A false breakout does not prove a completed reversal
Terminology varies. Fidelity calls a quick return through the breakout price a false breakout; its narrower failed-breakout definition also requires a breakout in the opposite direction. This page records the return and subsequent confirmation separately. Its custom short setup trades after local downward confirmation, before any required break of the entire old range. It is not evidence that a full opposite breakout has already happened.
- An upper wick above resistance followed by an inside close is a rejected intrabar probe. It does not meet this page's prerequisite of a completed close above resistance.
- A later price decline alone does not identify the event. The old boundary, breakout time and return deadline must all be known.
- A failed-breakout setup can itself fail when price resumes the original move. Do not infer stop hunting, trapped institutions or trader intent from candles alone.
Freeze the boundary before evaluating a return
The following upper-breakout protocol is an illustrative research specification, not a published trading edge or recommended setting. Use standard candles, one fixed symbol, timeframe and session, and sufficient history. It deliberately defines only the upper-breakout short setup; a lower-breakout long version needs its own reversed rules and test.
- Evaluate overshoot before return or confirmation. A return-bar low at or below L leaves no possible confirmation close satisfying both comparisons, so that event cannot produce an entry.
- Ignore additional breakouts while an event or trade is active. Retire a traded event once flat; begin scanning for a new b0 on the following bar. A new event gets a fresh preceding window.
- Here reentry means price returning inside the frozen range. It does not mean adding to an existing position or taking a second trade from the same event.
| Decision | Rule fixed before testing |
|---|
| Reference window | At a candidate bar b0, H is the highest high and L the lowest low of the preceding 20 completed bars, excluding b0. Require valid data and H > L. |
| Breakout close | Require the previous close to be within [L, H] and the b0 close strictly above H + 2 minimum ticks. Also require the b0 low to be at least L, excluding a bar that crosses both range edges. |
| Frozen event | Store H, L, the tick size and b0 timestamp. Do not recompute the active boundaries from later candles. A rolling high can otherwise absorb the breakout. |
| Return deadline | Inspect b1, b2 and b3 after b0. The first close strictly between L and H is the return bar r. A close exactly at H is not a return. |
| Expiry or overshoot | Without a return by the b3 close, expire the event. A close at or below L before entry retires it as an overshoot; do not chase a move that has already reached the intended target. |
| Separate confirmation | On r+1 only, require L < close < the low of r. If this fails, retire the event. A return on b3 may confirm on b4; b4 cannot be a late return. |
Separate following a breakout from fading its failure
A continuation strategy buys because the upper boundary has been exceeded. The failure strategy below waits while that continuation thesis is challenged, then sells only after its own confirmation. A losing continuation trade is not automatically a short signal. Compare these ideas as separate runs instead of reversing every stopped position.
- Do not select continuation trades using knowledge that the next three bars never returned. That would use future information at b0.
- Returning inside H is not identical to breaking below L. A test of the latter would enter later and answer a different question.
| Question | Continuation experiment | Failure experiment on this page |
|---|
| Direction after an upper break | Long, with the breakout. | Short, against the original breakout. |
| Information available at entry decision | The b0 close and earlier data. | The breakout, an eligible return and the r+1 confirmation close. |
| Return inside the range | May invalidate the long thesis under its own exit rules. | Creates a candidate; it is not yet an entry. |
| Separate trade specification | Define its own stop, target and holding limit before testing. | Use the fixed stop, L target and ten-bar time exit below. |
| What a comparison must retain | All eligible breakouts, including those that later fail. | All accepted failure signals, including those that resume upward. |
Entry, invalidation and exits for the short version
After r+1 closes with confirmation, submit one market short. With TradingView's default bar-close strategy timing, its earliest historical fill is normally the next bar open. Record the actual simulated entry E rather than assigning the confirmation close. Waiting for a completed close avoids deciding from a changing intrabar close, but delays the decision.
- A stop is an order level, not a guaranteed fill price. Gaps, spread, slippage and short-borrow charges can make the realized loss larger than S minus E.
- When a bar touches both exits, examine the emulator's assumed path and available intrabar data. Never choose whichever sequence produces the preferred result.
| Component | Explicit instruction |
|---|
| Protective stop S | Highest high from b0 through r+1, inclusive, plus 2 minimum ticks. Freeze S when confirmation completes; submit the protective buy-stop with the entry. |
| Profit target | Buy-limit at frozen L, submitted with the entry. Never move L down after price falls. |
| Position rule | One short position, no pyramiding and no second entry from this event. The arithmetic example uses 100 shares with sufficient capital and available borrow. |
| Time exit | Count the fill bar as bar one. If still open at the close of bar ten, request a market exit for the next available fill. Keep protective orders active until flat, then cancel leftovers. |
| Failure after entry | A stop fill invalidates the short. A close back above H alone does not add an exit in this baseline; introducing that exit requires a separately labeled test. |
| Entry outside (L, S) | Keep the actual fill in the record and exit at the first executable opportunity. Do not delete a gap trade, invent a fill at the old close or calculate a positive reward from an already-passed target. |
Worked sequence: three bars before the short decision
These fictional dollar prices illustrate the rule, not historical performance. Assume the preceding 20 bars give H = 100, L = 98, the previous close is 99.90, and the tick size is 0.01. The breakout threshold is 100.02. No bar before the entry closes at or below 98.
- A b2 close of exactly 100 would not qualify as a return. A first inside close at b4 would be too late. A b3 close of exactly 99.60 would fail the strict confirmation rule.
- The first return close is 99.80, but this strategy enters at the assumed 99.40. That 0.40 delay removes $40 of gross target distance for 100 shares and adds $40 of stop distance compared with the unavailable 99.80 entry. It is not a separate commission.
- If b3 instead closes at 100.10, the confirmation fails and the event is retired. A later selloff cannot retroactively restore this signal.
| Bar | High / low / close | Decision known at this close |
|---|
| b0 | 100.90 / 99.70 / 100.60 | 100.60 > 100.02. Freeze the old range; start the return clock. |
| b1 | 100.70 / 100.10 / 100.20 | Still above H. No return and no short signal. |
| b2 = r | 100.30 / 99.60 / 99.80 | First close inside (98, 100). Store return low 99.60; wait. |
| b3 = r+1 | 99.90 / 99.35 / 99.50 | 98 < 99.50 < 99.60. Confirmation passes; submit the short. |
| b4 open | Assumed fill E = 99.40 | S = 100.90 + 0.02 = 100.92. Target stays at 98. |
Calculate reward after confirmation and costs
Using the same hypothetical 100-share short, assume commission of $0.02 per share on each side, or $4 total. The target and ordinary stop branches assume exact fills at their levels. The gap branch supplies its own worse fill. Additional spread, borrow and financing costs are excluded from this arithmetic and must be added to a real test.
- Gross stop distance is 1.52 per share and target distance is 1.40. The target-to-stop distance ratio is about 0.92, before costs; a visible reversal does not imply an attractive payoff.
- For only the exact target and ordinary stop outcomes, arithmetic break-even win rate is 156 / (136 + 156), about 53.4%. This is not an estimated success rate. Gaps, time exits and variable costs invalidate that two-outcome shortcut.
- Do not deduct the illustrated 0.28 gap difference again: it is already included in the 101.20 exit fill.
| Alternative outcome | Net calculation | Result |
|---|
| Target fill at 98 | (99.40 - 98.00) × 100 - 4 | +$136 |
| Stop fill at 100.92 | (99.40 - 100.92) × 100 - 4 | -$156 |
| Gap stop fill at 101.20 | (99.40 - 101.20) × 100 - 4 | -$184 |
| Five target wins, five ordinary stop losses | 5 × 136 - 5 × 156 | -$100 |
Validate the event log before optimizing parameters
Keep one row per breakout with frozen levels, b0 time, return age, confirmation result, expiry reason and any order fills. An event that expires is different from a losing trade. Review accepted, rejected and late-return cases before comparing results; otherwise an indexing mistake can look like an edge.
- Record symbol, exchange, timeframe, session, chart type, test dates, commission units, slippage and script version. Report return frequency separately from net trade win rate.
- Freeze a development period and a later evaluation period before searching. Compare return windows of 1, 2 and 3 bars while holding the other rules fixed; treat those values as research choices, not universal settings.
- Repeat each retained setting on the untouched dates and a predeclared higher-cost case. Keep trade count, average net trade and drawdown beside net profit; do not select a setting from the later evaluation results.
- A close-confirmed strategy may miss rapid reversals and may enter after much of the move. A tick-triggered version changes the available information and must be tested separately.
Draft the frozen rules, then compare the exposed inputs
Give Pine Script AI Coding Agent the event table, strict comparisons, expiry rules, entry timing and exit definitions. Ask it to plot the frozen levels and label breakout, return and confirmation on their actual bars. Review the generated source, request syntax validation, and check the worked sequence and rejected cases against the strategy output.
- Use Strategy Optimizer to compare exposed inputs such as the return-window length over the small range fixed above. Keep the held-out dates as a separate validation run; a parameter search does not establish future profitability.
- Export supported TradingView trade data to Backtest Deep Report to inspect its available trade and risk metrics, including expectancy and drawdown. Retain the event log separately: a trade export alone cannot describe expired candidates.
- The next step is a reviewed research script and reproducible backtest. This page provides no tested preset, automatic live execution or measured return claim.
