How to screen for breakout candidates

Write a selection question

Write whether you want a new window-high event or candidates still approaching a high. The default high-reaching rule and the just-below-high variant answer those separate questions. A manually drawn resistance line is a chart observation rather than an automatic pattern classification.

Read field units and lookback periods

The default uses the session high, distance from the 50-session high and relative volume. It requires the close within 1% of that high and volumeRatio at least 1.5. A high of 100, close of 99 and ratio of 1.5 satisfy those boundaries; a close of 98 fails the proximity requirement.

Start from the canonical preset

Open the paired screen and choose the rule below. It includes the reference population as well as the selection conditions. Use the same expression in CleverScan so the learning example and the screen answer the same question.

securityType = "Stock" and indexMembership contains "SP500" and (high * 1.000001 >= close / (1 + hi50FromHighPct / 100) and hi50FromHighPct >= -1 and volumeRatio >= 1.5)

Compare another question before combining rules

Choose the 20-session variant for a shorter window and the breakdown variant for a downward event. Do not replace a resistance question with an all-time-high claim. Near-support research uses its own low-side conditions followed by chart review.

securityType = "Stock" and indexMembership contains "SP500" and (hi50FromHighPct >= -3 and hi50FromHighPct < -0.5 and atrPct14 < 3)

Use a low-side shortlist to research support

The shared preset below selects closes within 3% of the 20-session low with RS Rating at least 80. Proximity to a window low is a shortlist condition. Draw the actual tested support zone on each chart, inspect prior touches and define whether confirmation means an intraday touch or a closing hold. A window low alone does not establish support.

securityType = "Stock" and indexMembership contains "SP500" and (hi20FromLowPct <= 3 and rsRating >= 80)

Inspect a match and a boundary exclusion

Draw the earlier price boundary, compare the closing candle with its wick and inspect the next retest. The default can match a wick reaching the high while the close remains just below it. Keep false exits in your research log.

Change one condition at a time and compare a row just inside the threshold with one just outside it. A zero-match result means the selected population did not meet the whole rule for that session; it is not a reason to silently change the question.

Save a reproducible research definition

Save the expression through the editor’s Library workflow. Record the selected session, population and sort so a later run can be compared meaningfully. A saved scan preserves rules, while a watchlist preserves symbols; their results can change independently.