification. Its numerical settings are teaching choices, not optimised parameters, personalised advice or evidence of profitability. All prices below are hypothetical.

This long-only teaching example tests a return away from a lower boundary towards the middle of a range. It uses fixed levels for one observation session. The purpose is to make each decision reviewable, not to claim that these times or thresholds identify profitable ranges.

Fidelity: support, resistance and role reversal explains the support/resistance convention. A support label describes past observations; it does not prevent price breaking lower. Adding an RSI threshold would change the specification and should be evaluated as a separate variant.

A fully stated example range specification.

The middle-50% condition gives repeated visits a specific meaning: the price must leave an edge before a further visit counts. It is a reproducibility convention, not a claim that two tests make a boundary strong.

The midpoint can leave less room than expected.

Assume an eligible frozen range from 1.1000 to 1.1040. Its width is 40 pips, its midpoint 1.1020 and its bottom-10% boundary 1.1004. A completed signal bar visits 1.1002 and closes at 1.1006. Suppose the next ask is 1.1008 and the spread passes the filter.

The bid stop is 1.0998, 10 pips below entry. The midpoint is 12 pips above entry. The gross ratio is therefore 1.2:1, even though the entire range is 40 pips wide. The distance you can trade from the available entry matters more than the total size of the shape.

At a hypothetical USD 1 per pip and USD 0.70 additional round-trip commission, an exact stop fill loses USD 10.70 and an exact midpoint exit gains USD 11.30. The net ratio is about 1.056:1. Ignoring other exits and slippage, the corresponding break-even win rate is approximately 48.64%.

If the available ask were instead 1.1015, the five-pip distance to the midpoint would be less than the seventeen-pip stop distance, so the trade would be skipped. Do not use the earlier signal close as an invented fill.

The failure is a genuine change of state.

Price can break through 1.1000 and continue lower. Repeatedly buying the falling market because it used to be the range low abandons the one-entry rule and can concentrate losses. The boundary should not be moved down after each loss merely to preserve the range label.

An ordinary stop can fill beyond its trigger in a fast move. A scheduled-event filter cannot remove unscheduled news. Range-trading examples that show only gentle oscillations omit the failure that most needs to be understood.

A second awkward case is a range that remains intact but never offers a qualifying rejection or tradable entry. Record no trade. A rule does not need to produce a position every session to be a coherent study specification.

Evaluate the whole entry and exit process.

Preserve the frozen boundaries, eligibility visits and timestamps. Verify that all reference candles were complete at 09:00. Use bid/ask data to model entry and exits, and handle a bar crossing both stop and target with finer data or a conservative declared rule.

Separate the counts for eligible ranges, signals, accepted entries and completed trades. A high percentage of ranges that “held” is not the win rate of the executed method. Costs and the first available entry can change that result substantially.

Compare the process on a later untouched period, including days when ranges broke. Use the risk/reward calculator for scenario arithmetic and the strategy research framework for reporting limitations.

Work through it yourself.

Record the plan, execution assumptions, outcome and lessons from a practice trade.

Free to download without registering. Practice examples explain the method; they do not establish a profitable strategy.

Common range-trading questions.