Marshall, Young & Rose (2006): Candlestick Patterns Found No Value on DJIA Stocks
The 2006 academic study used bootstrap testing on candlestick rules across DJIA stocks in the 1990s and found little evidence they worked — and backtests across 903 assets tell a similar story.
The Marshall, Young and Rose Finding
In 2006, researchers Marshall, Young and Rose published a study examining candlestick trading patterns applied to the 30 stocks in the Dow Jones Industrial Average over a period spanning roughly the 1990s into the early 2000s. Using bootstrap statistical methods — a technique that simulates many alternative price paths to guard against data-snooping bias — they concluded that candlestick rules showed no or little value for investors.
The paper attracted attention because candlestick analysis had a devoted following in retail trading communities, and rigorous academic tests of popular patterns were comparatively rare. When the answer came back negative, some dismissed it as irrelevant to modern markets. Others treated it as a final verdict on all candlestick analysis. The honest reading is narrower: specific patterns failed a rigorous test on a specific set of large-cap stocks over one particular decade.
Why DJIA Stocks Are Hard Ground for Pattern Rules
The Dow components are among the most liquid, most-analysed equities in the world. Institutional participation is high, price discovery is fast, and widely known patterns face rapid arbitrage. That headwind is not unique to candlesticks — it affects most rules-based systems applied to heavily traded large-caps.
Our backtests across 903 assets and four timeframes (1-Hour, 4-Hour, Daily, Weekly) reflect the same dynamic. For the Index asset class, the top-performing indicators are DeMarker, ROC (30), Williams %R (7), and EMA 100 Trend — momentum and oscillator tools. No candlestick formation appears among the top performers for that class.
What Candlestick Backtests Show Across the Broader Universe
Zoom out to all 903 assets and the picture grows more mixed, but not dramatically different. The Hammer pattern does reach top-performer status in the Commodity class with two wins, suggesting that pattern recognition can pick up signal in specific, less-arbitraged contexts. But for Stocks — the class closest to what Marshall, Young and Rose studied — the leading indicators are Fibonacci Pivots (22 wins), Projection Bands (16 wins), Intraday Momentum Index (16 wins), Camarilla Pivots (16 wins), and Markov Regime Detection (14 wins). Candlestick formations do not appear.
The broader numbers add essential context. Across all 382 indicators and 903 assets tested, only 26% of indicator-asset combinations beat a simple buy-and-hold strategy. High win-rate figures can mislead: some indicators post median win rates above 70% per trade, yet fewer than one in ten of their tested configurations actually beat the market on a risk-adjusted basis. The metric that matters is Sharpe ratio versus buy-and-hold — not whether more individual trades closed in profit than at a loss.
What the Bootstrap Method Was Actually Measuring
Marshall, Young and Rose's use of bootstrap statistics was methodologically appropriate. Standard significance tests applied to trading rules suffer from a well-known problem: test enough patterns and some will appear to work by chance. The bootstrap approach asks whether a pattern's historical returns are distinguishable from what you'd expect across many randomly generated price paths with the same statistical properties. For the patterns they tested, the answer was essentially that you could not rule out luck.
This does not mean candlestick patterns are universally worthless. It means they failed a specific test on a specific set of assets over a specific decade. Context — asset class, timeframe, transaction costs, market regime — shapes whether any rule works. A test that was negative on DJIA large-caps in the 1990s does not settle the question for commodities, crypto, or lower-liquidity instruments in the 2020s.
Hypothetical Backtests, Not Financial Advice
Every result on this site — 660,005 backtests across 903 assets and four timeframes — is hypothetical. The tests apply realistic cost assumptions, but past simulated performance does not predict future real-world returns. No indicator that topped a backtest is guaranteed to keep outperforming. Nothing here is financial advice, and nothing here should be read as a recommendation to trade any asset or strategy.
The Marshall, Young and Rose study is academic research published in 2006. It reflects how candlestick rules performed on heavily traded US equities in one decade, tested with the statistical tools available at the time. Treat it as one data point in a larger picture, not a final answer about technical analysis at large.
Questions, answered
What exactly did Marshall, Young and Rose conclude about candlestick patterns?
Their 2006 study applied candlestick trading rules to the 30 DJIA stocks over roughly the 1990s, using bootstrap methods to control for data-snooping. The conclusion was that the patterns showed no or little evidence of value — returns were not statistically distinguishable from what could arise by chance when the same rules were applied to randomly generated price paths with similar properties to the actual data.
Do candlestick patterns ever outperform in backtests?
Occasionally and in specific contexts. The Hammer pattern reaches top-performer status in the Commodity asset class in our tests. But candlestick formations do not appear among the leading indicators for Stocks, Index, Forex, or most other asset classes. The evidence is asset-specific and generally weaker than momentum oscillators or pivot-level systems.
Are your backtest results real trading returns?
No — all results are hypothetical simulations. They model how a rule would have performed given historical prices and assumed transaction costs. They do not capture real-world slippage beyond what is modelled, the market impact of actual order flow, or the difficulty of following a systematic rule under live conditions. These results are not financial advice and should not be treated as a performance guarantee.
What indicators actually perform well on index and stock assets?
In our backtests, the top performers for the Index class include DeMarker, ROC (30), Williams %R (7), and EMA 100 Trend. For the Stock class, Fibonacci Pivots, Projection Bands, Intraday Momentum Index, and Camarilla Pivots lead. These are asset-specific results from hypothetical tests — not recommendations, and not necessarily the indicators that will lead in future conditions.
Every figure here comes from our own out-of-sample backtests, costs included — not a course or a guess. Educational information only — not investment advice. Hypothetical backtested results; past performance does not guarantee future results. Trading involves risk of loss.
Keep reading
Get the weekly edge report
The best-performing indicator per asset, what changed this week, and the honest caveats — straight to your inbox.