3 White Soldiers Candlestick Pattern: Strength vs Exhaustion

Define the bodies, opens, and wicks before counting signals

TradingView describes three consecutive long bullish bodies, opening within the preceding body and closing near the high. In a detector, words such as long and near need explicit thresholds. The following is an illustrative specification, not a claim about TradingView's internal detector or an optimal trading rule. Number the candles from oldest to newest. Require every close to exceed its open and require close 1 < close 2 < close 3. For candles 2 and 3, require the open to be strictly between the prior open and prior close. If an open equals the previous close, this strict version rejects it; an inclusive version can be tested separately for continuously traded data.

• Body size: require each real body to exceed the mean absolute body size of the 20 bars before candle 1. Freeze that baseline for all three candles.
• Upper wick: high minus close, divided by high minus low. An illustrative maximum is 15% on each bullish candle.
• Lower wick: open minus low. An optional maximum of 20% of the range makes the short-shadow requirement explicit.
• Reject zero-range candles before dividing. Body lengths need not increase monotonically.
• For a reversal study, define the decline using only bars before candle 1; do not infer it from the three-bar diagram.

An illustrative pattern with rising closes and equal bodies

Assume a decline preceded the three candles and the frozen mean body is 2 price units. Each body below measures 3 units. The second open, 102, is inside 100 to 103; the third open, 104, is inside 102 to 105. Closes rise from 103 to 105 to 107. Each upper wick is 0.3 within a 3.6 range, or about 8.3%, and each lower wick has the same proportion. This passes the illustrative geometry rules even though the bodies do not grow. Change candle 3 to a high of 110 while keeping its other prices: its upper wick becomes 3 / 6.3, about 47.6%, and it fails the 15% limit. A green candle alone is insufficient.

Trend initiation and overextension can share the same shape

The location of the pattern changes the question. After a decline, it can be studied as a candidate reversal. After an existing rally, the same geometry does not demonstrate that a new uptrend has begun. Neither location proves that buying pressure will continue. Measure the distance already traveled and the room to a resistance level identified before entry. In the hypothetical example, a fill at 107 with an exit trigger at 99.7 has a price risk of 7.3 units. If the next resistance is 109, the distance to it is only 2 units, about 0.27 times that risk before costs. The formation still passes; a study requiring at least twice the initial risk in room would skip it. Moving the stop closer just to improve that ratio changes the exit hypothesis.

Separate confirmation, entry, and failure

The final body and wick are known only when candle 3 closes. TradingView's bar-state documentation distinguishes this closing update from intrabar updates. A marker based on an unfinished third candle can disappear before the close. Closed-bar evaluation addresses that specific change; it is not a blanket guarantee against every form of repainting. An order after confirmation, a breakout above the third high, and a pullback order are different entry rules. A breakout can fill after a gap at a worse price. A pullback order may never fill. Neither the third close nor the midpoint is automatically an executable entry price. For a stated test, a completed close below the three-bar low could invalidate the reversal thesis. A stop triggered below candle 3's low is a tighter alternative and will produce different trades. Choose one in advance. A later decline does not erase a historically valid pattern; it records a failed trading hypothesis.

Test the structure and the extension filter separately

Start with a detector that records every completed pattern under one definition. For an illustrative extension measure, divide the third close minus the first open by the 14-period ATR measured on the bar immediately before candle 1. Require a positive ATR. Compare an unfiltered run with a run that rejects values above 2, keeping all other settings fixed. The cutoff is an experiment, not a recommended setting. Use the same symbol, timeframe, sessions, date range, entries, exits, and position sizing in both runs. Record how many signals the filter removes as well as net results, drawdown, and trade count. Keep a separate date range for evaluation after choosing the cutoff. A smaller sample or a better in-sample result alone does not show that the filter helps. TradingView documents synthetic-price problems on non-standard charts and provides commission and slippage settings. Use ordinary price candles for this test, state the fill assumptions, and include costs. This page provides a rule specification, not a live market scanner or a measured performance result.