Drawing support and resistance — boundaries, timing and tests

Draw a support or resistance area so its original boundaries and creation time can be reconstructed later. The aim is a testable observation region, not a line that must stop the market. This article supplies one explicit method and counterexamples, rather than a universal best drawing style.

A line and a zone describe different conditions

Support is considered a possible region for a decline to pause; resistance for a rise to pause. Prior lows and highs can provide anchors. ChartSchool discusses both exact levels and wider zones. A line is not inherently wrong, and a rectangle does not guarantee a better outcome.

A prior reversal is a price observation, not an inventory of resting bank orders or proof that an institution still defends the area. Record what the chart establishes and what remains a hypothesis.

Fix the data and point-selection rule

Record market, interval, feed, timezone and price field. A multi-timeframe process may provide additional context, but a weekly region is not automatically stronger than an hourly one. There is no universal prohibition on M5 levels.

For this example, a low qualifies only if it is strictly below the Lows of two completed bars on each side. Equal Lows fail. The point is available only after the second right neighbour closes; an earlier plotted marker does not make it available earlier. Define the mirrored High rule separately and relate it to market structure. Two neighbours is a chosen parameter, not a verified optimum.

Construct and freeze one supplied pair

MT5 charts can use Bid or Last. A body spans Open and Close; a wick spans the recorded Low/High range, without proving that trades occurred at every intermediate price. A Bid extreme is not automatically the Ask price for buying.

In our fictional EUR/USD model, use only the Lows of two consecutive qualifying pivots. Predeclare a maximum difference of 5 pips inclusive, with pip 0.0001. If the pair fails, create no region. Use minimum and maximum Low as boundaries without an added buffer.

These qualifying points are supplied assumptions; the table does not contain all neighbouring prices needed to detect them independently. Their difference is 4 pips, within the chosen tolerance. The full region exists after bar 11, not already at bar 3.

Horizontal boundaries and a sloping line are separate rules

The rectangle remains 1.0848–1.0852. A line connecting the two different-price anchors changes with time and therefore tests another condition. Do not replace the rectangle with a trendline after a break; record separate constructions in the technical-analysis workflow.

For this specific pair, the line is 1.0848 + (0.0004/6)(t − 3), giving 1.0854 at t = 12, already above the rectangle. This arithmetic is an additional comparison, not the chosen region rule or a native-platform algorithm. A trend-following system needs to state which condition it uses.

Contact, break and role change need separate definitions

Define range contact by High ≥ 1.0848 and Low ≤ 1.0852. Define a support break by Close strictly below 1.0848; equality does not suffice. This is a test convention, not a promise that price will stay outside.

After bar 14, treat the area as a candidate resistance. Bar 15 supplies a contact from below, but bar 16 closes above the upper boundary and ends that hypothesis under our rule. Role change is an observed scenario, not an automatic outcome of every support break.

Now consider a separate record containing only two successive Bid quotes, 1.0856 and 1.0844. Its range intersects the region, yet neither recorded quote lies inside it. A range-intersection label is different from an observed in-region price, and neither alone establishes an order fill. Missing ticks also cannot prove no such quote existed.

Quote fields and actual fills determine the cash calculation

Separately assign Bid 1.0852, Ask 1.0854, buy fill at Ask 1.0854 and later sell fill at Bid 1.0840. The difference is −14 pips. For 10,000 EUR it gives −14 USD in price P/L; subtract combined commissions 2 USD to get −16 USD, assuming no financing or conversion.

Do not subtract the spread again: those actual Bid/Ask fills already contain it. No order is promised to execute at a rectangle boundary. A confluence condition can specify context without itself supplying fills, sizing or a guaranteed independent confirmation.

Preserve original boundaries and the full sample

Do not expand an old region after seeing a lower Low. A new version may be recorded with its own creation time, while the old record remains intact. Define whether another contact requires price to leave first; several bars in one region are not automatically independent attempts.

The diagram uses five separate fictional paths. A,C,E contact 100–102 then reach at least 103; B,D do not return. Reaction among contacts is 3/3 = 100%, while contact plus reaction across all trials is 3/5 = 60%. Neither proportion is a trade win rate or a promised opportunity frequency for a real market.

Keep a reproducible record

Save source, interval, price field, selection rule, anchor confirmation times, boundaries, activation and expiry. Retain contacts, breaks, no-contact cases and failed role changes. Bar 16 ends this example’s resistance observation; choose any different expiry before testing.

If adding Fibonacci levels, record their anchors and convention separately. Matching price-derived marks does not automatically supply independent evidence. Pick a display rule for readability, then evaluate a complete hypothesis on data not used to fit it. A month, a fixed line count or consistency alone does not certify profitability.