How to Scan for Hammer Candlesticks

A hammer scan is two jobs. A screener does the first exactly: the three ratio tests are arithmetic on the open, high, low and close, and a machine applies them without argument. The second job, deciding whether the candle formed somewhere that matters, is one no screener can do, and it is the one that determines whether any of the results are worth trading.

Getting value out of scanning comes from knowing which half you are automating.

The three filters a screener can apply

All three are measured against the candle’s total high-to-low range, which is high minus low.

Lower shadow ≥ 66% of range. The lower shadow is the distance from the low up to whichever of open or close is lower. This is the rejection the pattern is named for, and without it there is no pattern.

Real body ≤ 33% of range. The body is the absolute difference between open and close. A large body means the session had directional conviction, which is a different message entirely.

Upper shadow ≤ 10% of range. The upper shadow runs from the high down to whichever of open or close is higher. A significant one means sellers capped the rally too, and the candle is closer to a spinning top.

Most screeners expose open, high, low and close as variables, which is all three conditions need. Where a platform offers a built-in “hammer” filter, it is worth checking what thresholds it uses before trusting it, because implementations vary and some are considerably looser than the ratios above.

The full reasoning behind each threshold is on our hammer candlestick page.

The filters that make the list tradable

Shape filters alone return a list dominated by things you would never trade.

Liquidity. Set a floor on average daily volume. A perfect hammer on a stock trading a few thousand shares a day is not an opportunity, because the spread will take more than the setup offers. The cost side of that is in trading costs.

Price. A floor here removes the very low-priced names where a single tick is a large percentage move and the ratios become meaningless.

Range. A minimum candle range, expressed in ATR or as a percentage of price, removes candles so small that their proportions are an accident of rounding rather than a description of a session.

The filter that does the most work

Add a condition for a preceding downtrend, and watch the result count collapse.

This is the single highest-value filter and the one most often left out. Long lower shadows occur constantly, in rising markets and falling ones, and the hammer only means anything after a decline. Without a trend condition the scan is returning candle shapes rather than setups.

Implementations vary by platform, but something like a close below a medium-term moving average, or several consecutive sessions of lower highs and lower lows, works. The precise formulation matters less than having one at all.

Expect this filter alone to remove most of the list. That is the filter working.

What the screener cannot do

Everything that remains is a judgement, and it has to be made chart by chart.

Is the hammer at a level you had already marked? A prior swing low, a support zone, the bottom of a range. A hammer in open space is a candle shape; a hammer at a level you identified before it printed is a setup. Reading those levels is covered in support and resistance.

What did volume do? Heavy volume on the hammer means real size changed hands at the lows. Thin volume means the bounce came from an absence of sellers, which is a much weaker thing to bet on. Some screeners can filter on relative volume; none can tell you whether the figure makes sense for that particular stock.

What is the higher timeframe doing? A daily hammer against a falling weekly trend is a different proposition from the same candle in a weekly uptrend.

Is this the fourth hammer in a fortnight on the same chart? Repeated hammers that keep failing say supply is heavy, not that a reversal is overdue. No screener has that memory.

A workable routine

Run the scan after the close, so the candles are final.
Apply the three ratio filters, plus liquidity, price and minimum range.
Apply the downtrend condition.
Open every remaining chart. If there are more than you will genuinely look at, tighten the trend condition rather than skimming.
Mark the ones sitting at a level you can name, and discard the rest.
For survivors, set the stop below the hammer’s low first, then size from that distance using ATR-based stop placement and the position size calculator.

Step six comes last for a reason. A hammer with a very long lower shadow is the most convincing version of the pattern and carries the widest stop, which means the smallest position. Deciding size before looking at the stop distance gets that backwards, and it is the most common way a good scan turns into a badly sized trade.

The shape filter catches two patterns, not one

Worth knowing before you trust a result list: the three ratio tests describe the hanging man exactly as well as they describe the hammer. The two candles are identical. Only the preceding trend tells them apart, and the hanging man is bearish.

So a scan running shape filters alone returns a mixture of bullish hammers in downtrends and bearish hanging men in uptrends, with no marker distinguishing them. Acting on that list as though every result were a hammer means taking roughly half of the trades in the wrong direction.

The downtrend condition is what separates them, which is the second reason it is not optional. It is not improving the quality of the list; it is deciding what the list contains. The same logic applies to the inverted hammer, whose shape is identical to the shooting star.

If your platform’s built-in hammer filter returns results in obvious uptrends, that is what has happened, and it is a reason to build the conditions yourself rather than trust the preset.

On scanning more generally

A scan is a way of looking at more charts than you could by hand, not a way of avoiding looking at charts. The output is a list of candidates, every one of which still needs the judgement a screener cannot supply.

The failure mode is a list long enough that it gets skimmed, because a skimmed list produces trades chosen on shape alone, which is the thing every piece of advice about this pattern warns against. A short list you read properly beats a long one you scroll past, and tightening the trend condition is usually the fastest way to get there.

Frequently asked questions

Can a screener find hammer candlesticks reliably?

It can measure the shape exactly, because the three ratios are pure arithmetic on the open, high, low and close. It cannot judge whether the candle formed at a level that matters, which is the part that decides whether the pattern is worth trading.

What filters should I use?

Lower shadow at least 66% of the candle's range, real body no more than 33%, upper shadow no more than 10%. Add a liquidity floor and a price floor so the results are tradable, then add a downtrend condition so you are not returning hammers from the middle of ranges.

Why does my scan return hundreds of results?

Almost always because there is no trend condition. Long lower shadows are common in every market condition, and without a preceding decline most of what returns is noise. Adding a requirement for several sessions of lower highs and lower lows usually cuts the list by most of its length.

Should I scan intraday or end of day?

End of day for daily hammers, run after the close so the candle is final. A hammer that exists at two in the afternoon frequently does not exist at the close, and scanning intraday produces a list of candles that have not finished forming.

How many results should a good scan return?

Few enough to look at every one properly. If the list is longer than you will genuinely review chart by chart, the filters are too loose, and a list you skim is worse than a shorter list you read.