Continuation candlestick patterns: trend before pattern

Establish the trend before the first pattern candle

A bullish shape in a sideways market is not evidence of an uptrend continuation. Define the trend independently of the pattern and record it before the setup begins. TradingView offers trend-detection choices for its three-methods indicators, including price relative to SMA50, alignment with SMA50 and SMA200, or no trend detection. A shape detected with the filter disabled still needs context.

• Illustrative research filter: before the first pattern candle, require the preceding close above its 20-period SMA and that SMA above its value five bars earlier. Reverse both comparisons for a downtrend.
• Freeze that initial classification for the setup. Using the eventual breakout to decide whether the earlier market was trending introduces hindsight.
• Label setups that meet neither direction as neutral. This is a mechanical classification, not proof that the market has no trend under another definition.
• Use the same symbol, timeframe and session throughout the comparison. A daily uptrend and an intraday downtrend answer different questions.

Separate consolidation, resumption and failure

Consolidation is the candidate stage. Resumption requires the chosen price condition to complete. Failure can happen before a signal exists or after a confirmed signal has already produced a trade. Keep those outcomes separate: a rejected setup and a losing trade should not share one count.

Compare the shapes without giving every pause a direction

TradingView describes rising three methods as a long bullish candle, three smaller bearish bodies within the first range, and a long bullish fifth candle closing above the first close. Falling three methods reverses the colors and finishes below the first close. Requiring every middle wick inside the initial range is a stricter containment variant; record that choice instead of switching definitions between examples.

Example: a completed shape can still be below resistance

In this hypothetical rising-three-methods sequence, candle 1 closes at 108 and reaches 110. Candle 5 closes at 109. It satisfies the close-above-close condition, but a stricter entry rule requiring a close above 110 has not fired. An intrabar high of 111 does not satisfy that closing-price rule. A script should display these as separate conditions.

• If candle 5 instead closes at 107, this five-candle sequence has not completed the rising-three-methods condition.
• Under a strict full-range containment rule, a middle low below 99 rejects the setup before confirmation.
• If a later candle closes below 99 after confirmation, label it a failure of the continuation hypothesis under this example rule. This does not retrospectively erase the original signal.
• A break below 99 does not by itself establish a lasting downtrend. Keep trend reversal and setup failure as different observations.

Compare patterns with one trend filter and retain failed samples

A useful experiment changes the pattern condition while holding the surrounding rules fixed. The following protocol is an illustrative research design, not a claim that these settings are profitable. Choose numerical body-size thresholds before running it; terms such as long and small are insufficient for reproducible detection.

• Separate bullish and bearish candidates, then split each by the frozen pre-pattern trend classification: aligned, neutral or opposite. Retain all groups in the report.
• For each candidate, record its first candle, final confirmation time, containment result and whether it expired. Exactly five candles is one three-methods definition; a longer pause belongs to a separately named variant.
• For a fixed ten-bar outcome window after confirmation, classify closes relative to the full completed pattern range: first close beyond the trend-side boundary is continuation, first close beyond the opposite boundary is failure, and neither is unresolved.
• Record both first-boundary outcome and the tenth-bar close. A setup can resume first and reverse later; reporting only the initial move hides that path.
• For a trading comparison, keep position sizing, entry timing, stop, target, time exit and costs identical across pattern variants. Report trade count, average net result and drawdown alongside win rate.
• Choose settings on one period and evaluate on a separate later period. Compare nearby thresholds and an unfiltered baseline before attributing any result to the pattern.

Match the TradingView signal time to the trade test

An unfinished candle can cross a boundary and close back inside it. TradingView documents that realtime highs, lows and closes change until the bar completes. For this close-confirmed method, evaluate and alert at the final candle close. Waiting for the close addresses that particular source of signal changes; it does not guarantee that every script avoids repainting.

• Keep recognition time separate from fill time. With default strategy calculation and order-processing settings, a market order created at a bar close normally fills at the next bar open.
• Include commission and slippage. A gap between confirmation and entry can change the distance to the stop.
• Use standard OHLC candles for the baseline. Synthetic chart prices can change both the detected formation and simulated fills.
• Do not assume a volume filter improves results. Test it separately, document the feed and compare the same sample and execution rules.