StreetAlpha
Free research

Do popular trading strategies work? We backtested them.

Golden crosses, Turtle breakouts, the Minervini trend template, Weinstein stages, earnings gaps and more, each tested on thousands of US stocks since 2018, plus crypto strategies on Bitcoin and Ethereum since 2013. Most did no better than picking a stock at random. Here is every result, including the ones that failed.

Stocks tested3,052
History2018 on
Strategies38
Positive result4

How we test

Against random picks. Every signal is compared with random stocks bought on the same day and held the same way.

Costs included. A round trip trading cost is taken off every trade.

Rules set first. Rules are written down before we look, then checked once on recent months held back.

Real stocks. Liquid US stocks above $10, including companies that later stopped trading.

Every result

Positive Did better than the average stockMixed Worked in some periods onlyNo edge Same as random picksDid worse Did worse than the average stock
StrategyCategoryVerdictKey resultDetails
Momentum Leaders (12 month strength)MomentumPositive+1.07% a month over the average stock; volatile, not yet conclusive
Golden crossChart setupsNo edge0.40% behind the average stock over 60 days; flipped in the last 18 monthsRead the test
Death cross as a sell signalChart setupsNo edgeStocks did 0.67% better afterwards, not worseRead the test
Turtle trading, 20 day breakoutChart setupsNo edge0.26% behind per trade, 40% winnersRead the test
Turtle trading, 55 day breakoutChart setupsNo edge0.46% behind per trade, 40% winnersRead the test
Weinstein Stage 2 breakoutChart setupsDid worse4.1% behind SPY over 26 weeksRead the test
Minervini trend templateChart setupsNo edge+0.09% over 20 days, the same as random picksRead the test
Tight range breakout (VCP style)Chart setupsNo edge0.10% behind over 10 daysRead the test
52 week high breakout on volumeChart setupsNo edge0.05% behind over 20 daysRead the test
RSI(2) pullbackChart setupsMixed66% winners but 0.04% behind; better in the last 18 monthsRead the test
Bounce off the 50 day averageChart setupsNo edge0.05% behind over 10 daysRead the test
Pullback and curlChart setupsNo edge0.09% behind over 10 daysRead the test
Volume shock breakoutChart setupsNo edge0.31% behind over 10 days; flipped in the last 18 months
Power earnings gap, mid capsEarningsNo edge0.26% behind over 20 daysRead the test
Power earnings gap, S&P 500EarningsNo edge0.21% behind over 20 daysRead the test
Beat and popped 3% or moreEarningsNo edge+0.50% over 20 days, not different from random picksRead the test
Earnings surprise driftEarningsNo edge1.71% behind over 60 days in the last 18 monthsRead the test
Beat but sold offEarningsNo edgeBounced no more than other dips in the same stocksRead the test
Strong on its own (residual momentum)MomentumMixedPassed at first, not confirmed in the last 18 months
Correction deniersMomentumDid worse1.80% behind over 20 days
Holding up on down daysMomentumNo edge0.35% behind over 20 days
Hot industriesMomentumDid worseLeaders lagged the average stock the next week
Highest buyback yield, S&P 500FundamentalsNo edge+0.12% a month over the average member, not different from luckRead the test
Shrinking share countFundamentalsNo edge+0.10% a month, ahead in 52 of 89 monthsRead the test
Biggest stock issuersFundamentalsDid worse0.31% a month behind the average memberRead the test
Heavy buyers after the buyback blackoutFundamentalsNo edgeDid no better once their buyback window openedRead the test
Dip below the buyback priceFundamentalsNo edgeTrailed similar dips by 0.38% over 20 daysRead the test
AI headline scoringNews and AINo edgePrices moved before the next open
Indicator model, next day directionNews and AINo edge49.9% right, a coin flip
Bitcoin and Ethereum trend followingCryptoPositiveAbout the same return as holding with a far smaller worst drop; held up in the last 18 monthsRead the test
Selling crypto on big drops during the dayCryptoDid worseCut Bitcoin from about 79% a year to 49%; most drops bouncedRead the test
Buying a big one day jump off a crypto bottomCryptoPositiveA small gain on both coins, outside bear markets only; few signalsRead the test
Rotating between Bitcoin and EthereumCryptoMixedFar better in the last 18 months, deeper worst drop before thatRead the test
Range trading sideways crypto marketsCryptoDid worseWorse in every period on both coinsRead the test
Altcoins only in an uptrend (SOL, XRP)CryptoPositiveAbout flat in the last 18 months while the coins fell about 45%Read the test
Buying crypto capitulation wicksCryptoNo edgeRarely fired and added nothingRead the test
Crypto bounce trades after oversold daysCryptoNo edgeLess than half a point a year either wayRead the test
Weekly buys (DCA) while crypto is beaten downCryptoDid worseSame return, deeper worst dropsRead the test

Results are against the average eligible stock over the same days, after costs, from mid 2018 to May 2025, unless stated. Crypto results are against simply holding the coin, from 2013 for Bitcoin and 2016 for Ethereum. The most recent 18 months are held back and checked separately.

What did hold upSlow, 12 month momentum, and a few timing strategies of our own. We track them live in our rules based portfolios.
See the portfolios

Backtest questions

Why do most popular trading strategies fail?

Anything a screener can find in seconds is found by everyone. Once a pattern is widely traded, the extra return tends to disappear. In our tests, most chart setups did about the same as picking a stock at random on the same day.

What did work?

Buying the stocks with the strongest gains over the last 12 months, leaving out the most recent month, held up best. It beat the average stock in our data, though with large swings from month to month.

Does a failed backtest mean a setup is useless?

Not always. Some setups are useful context for timing a trade in a stock you already like. They just did not beat the market on their own.

How do you test a strategy?

Every signal is compared with random stocks bought on the same day and held the same way, after trading costs. The rules are fixed before we look at results, and the most recent 18 months are held back and checked separately.

Backtest results are hypothetical. They are based on historical prices, include a simple trading cost and leave out taxes and other costs of real trading. Past results do not guarantee future returns. For informational and educational purposes only. Not investment advice, and not a recommendation to buy or sell any security.