FREDRead a chart
Tested patterns

Do chart patterns actually work? We tested 1,902,874 trades

Short answer: most don't, on their own. 26 of the 28 setups Fred tested lost money on average after spread, across 8 markets and 3 timeframes from 2020 to 2026. Fred's edge is finding the versions that do work: 38 setups made money in both 2020–23 and 2024–26.

Results at a standard exit: stop just past the pattern, take profit 1.5× the risk. "Per trade" is the average result as a share of the amount risked. Checked 28 October 2026.

SetupTrades testedAverage per tradeHit take profitVersions that work
Evening fade (sell)37,766+2%32%8 green
London breakout (buy)13,511+1%29.8%3 green
Sweep + displacement + FVG + volume (buy)8,137−1%25.5%–
Sweep + displacement + FVG + volume (sell)9,663−2%26.9%–
London breakout (sell)12,068−3%30%1 green
NY open breakout (buy)6,884−3%28.8%–
Sweep + displacement + FVG (buy)16,222−3%24.9%1 green
Bull flag1,072−3%30.6%–
Sweep + displacement + FVG (sell)18,423−4%26.3%–
Double bottom98,075−5%21.6%5 green
NY open breakout (sell)5,804−5%29%–
Evening star33,812−7%34%–
Double top96,630−7%22.5%2 green
Bearish engulfing198,295−8%34.6%4 green
Yesterday's level sweep (buy)52,191−8%34%2 green
Morning star31,023−8%32.4%1 green
Liquidity sweep (bear)131,497−8%35.1%2 green
FVG retrace (bear)14,469−9%32.2%–
Liquidity sweep (bull)119,186−9%34%–
Yesterday's level sweep (sell)60,872−9%34%–
Bullish engulfing191,228−10%33.2%2 green
Trend pullback (buy)237,215−10%32.1%6 green
Trend pullback (sell)213,557−10%32.7%–
FVG retrace (bull)16,263−10%31.3%–
Shooting star121,131−11%34.8%1 green
Hammer121,677−12%33.8%–
Bear flag933−12%27.9%–
Evening fade (buy)35,270−19%22.4%–

Why most patterns lose

A pattern that "looks right" still has to beat the broker's spread, and most only work at certain times, on certain markets, or with the trend. Averaged over everything, the losing trades win. That's why Fred rates every setup green, amber or red, and shows the real odds for the one on your chart.

Got a chart open? Upload a screenshot and Fred checks it against six years of tests.

Read my chart free

5 free reads. For learning and demo practice. Not financial advice.

Questions

Do candlestick patterns work?

On their own, mostly not. Fred tested 28 setups on 1,902,874 trades from 2020 to 2026, after real broker spreads, and 26 of them lost money on average. Some versions do work: on the right market, timeframe and time of day, and in the right direction for the trend, 38 setups made money in both 2020–23 and 2024–26.

Which chart pattern has the best results?

Taken as a whole, the evening fade (selling a break of the previous candle between 6pm and 9pm UK after a strong rising day) was the only setup that made money in both periods without extra filters. Fred's green-rated versions of other setups also made money in both periods.

How were the patterns tested?

Every signal was traded at the next candle's open with the real MT5 spread, a stop just past the pattern and a take profit 1.5 times the risk, closing after 48 candles if neither was hit. Rules were set on 2020–23 and checked on 2024–26.

Back-tested results, not live trading. Past results don't guarantee future ones. See also AI chart tools compared and Fred's live results.