Bear-Market Rallies: Pullback Shorts vs Breakdown Shorts

Make the setup definitions executable

For a pullback, define where the zone comes from and when it was known. A moving-average zone can be calculated at the last completed candle; a structural zone depends on a specified swing-confirmation rule. If the zone is redrawn after future candles arrive, the apparent historical entry may not have been available at the time.

For a breakdown, define the range using bars that existed before the break. State whether the trigger is a traded price below the boundary, a completed close or a retest rejection. Those are separate entry models. A rule should not switch between them retrospectively according to which would have won.

The entry and invalidation should also be specified independently. A pattern can disappear before an order is accepted, or the fill can make the original risk-reward unattractive. Describe whether the robot cancels, recalculates or continues. A setup chart should include the actual order price rather than placing an arrow on the most convenient candle.

Compare location, confirmation and opportunity cost

Waiting for confirmation reduces some early entries while giving up price. A pullback trader may obtain an attractive entry near resistance but miss a market that never rebounds. A breakdown trader may capture immediate continuation but sell after a move already becomes extended. Neither trade-off can be resolved by showing one successful screenshot.

Build a matched experiment. Use the same regime filter, risk budget, exit method and date range, then vary only the entry model. Record signals with no fill, missed moves, repeated stop-outs and successful trends. If each variant also uses a different exit and risk method, its result cannot isolate the entry choice.

Analyze the failure periods. Repeated false breaks may cluster in a range. Repeated pullback losses may cluster when the trend reverses. A useful filter should be tested on fresh data to see whether it removes such clusters without merely deleting the losing historical candles.

One setup should not create unlimited attempts

Define a setup identifier: for example, the range formation time or the confirmed trend-change bar. Specify how many entries are permitted and what event resets the count. Otherwise a robot can sell the same failed break repeatedly, generating a very different trade sample from the illustrated single attempt.

A pending order needs an expiry policy. If it remains active after the regime changes, the robot can enter a short that no longer matches the rationale. Canceling stale orders is part of execution logic and should appear in testing, not only in the manual.

Finally, annotate both winning and losing examples with entry, initial stop, exit and realized R. Use the original stop distance as the denominator, so moving a trailing stop does not make the reported reward ratio change retroactively. This produces examples that explain the rule rather than merely decorating the article.