In this lesson
A hammer looks like it tells a story on its own. The price fell hard during the candle, then buyers pushed it all the way back up, leaving a long tail below a small body. The shooting star is the same story upside down: a rally that was sold off before the candle closed.
Both patterns are in every beginner's guide. In this lesson we test them on EURUSD, GBPUSD, USDJPY and gold, October 2024 to September 2026, with the same method as lesson 2. We also run the test twice, with a strict and a loose definition, because the two give very different answers.
What hammers and shooting stars are
- A hammer has a small body near the top of the candle, a long lower wick and little or no upper wick. It appears after the price has been falling, and it is read as a signal to buy.
- A shooting star has a small body near the bottom, a long upper wick and little or no lower wick. It appears after the price has been rising, and it is read as a signal to sell.
The pattern's last candle opened 2026-09-11, 09:00 UTC. White = up candle, dark = down candle.
The exact rule we tested (hammer)
The body is no bigger than a third of the candle's full range. The lower wick is at least twice the size of the body. The upper wick is no more than a tenth of the range, so the long lower wick makes up most of the candle. And the price had been falling: the close before the hammer was lower than the close five candles earlier. The shooting star is the mirror image, after a rise.
The pattern's last candle opened 2026-07-09, 05:00 UTC. White = up candle, dark = down candle.
We entered at the next candle's open, with the stop just beyond the end of the long wick: below the hammer's low, or above the shooting star's high. The target was the same distance away (1:1), or twice as far (1:2). Spread paid on every trade.
Strict hammers are rare
With the rule above, a real hammer is not an everyday candle. On the 1-hour charts we found 194 on EURUSD in two years, and 149 to 213 per market. On the 4-hour charts there were only 32 to 39 per market, too few to judge anything, so this lesson works with the 1-hour results.
| Market | Next candle right | Won (1:1) | Random, same stop |
|---|---|---|---|
| EURUSD 1-hour, 194 signals | 51% | 49% −0.02R a trade | 44% |
| GBPUSD 1-hour, 181 signals | 56.4% | 44.8% −0.10R a trade | 42.2% |
| USDJPY 1-hour, 213 signals | 47.9% | 44.6% −0.11R a trade | 45.3% |
| Gold (XAUUSD) 1-hour, 149 signals | 54.4% | 51.7% +0.03R a trade | 42.9% |
| EURUSD 4-hour, 39 signals | 43.6% | 42.1% −0.16R a trade | 49.1% |
| GBPUSD 4-hour, 32 signals | 56.2% | 46.9% −0.06R a trade | 56.8% |
| USDJPY 4-hour, 38 signals | 57.9% | 57.9% +0.16R a trade | 43% |
| Gold (XAUUSD) 4-hour, 33 signals | 45.5% | 54.5% +0.09R a trade | 48.4% |
1:1 = target as far away as the stop. Spread paid. Bold = the pattern won more often than random trades with the same stop. Under each win rate: the average result per trade in R (the amount risked).
The next candle went up 47.9% to 56.4% of the time, depending on the market: from a little worse than a coin flip to a little better.
The trades. Hammers beat random trades with the same stop on 3 of 4 markets. On EURUSD, 49% of the 1:1 trades won, against 44% for random trades. But the average result was still below zero on 3 of the 4 markets, and the best one, gold, made only +0.03R per trade.
Is that more than luck? With about two hundred trades, a win rate is a rough estimate. On EURUSD the range we can be 95% sure of runs from 42% to 56%, wide enough to include the random result. The whole range sat above the random trades on 1 of 4 markets, gold, and even there it did not clear random trades in the same direction, which won 47.4% (more on that comparison below). Two years of strict hammers are not enough to tell a small edge from chance.
The loose version many traders use
Many guides are less strict. They allow an upper wick of up to a quarter of the candle, and some count any small body with a long wick. We ran that version too.
The loose rule found 2,065 hammers on the 1-hour charts, against 737 for the strict one, and they beat random trades on 4 of 4 markets. That sounds like a better pattern. It is not, and here is why.
A hammer is always a buy. The random trades we compare with are half buys and half sells. So if a market mostly rose during the test, any buy signal beats those random trades, whether the pattern means anything or not. And some of these markets did rise: on the 4-hour charts, 54.1% of USDJPY candles and 53.6% of gold candles closed up.
So we also compared every pattern with random trades in the same direction, with the same stop. For loose hammers on the 1-hour charts:
- EURUSD: hammers won 47.8%, random trades 44.5%, random buys 48%.
- Gold: hammers 46.3%, random trades 42.5%, random buys 48.5%.
On both markets, the loose hammer's lead over random trades came from buying, not from the candle. Buying at a random moment did as well or better.
Why this check matters
Many pattern tests compare a buy signal with zero, or with random trades in both directions. In a rising market, that makes every buy signal look good. The fair comparison for a pattern that only buys is a strategy that only buys. Keep this in mind when you read anyone's backtest, including your own.
Shooting stars
Strict shooting stars were as rare as hammers: 156 to 240 per market on the 1-hour charts. They beat random trades on 3 of 4 markets, and the average result was above zero on 0 of 4. The worst was GBPUSD: 40.2% of trades won, against 45.5% for random trades.
| Market | Next candle right | Won (1:1) | Random, same stop |
|---|---|---|---|
| EURUSD 1-hour, 240 signals | 53.3% | 48.3% −0.03R a trade | 45.1% |
| GBPUSD 1-hour, 189 signals | 50.3% | 40.2% −0.20R a trade | 45.5% |
| USDJPY 1-hour, 175 signals | 52% | 45.1% −0.10R a trade | 42.1% |
| Gold (XAUUSD) 1-hour, 156 signals | 50% | 50% +0.00R a trade | 48.6% |
| EURUSD 4-hour, 42 signals | 61.9% | 54.8% +0.10R a trade | 44.4% |
| GBPUSD 4-hour, 43 signals | 37.2% | 32.6% −0.35R a trade | 38% |
| USDJPY 4-hour, 43 signals | 32.6% | 37.2% −0.26R a trade | 43.3% |
| Gold (XAUUSD) 4-hour, 43 signals | 48.8% | 46.5% −0.07R a trade | 41.4% |
1:1 = target as far away as the stop. Spread paid. Bold = the pattern won more often than random trades with the same stop. Under each win rate: the average result per trade in R (the amount risked).
The 4-hour charts again had too few signals to judge, about forty per market.
What to take from this
How you define a pattern decides what you are testing. The strict hammer is rare, and two years of it cannot show an edge either way. The loose hammer is common, and it looked better than random trades only because it always bought in markets that were rising.
If you trade these candles, write your exact rule down, including how long the wick must be and what must have happened before it, then test it against random trades in the same direction. Lesson 8 shows how many trades you need before a result means something, and lesson 9 how to run the test yourself.
Key takeaways
- A hammer is a small body with a long lower wick after a fall (a buy); a shooting star is the mirror after a rise (a sell).
- With a strict rule, they are rare: 149 to 213 hammers per market on 1-hour charts in two years, and only a few dozen on 4-hour charts.
- Strict hammers beat random trades on 3 of 4 markets, but none beat random buys by more than luck could explain.
- The loose rule found 2,065 hammers and looked better, but only because it always bought: against random buys, its lead disappeared on EURUSD and gold.
- Always compare a pattern that only buys (or only sells) with random trades in the same direction.
Questions
Is the hammer candle a reliable buy signal?
Not that we could show. Strict hammers on EURUSD, GBPUSD, USDJPY and gold from October 2024 to September 2026 did not beat random buys by more than luck could explain, and the loose version did no better than random buys.
What is the difference between a hammer and a shooting star?
A hammer has its long wick below a small body and comes after a fall; it is read as a buy. A shooting star has its long wick above a small body and comes after a rise; it is read as a sell.
Why compare a hammer with random buys and not just random trades?
Because a hammer only ever buys. In a market that mostly rose, any buying rule beats random trades in both directions. Random buys show what the direction alone was worth.
How long should the wick of a hammer be?
There is no single standard. In our strict rule the lower wick is at least twice the body and the upper wick at most a tenth of the candle, so the lower wick is most of the candle. Looser rules find far more hammers, but those did no better than random buys.
Where should the stop go on a hammer?
The usual place, and the one we tested, is just below the end of the long lower wick. For a shooting star, just above the end of the upper wick.
Candlestick patterns, tested
- 1. How to read a candle
- 2. Engulfing patterns, tested
- 3. Hammers and shooting stars, tested
- 4. The doji: indecision or reversal? soon
- 5. Inside bars and breakouts, tested soon
- 6. Morning and evening stars soon
- 7. 1-hour, 4-hour or daily: where costs bite soon
- 8. Luck or edge: reading a test honestly soon
- 9. Test a pattern yourself soon
Educational content, not financial advice. Trading forex and CFDs carries a high risk of losing money. Examples use round numbers to show the method; check your own broker's contract specifications and your firm's rules.