Trading Outpost

Why Your Candlestick Patterns Keep Failing

This video makes no numerical claim. It states no win rate, no hit rate, no backtest, no price, no date and no benchmark, and nothing on screen states one either. What it does assert is a set of definitions and a set of market mechanics, and those are what is chased to a primary source below.

Every chart in the video is a synthetic series generated for teaching, and each one says so on screen. No chart is presented as a record of what any market did.

What a candlestick encodes

Claim. Every candle shows four things for its period: where price opened, where it closed, how high it went and how low it went. The body spans the open and the close, and the wick spans from the body out to the extreme actually traded.

Claim. A bullish candle is one that closed higher than it opened. A bearish candle is one that closed lower than it opened.

CME Group’s technical analysis course states this directly: candlestick charts take the same information as a bar chart, open, high, low and close, and represent it as a body and a wick, where “the body of the candle, the thicker middle portion, shows the open and closing prices during the time frame” and “the wick, illustrated by a thin line at the top and bottom of the body, shows the highest and lowest prices traded over the time frame”. It also gives the colour convention the video uses, that a candle closing below its open is drawn one way and one closing above its open the other.

The pattern names

Claim. Engulfing candles, pin bars, doji, hammers, shooting stars, morning stars and evening stars are the named candlestick patterns a new trader learns first, and an engulfing candle is one whose body covers the body of the candle before it.

These are the definitions set out in Steve Nison’s Japanese Candlestick Charting Techniques, the work that introduced the Japanese candlestick vocabulary to Western markets and remains the reference the names are taken from. An engulfing pattern there is defined by the real body of one candlestick completely covering the real body of the previous one; hammer, shooting star and doji are treated as single candle reversal patterns, and the morning and evening stars as the multi candle star formations.

Trend as a sequence of highs and lows

Claim. In an uptrend buyers are generally creating higher highs and higher lows, and in a downtrend sellers are generally creating lower lows and lower highs. A bullish pattern that leaves that sequence intact has not yet proved a reversal.

This is the structural definition of trend from Dow Theory, codified after Charles Dow’s death by William Peter Hamilton in The Stock Market Barometer (1922) and Robert Rhea in The Dow Theory (1932): an uptrend is a sequence of higher highs and higher lows, and a downtrend its mirror image. The HH, HL, LH and LL labelling the video uses on screen is the direct descendant of that rule.

What happens when a stop is triggered

Claim. Price pushing through an obvious high or low triggers stops, and that triggering is itself a source of the move that follows.

A stop order is an instruction to buy or sell once price reaches a specified stop price, and the SEC is explicit that the stop price is a trigger rather than a guaranteed execution price: when the stop price is reached, the stop order becomes a market order and executes at whatever the market is then offering. That is the mechanism by which a cluster of stops sitting beyond a level turns into a burst of market orders the moment the level is crossed.

Why orders sit beyond obvious levels

Claim. Many dramatic candlestick patterns form around liquidity grabs: price runs into an area where orders were likely sitting, fails to continue, and rejects. Price pushing above an obvious high triggers breakout buyers and short stops; pushing below an obvious low triggers breakout sellers and long stops.

Carol Osler’s study in the Journal of Finance is the first work built on individual currency stop-loss and take-profit order data, and it documents exactly this clustering: take-profit orders cluster strongly at round numbers, and stop-loss orders cluster strongly just beyond them, at the round numbers that are commonly used as support and resistance levels. Osler uses that clustering to explain two long-standing technical claims, that trends tend to reverse at predictable support and resistance levels, and that moves tend to be unusually rapid once those levels are crossed. Stop-loss orders in particular are found to intensify trends rather than damp them.

Sizing a stop against normal noise

Claim. A stop that sits inside normal candle noise can be taken out before the setup has a chance, because markets have noise, spreads and volatility around any level.

The standard measure of that noise is the true range and its average, introduced by J. Welles Wilder in New Concepts in Technical Trading Systems (1978). Wilder defines true range as the greatest of the current high less the current low, the absolute difference between the current high and the previous close, and the absolute difference between the current low and the previous close, and averages it to produce a per period volatility figure. The volatility system in the same book places its stop at a distance derived from that figure, widening the gap from price as volatility rises, which is the practice the video describes when it says an invalidation should sit outside normal noise rather than on the line itself.

Not checked

These are assertions the video makes that could not be chased to a primary source. They are judgements about how traders behave and about which considerations matter most, not measured results, and none of them is presented on screen as a figure.