Evening Star Candlestick: Confirmation and Invalidation

Define the evening star candlestick formation

TradingView and thinkorswim describe the same sequence: an uptrend, a long bullish first candle, a small second candle that gaps up, and a bearish third candle closing below the first candle midpoint. Their official descriptions are linked in Sources. The second body can be bullish or bearish. Terms such as long, small and uptrend need numerical definitions in a detector. The following specification is illustrative, not a claim about either platform default. Measure the trend and reference body average before candle 1. This prevents candle 3 from changing whether an uptrend existed before the pattern.

Calculate third-candle penetration

Let O1 and C1 be the first candle open and close, and C3 the third candle close. Because candle 1 is bullish, its body B1 = C1 − O1 is positive. Penetration = (C1 − C3) / B1 × 100%. A value above 50% means C3 is below the first body midpoint. Exactly 50% does not pass the strict below-midpoint rule used here. In this hypothetical example, O1 is 100 and C1 is 108, so the midpoint is 104. Candle 3 closes at 103: (108 − 103) / 8 × 100% = 62.5%. Assume the preceding five closes are 95, 96, 97, 98 and 99, and the preceding 20-body mean is 3. Candle 1 has body 8; candle 2 has body 0.5, below the illustrative 0.9 limit. All specified tests pass. These invented prices demonstrate arithmetic, not performance.

Separate body gaps, full gaps and no-gap variants

A body gap and a gap between the entire candle ranges are different tests. In the specification above, the lower edge of the second body must exceed C1. A stricter full-range gap would require the second low to exceed the first high. The example has a body gap, but its second low equals the first high at 109, so it fails that stricter test. A no-gap adaptation can retain the advance, small middle body and third-candle penetration while dropping the body-gap requirement. For example, replace candle 2 with open 108, high 110, low 107.5 and close 108.5. Its body remains small, but it touches the first close and fails the strict gap rule. Label it as a no-gap variant when testing continuous-session markets or intraday bars. Changing the definition changes the sample; it does not establish that either version is more profitable.

• Record the symbol, timeframe, session setting and gap definition with each test.
• The illustrative rule does not require a second gap before candle 3; adding one creates another variant.
• A doji middle candle produces a doji-star variant. A smaller body alone supplies no measured win rate.

Distinguish incomplete confirmation from a failed reversal

If candle 3 closes at 106 in the example, there is no completed evening star under the chosen rule. Calling a later decline a successful evening star would change the definition after seeing the outcome. If candle 3 closes at 103, the pattern is complete. An optional follow-through rule could then require candle 4 to close below 102 without trading above 112. This is a separate one-bar test: discard the candidate if candle 4 fails, rather than waiting indefinitely for a convenient breakdown. Any signal based on candle 4 belongs at its close, not on candle 2 or candle 3. The alternatives below use the same completed pattern. A rise above 112 invalidates this illustrative bearish setup, while leaving its historical three-candle shape intact. A stop order above the high and an exit based on a closing-price invalidation are different execution rules; specify which one a strategy uses.

Test the rule at the time it becomes available

Evaluate the pattern after candle 3 closes. During an open candle its close, body and penetration can change; TradingView documents that barstate.isconfirmed becomes true on the closing update. Adding candle 4 follow-through delays the decision by another complete bar. In a default TradingView strategy, a market order created at a bar close can first fill on the next available tick, normally the next bar open. A backtest that uses the confirmation close as an automatic fill needs a different, explicit execution setting. Gaps, spread, commissions and slippage can change the realized entry or exit. Use standard OHLC charts for the baseline; synthetic chart prices can distort both the pattern and simulated fills. Before comparing settings, define the short entry, stop, target or time exit, and whether overlapping candidates are allowed. Keep those rules and the evaluation period fixed while changing one pattern threshold. Record rejected candidates separately from executed trades, and review a later holdout period before drawing conclusions. Neither the official pattern definition nor the examples here supplies a universal success rate.