Candlestick patterns are the oldest tool in technical analysis and the noisiest when applied without judgement. Open any chart with every pattern marked and you will see dozens of labels and no hierarchy. The way out is not a better shape detector but a better question: does this pattern appear somewhere that matters? The same idea is applied automatically in Edo Candle Context.
1. Where it forms
The same hammer in the middle of a range and at a support that price has respected three times are two different events. Check whether the pattern's extreme sits at a level built from pivots, at the latest swing or at the end of a stretched move. A pattern in no man's land is shape without context.
2. What came before
A hammer needs a decline to reverse and a shooting star needs an advance. Compare the close just before the pattern with the close a few candles earlier to see which leg price is coming from. A reversal pattern that appears after a move in its own direction is a continuation shape with the wrong name.
3. How well formed it is
Size against the average range, strength of the close and volume separate a convincing candle from one that barely meets the textbook definition. Express every definition as proportions of the body, the wicks and the ATR so the same thresholds work on an index at 5,000 points and on a currency pair at 1.10.
4. What happened next
A pattern is a hypothesis. Define confirmation beforehand, for example a close beyond the pattern in its direction within five candles, and invalidation just as clearly, a close beyond its other extreme. A pattern left Pending forever teaches nothing.
5. How often it has worked here
Count the confirmed and invalidated patterns on the symbol and timeframe you actually trade and always read the rate together with the number of events. A 70% rate over 10 events is an anecdote, a 59% rate over 1,000 is a measurement, and neither is a promise about the next candle.
