Sign inCreate account
← Strategy library

Candlestick Patterns: What the Testing Showsthe page that explains why the others are rated low

Evidence: WeakBeginnerPrice action · Reference

Doji, hammer, shooting star, morning star, three white soldiers. There are dozens of named patterns, thousands of courses teaching them, and remarkably little evidence that any of them predicts anything on its own.

This page is not an argument that candles are useless. It is an argument about what they are for: a candle is a compact description of what happened inside one bar, and that is genuinely useful for placing a stop and for timing an entry you had already decided on. It is not a forecast.

Every price-action page in this library is built that way. This one exists so the reason is written down once rather than implied ten times.

Category
Price action · Reference
Timeframe
Any
What a candle is good for
Stop placement, entry timing, seeing who won a bar
What it is not good for
Predicting the next bar on its own
Tools
Your own trade record, sorted by whether a level was present

Valid setup

PA-10 · CANDLESTICK PATTERNS — TWO IDENTICAL CANDLES the same rejection candle twice · one at nothing in particular, one at a level that had already held REFERENCE 0.85226 0.85333 0.85439 0.85546 0.85652 0.85759 0.85866 SAME CANDLE, NO LEVEL — nothing happens SAME CANDLE, AT THE LEVEL A LEVEL — the only thing that differs between the two 1 2 0.85400 ① A textbook rejection candle in open space ② The identical candle, at a level that held before ③ The pattern is the same. The context is not ④ The level is what the evidence supports ⑤ The candle supplies the stop and the moment ⑥ Record whether a level was present. That column explains more

Exact rules

Entry

  • Require a reason before the candle: a level, a session boundary, a sweep, a higher-timeframe trend. If the only thing you can say about a setup is the pattern's name, there is no setup.
  • Fix the pattern's definition in code. Most published disagreement about whether patterns work is really disagreement about what counts as one.
  • Require a size filter. Nearly every pattern on a low timeframe is inside the spread, and a pattern smaller than the spread is not an observation about the market.
  • Test the level without the candle and the candle without the level. One of those will carry almost all of the result, and it is worth knowing which.

Stop loss

  • This is where candles genuinely earn their place: the extreme of a rejection candle is a precise, non-arbitrary invalidation point.
  • A tighter stop for the same thesis raises the multiple on the same bet. That is arithmetic, and it is a real benefit that requires no predictive claim at all.
  • Risk as the underlying strategy specifies. The candle does not change how much to risk.

Take profit / exit

  • Take targets from structure — the next level, the range height — not from the pattern. No pattern implies a distance.
  • If you keep records of only one thing, record whether a level was present. That single column will explain more of your results than the pattern column will.
  • The honest summary: candles are a measuring tool, not a signal. Used that way they are useful and modest. Used as forecasts they produce the results the testing describes.

Build Candlestick Patterns: What the Testing Shows

Reading this page is free and always will be. Taking it is what costs — the complete MQL5 for the rules above, yours to keep, permanently. Five of them cost $10 and they never expire.

Sign inWhat it costs

More from Price action

Previous-Session High / Low Sweepliquidity taken, then reversalRound-Number Rejectionthe 00 and 50 figuresBreak and Retest (Acceptance)intraday level, continuation

Trading carries risk and most retail accounts lose money. Nothing on this page is financial advice or a recommendation to trade. The chart is an illustration of the rules on constructed candles — it is not a record of a trade and not a backtest.