Trading Education Platform SystemSubscribe for Free
Edition 1 · 25 patterns · tested 8 Oct 2026

Everyone sells a pattern.
We tested them all the same way.

Head and shoulders, double tops, RSI, MACD, order blocks, fair value gaps and 19 more. Same 29 coins, same costs, same rules written before we looked at any result, and a comparison with entering at random. 111,674 simulated trades in total.

See the resultsWhat is R?
  • Rules fixed in advance
  • Real fees and funding
  • Not investment advice
At a glance

25 patterns, one standard

Lose money after costs5Average result below zero with a clear margin.
No detectable edge12Average close to zero or too uncertain to tell from luck.
Too few trades to judge6The rule fires rarely, so a small edge cannot be ruled out.
Small positive, not confirmed2Slightly above zero in part of the data, not in the rest.

Results are in % of the amount risked per trade (and in R), after a 0.14% round-trip cost and real funding. The "better than random entries" column compares each pattern with 1,000 runs that enter at random times with the same stops and targets. Random entries lose money to costs, so beating them is necessary but not sufficient: the average must also be clearly above zero. How we test.

Chart patterns

Chart patterns

Shapes drawn on the chart that courses say predict the next move.

PatternVerdictAverage per tradeTradesBetter than random entries
Head and shoulders
Three peaks, the middle one the highest.
No detectable edge after costs
-4.9% -0.049R
686
28%
Double top and double bottom
Price hits the same level twice and fails.
No detectable edge after costs
+0.1% +0.001R
2,600
86%
Triangles
Highs and lows squeeze toward each other.
No detectable edge after costs
-1.3% -0.013R
4,354
23%
Flags and pennants
A sharp move, a short pause, then, courses say, the move continues by the same amount.
Too few trades to judge
+14.1% +0.141R
173
90%
Rising and falling wedges
Both lines slope the same way and squeeze.
No detectable edge after costs
-2.8% -0.028R
3,356
55%
Cup and handle
A rounded bottom, a small pullback, then a breakout.
Too few trades to judge
+18.7% +0.187R
28
82%
Candles

Candles

One to three candles in a row, taught as turning points.

PatternVerdictAverage per tradeTradesBetter than random entries
Engulfing candle
One candle completely covers the body of the previous one.
No detectable edge after costs
+0.0% +0.000R
2,720
96%
Pin bar and hammer
A candle with a long tail and a small body.
Loses money after costs
-13.7% -0.137R
1,272
1%
Inside bar
A small candle inside the range of the previous one.
No detectable edge after costs
+4.0% +0.040R
8,063
100%
Morning and evening star
Three candles: a big one, a tiny one, a big one the other way.
Too few trades to judge
-9.8% -0.098R
361
17%
Indicators

Indicators

Rules built on RSI, MACD, moving averages, Bollinger bands, Fibonacci and Ichimoku.

PatternVerdictAverage per tradeTradesBetter than random entries
RSI overbought and oversold
RSI below 30 means "oversold", above 70 "overbought".
Loses money after costs
-9.9% -0.099R
6,745
0%
RSI divergence
Price makes a lower low, but RSI a higher low.
No detectable edge after costs
-4.9% -0.049R
1,195
12%
MACD signal cross
The MACD line crosses its signal line.
No detectable edge after costs
-0.1% -0.001R
18,026
99%
MACD divergence
Price makes a new extreme, MACD does not.
No detectable edge after costs
-4.0% -0.040R
1,400
14%
Golden cross and death cross
The 50-day average crosses the 200-day average.
Too few trades to judge
-11.4% -0.114R
301
3%
Bollinger band touch
Price closes outside the Bollinger band.
Loses money after costs
-7.0% -0.070R
14,880
0%
Bollinger squeeze
The bands get as narrow as in the last 100 candles.
No detectable edge after costs
+4.6% +0.046R
3,114
100%
Fibonacci 61.8% retracement
After a move, price pulls back 61.
Too few trades to judge
-10.7% -0.107R
313
7%
Ichimoku cloud breakout
Price breaks out of the Ichimoku cloud.
No detectable edge after costs
-2.5% -0.025R
7,847
68%
Smart Money and levels

Smart Money and levels

Order blocks, gaps, liquidity sweeps, structure breaks and support and resistance.

PatternVerdictAverage per tradeTradesBetter than random entries
Order block
The last opposite candle before a strong move.
Small positive result, not confirmed
+8.6% +0.086R
1,867
100%
Fair value gap
Three candles leave a gap that the middle one covers.
Small positive result, not confirmed
+6.7% +0.067R
10,521
100%
Liquidity sweep
Price pokes above an old high, then closes back below.
Loses money after costs
-9.8% -0.098R
9,364
0%
Break of structure and CHoCH
Price closes beyond the last swing high or low.
No detectable edge after costs
+2.6% +0.026R
9,373
98%
Support and resistance bounce
A level that price touched several times.
Loses money after costs
-12.4% -0.124R
2,740
1%
Supply and demand zones
A tight base of small candles before a sharp move.
Too few trades to judge
+9.3% +0.093R
375
100%
Read this first

What this is not

Not proof that nothing worksWe say "no detectable edge after costs, in these data, with these rules". Different rules, markets or periods may behave differently.
Not a forecastA result in the past says nothing certain about the future, in either direction.
Not our strategiesWe publish tests of popular patterns. We do not publish or comment on strategies we use ourselves.
Not adviceEducation only. Trading leveraged products carries a high risk of loss.
Questions

Pattern Reality Check: common questions

Do chart patterns work in crypto?

We tested 25 popular patterns and indicator rules on 29 crypto perpetual futures, with fixed textbook rules, real trading costs and funding. After costs, 5 lost money, 12 showed no detectable edge, and 6 had too few trades to judge. 2 showed a small positive average that did not hold up in later data. This is a statement about these rules in this data, not about every possible way to trade.

Why do courses teach patterns if they do not work?

Patterns are easy to see on a chart after the fact, and any course can show a chart where one worked. The test that matters is what happens when the same rule is applied to every case, before knowing the outcome, with costs. That is what this section does.

Does this mean no pattern or strategy works?

No. It means that these popular, textbook versions show no detectable edge after costs in our data. Other rules, markets or periods may differ. We publish tests of popular patterns; we do not publish or comment on strategies we use ourselves.

What is R?

R is the amount you decide to risk on a trade. A trade that loses its stop is -1R, one that gains twice the risk is +2R. Every result on these pages is also shown as a percentage of the amount risked. See what is R.