Reversal Candlestick Structure finds sixteen candlestick patterns on MT4 and MT5, and then does the thing almost no pattern indicator does: it counts how often each one showed up in the context that gives it its name.
The number it puts on screen is a context rate. It is not a win rate, and the difference matters.
A hammer and a hanging man are the same candle
Open the source and the two tests sit a few lines apart. Both ask for the same thing:
lower wick more than twice the body, body under half the range, upper wick under 2% of the range, and not a doji
That's it. Character for character, the geometry of a hammer and the geometry of a hanging man are identical. The same is true of the inverted hammer and the shooting star, which share a single test of their own with the wicks the other way up.
So what separates them? One line each. The hammer is that shape while the trend read is oversold. The hanging man is that shape while the trend read is overbought. Nothing else.
Most tools skip that second half and label the candle from its shape alone. Which means that on any chart where you've seen a "hammer" marked in the middle of a rally, you were looking at a hanging man wearing the wrong name.
What the percentage actually counts
Every pattern keeps two running totals. One counts every bar where the geometry matched. The other counts the subset where the geometry matched and the trend context agreed. The number on the dashboard is the second divided by the first.
Read it as a sentence: of all the hammer-shaped candles on this chart, this share of them actually occurred in a place where the pattern means what the textbook says.
Look at the gold H1 screenshot above. Hammer reads 0.00. Across 5028 bars, every single hammer-shaped candle on that chart appeared while the trend read was not oversold. Not one of them was a hammer. They were all hanging men, or nothing at all.
That's a fact about that chart, and it's the sort of fact you can't get from a tool that only draws shapes.
I want to be blunt about the limit of it, because the number looks like the kind of number people over-read. It says nothing about what price did next. It doesn't track outcomes, it doesn't measure profit, and a pattern showing 60% is not a pattern that works 60% of the time. It is a pattern that was correctly named 60% of the time. Those are different claims and only the second one is being made.
The context is two conditions, not one
The trend read is a stochastic over Trend Length bars, default 14, plus a smoothed version of itself controlled by Warmup Length.
Overbought requires both the raw stochastic and the smoothed one to sit at or above Threshold, default 80. Oversold requires both at or below 20. Asking for both is what stops a single spiky bar from flipping the context, and it's why the marks on the chart come in runs rather than one-offs.
Those runs are the small Wingdings arrows. They're the Show reversal potential layer, and they mark bars where the context alone says price is stretched, with no pattern required.
The same tool, three different answers
Here's the H4 chart of the same instrument:
Hammer goes from 0.00 to 3.85. Inverted Hammer goes from 10.00 to 0.00. Shooting Star nearly doubles. Same instrument, same code, same 5028 bars, different timeframe.
And on a different instrument entirely:
Bullish Engulfing reads 18.24 on EURUSD against 7.14 on gold. Bull Harami reads 20.00 against 0.00.
No universal table of these numbers exists, which is why the tool computes them on your chart.
The sixteen patterns
Eight bullish and eight bearish, each with its own on and off switch: hammer, inverted hammer, bullish engulfing, rising three, three white soldiers, morning star, bullish harami, tweezer bottom, and the bearish mirror of each.
They are not all equally strict. Rising three wants a wide-bodied green candle, then three red candles that all stay inside its range, then a green candle that closes above the whole thing. Three white soldiers wants each body to occupy more than 70% of its candle's range and each open to sit inside the previous body. Those are precise.
The single-candle patterns are the loose ones, and they're exactly the ones the context filter rescues.
| Ordinary pattern indicator | This | |
|---|---|---|
| Hammer against hanging man | Shape only, so both get the same label | Shape plus trend context, which is the only real difference |
| What it reports | That a pattern appeared | What share of that pattern's appearances were in context |
| Per instrument | The same rules everywhere | The counts are measured on your chart |
| Live bar | Often marked immediately | Excluded; patterns confirm on closed bars |
| Alerts | Usually one channel | Terminal, email, push or Telegram, named per pattern |
What it draws, and what it doesn't
Two-letter tags at the qualifying bars, the reversal-potential arrows, and the dashboard. Prefix exists so you can run two copies with different settings without their objects colliding.
No indicator buffers at all. Everything is a chart object, so iCustom gets you nothing and an EA can't read the patterns or the rates. The alerts are the only way out, and they're on by default here, one per pattern with the timeframe in the message.
No entries, no stops, no targets, no higher-timeframe read.
Where it's weak
The dashboard depends on how much history you've loaded. Lookback bar defaults to 5000. Two charts of the same symbol with different history depths will show different rates, and the panel prints the bar count so you can see which window you're reading. Set it to 0 for everything the terminal has.
The counts never reset. They accumulate across the whole window, so a regime that ended two years ago is weighted the same as last month.
The context filter is a stochastic, and that's all it is. No structure, no higher timeframe, no volatility. On a strong trend the stochastic can sit pinned at an extreme for a long stretch, and every pattern that forms in there inherits that context whether or not the trend is genuinely tired.
A flat range divides by zero. The stochastic divides by the range of the last Trend Length bars, and the wick percentages divide by the candle's own range. On a dead session where those come out equal, the arithmetic has nothing to protect it. It's rare on liquid symbols and it's real on illiquid ones.
Sixteen patterns and sixteen alert lines is a lot of noise. Notifications ship on. On a handful of charts that's a stream nobody reads, so turn off the patterns you don't trade before you turn on a channel.
Single-candle patterns still fire constantly. The filter changes what gets counted as a proper reversal, not how often the shape appears. The tags on the chart are the shape; the dashboard is the judgement.
Questions you'll have
Does Reversal Candlestick Structure repaint? No. Everything runs behind a new-bar check and the loop stops one bar short of the forming candle, so a tag is placed on a closed bar and stays there.
Is the percentage a win rate? No, and it's worth being clear. It's the share of that pattern's appearances that happened in the trend context matching the pattern's name. It says nothing about what price did afterwards.
Why does Hammer show 0.00 on my chart? Because every hammer-shaped candle in your loaded history appeared while the trend read was not oversold. That's a real answer about your instrument, not a fault. It's common on trending symbols.
What's the difference between a hammer and a hanging man here? Only the context. The geometric test is identical; the hammer needs an oversold read and the hanging man needs an overbought one.
Why do the numbers change when I switch timeframe? Because they're measured on the bars you have loaded. The patterns and the context are both computed on the chart timeframe, so H1 and H4 are two different measurements.
What are the small arrows above and below the candles? The reversal-potential marks. They flag bars where the trend read alone is stretched, with no pattern required, and they're controlled by Show reversal potential.
Can an EA read the patterns? No. There are no indicator buffers; every element is a chart object. Use the alerts.
Best timeframe? H1 and H4. On M5 and M15 the single-candle patterns appear constantly and the 14-bar stochastic spends most of its life mid-range.
MT4? Both platforms are included, one download each. Same sixteen patterns, same context filter, same dashboard.
Support
Send me a message with a screenshot, the symbol and the timeframe, and I'll tell you what the tool was doing on that bar.
Analysis tool. It places no trades and makes no profitability claim. Past price behaviour is not a guide to future price behaviour.