Dead cat bounce pattern: defining a failed recovery

The rebound is a candidate until it fails

CMC Markets describes the difficulty of separating a temporary rally from a genuine recovery while it is happening. IG uses a break of the prior swing low to classify the failed rebound. Neither observation supplies a universal percentage or duration. The thresholds below are a custom teaching protocol, not validated settings or a TradingView built-in detector.

A 100 to 70 decline and a 40% retracement

Assume a review starts at bar 4 close using a previously chosen window, bars 0 through 4. Bar 0 closes at 100; the lowest low in that window is 70 on bar 4, which closes at 72. Freeze 100 and 70 now. The decline is 30 units, or 30% of 100. This supplied window isolates the classification task; a market scanner would also need a repeatable rule for selecting windows and handling overlaps.

• Evaluate cancellation before activation or confirmation on each bar. If a bar reaches 85 and closes below 69.5, cancel the candidate; its OHLC values do not establish the intrabar order of events.
• The 0.5-unit buffer is specific to these hypothetical prices. Reusing it across differently priced assets would change its relative size.
• The rebound high of 82 below recovers (82 - 70) / (100 - 70) = 40% of the decline. Its gain from the low is 12 / 70 = 17.14%. Retracement percentage and percentage return use different denominators.

Hypothetical observations after the reference window
Bar 5: low 72, high 78, close 77.5 -> rebound activates
Bar 7: high 82, close 80          -> candidate remains active
Bar 8: low 69, close 71           -> wick below 70; no close confirmation
Bar 9: low 68.5, close 69         -> first close below 69.5; confirm

Assume all omitted bars stay above 70 and below 85.
All listed bars have highs below 85.
Record confirmation at bar 9 close, never at bar 7's rebound high.

A recovery counterexample must cancel the bearish candidate

Start from the same history through bar 7, then compare independent paths. Keeping the reference prices fixed makes the classification falsifiable. Cancellation rejects this particular hypothesis; it does not prove that a new bull market has begun.

• For this review, define an early breakdown failure as any close above 70 within five bars after confirmation. The third path meets that separate outcome rule.
• A double bottom requires its own trough and intervening resistance conditions. A rebound after a selloff alone establishes neither that formation nor a dead cat bounce.
• Do not erase confirmed events after a reclaim or extend expired windows until a desired result appears. Those changes alter the sample being evaluated.

Volume and market context can inform the review

Weak rebound volume can support a cautious interpretation, but it cannot confirm future failure. CMC Markets notes that genuine recoveries can also begin on weak volume. The price-only example above deliberately has no volume gate; adding one creates a different test.

• If testing a volume filter, record its formula and baseline in advance, such as rebound-bar volume relative to the preceding 20 completed bars. Use the same symbol, feed and session convention; missing volume is not zero volume.
• Record the catalyst known at the time of the decline and any subsequent announcement. Price candles alone cannot show whether short covering, new buyers or changed fundamentals caused the rebound.
• Keep the chart timeframe and broader trend visible. A recovery on an intraday chart can remain inside a daily decline. A label without its timeframe leaves the observation incomplete.
• No success rate is established here. Estimating one needs fixed rules, all eligible candidates, an outcome horizon and a separate definition of trade profitability after costs.

A confirmed pattern is not an executable short trade

Investor.gov explains that stock short sales generally involve borrowing shares, margin obligations and potentially unlimited losses as the stock rises. A chart event cannot establish available borrow or broker permission. A brief failed recovery also says nothing about whether an investor should sell an existing holding.

• Specify entry timing, position size, stop, maximum holding period and costs before testing trades. The bar 9 close of 69 is an observation, not a guaranteed fill.
• A hypothetical short filled at 69 and covered at 86 loses 17 per share before costs. A planned exit level does not guarantee that fill when a market gaps or liquidity is insufficient.
• Include borrowing charges and applicable dividend obligations for stocks. Derivatives require their own contract, funding and liquidation assumptions; a stock chart example does not establish those terms.

Check event timing on TradingView, then draft the rules

Advance a chart one completed bar at a time and record when each condition first becomes available. TradingView documents that pivot markers drawn on earlier bars can conceal the delay before a pivot is known. This example freezes an observed window low at bar 4; it does not rely on a future-confirmed swing pivot.

• If replacing the supplied window with a pivot detector, record the pivot bar and its confirmation bar separately. Never move the eligibility date backward to the plotted pivot.
• Give Pine Script AI Coding Agent the reference-window selection, rebound threshold, cancellation priority, close buffer, expiry and early-failure rule. Request a draft that logs each state transition, then inspect the code and request syntax checking.
• Compare the generated events with all three alternative paths before evaluating trades. Syntax success alone cannot verify the financial definition or the absence of hindsight.
• Use Trading Journal tags and trade notes to distinguish early entries, confirmed entries and reclaims in your executed trades. Keep a separate candidate log for setups with no trade so that trade statistics are not mistaken for detector accuracy.