Swing Breakout Sequence looks for one six-point structure on MT4 and MT5, and it is the most selective tool I sell. On gold H1, over 5003 bars, it drew a single model. On EURUSD H1 over the same depth it drew none at all.
That's the honest headline, and it's the thing to understand before buying rather than after.
The six points
Pivots come from a fast fractal, one bar either side at the default Pivot Period. The tool then asks whether the last six of them, oldest to newest, form a specific shape. For a bullish model:
- 1 a high, 2 a low, 3 a high, 4 a low, then a final low labelled x
- point 4 sits below point 2, so the pullback is making lower lows
- point 1 sits below point 3, so the run before it was building upward
- the final low x goes below point 4, which is the sweep
- the high between 4 and x never exceeds point 3
Five geometric conditions, and none of them is unusual on its own. What makes the model rare is the two that follow.
Two ATR gates, and a break of structure before any of it
Both of the first two legs have to be big. The distance from 1 to 2, and the distance from 2 to 3, each has to exceed ATR mult times ATR, 1.5 by default. A six-point shape made of small legs is thrown out no matter how neat it looks.
Then there's a precondition that runs on a completely separate fractal. Swing condition defaults to BOS, measured with Swing condition Length of 5 bars rather than the model's 1, and it has to have happened within Swing condition timelife bars, 50 by default.
So two pivot scales are running at once and doing two different jobs. The slow one decides whether the market has recently broken structure at all. The fast one looks for the entry shape inside that window. Neither is sufficient alone, which is the whole design.
Those are the BOS lines you can see across the chart above. They're the precondition layer, and they're drawn whether or not a model ever follows.
What happens after the model appears
Drawing the six points is not the signal. The model sits in state 0 and waits.
If price closes above point 2 within Valid model in candle bars, three by default, the model confirms. That's when the entry mark, the stop rectangle and the target appear, and when the alert fires. If those three candles pass without a close above point 2, the model is abandoned and nothing is ever drawn for it.
After confirmation there are exactly two exits. A close at or beyond point 3 is the target. A close back below the sweep low x is the stop. TP type chooses between using point 3 itself, which is the default, or a fixed reward multiple of the stop distance set by TP by R:R.
Bearish models are the exact mirror.
The panel, and what it is not
Top right, a small table headed sbs winrate: how many buy and sell models confirmed, and what share of them reached target before stop.
It is labelled winrate and I'd rather describe it accurately than let the label do the talking. It is an arithmetic count over the bars you have loaded, on one symbol and one timeframe, assuming a fill at the close of the confirming bar with no spread and no slippage. It is not a backtest and it is not a forecast.
On the gold chart above it reads one sell, zero. One sample. A single model that didn't reach target tells you almost nothing about the model and quite a lot about how rarely it appears, and only the second of those is a real measurement.
| Ordinary structure indicator | This | |
|---|---|---|
| How often it marks something | Every few bars | Once in thousands of bars at defaults |
| Pivot scales | One | Two, a fast one for the shape and a slow one for the precondition |
| Leg size | Not measured | Both early legs must exceed 1.5 ATR |
| The drawing | Is the signal | Is a candidate that expires in three candles |
| Exits | Usually none | Target and stop both defined by the model's own points |
Read this before you judge it on your chart
If you install this and see nothing, that is the expected result. Here's where to look first.
Swing condition is the strictest filter. Set it to none and the model no longer needs a recent break of structure, which is the single biggest change you can make. Swing condition timelife at 50 bars is the next one: it means the break has to be recent, and widening it lets older structure still count.
ATR mult at 1.5 is the third. Lowering it admits smaller legs.
Pivot Period at 1 is worth raising rather than lowering. A one-bar fractal produces a jagged zigzag, and a six-point alternating shape is genuinely hard to satisfy on jagged data. A slower pivot gives cleaner swings, and the ATR gates get easier to pass because the legs are bigger.
I'd rather say all that plainly than have you discover it on a live chart.
What it draws, and what it doesn't
The numbered points, the entry and stop rectangles, the target line, the BOS and CHoCH lines from the precondition layer, and the panel. Prefix keeps two copies from colliding if you run different settings side by side. Show pivot point exposes the raw fractal if you want to see what the zigzag is doing.
No indicator buffers. Everything is a chart object, so iCustom gets you nothing and an EA can't read the model, the entry or the stop. The alerts are the only way out, and they name the direction and the pivot period.
No zones, no order blocks, no higher-timeframe read.
Where it's weak
It is rare enough to feel broken. One model per five thousand bars is not a typo. If you want a tool that gives you something to look at every session, this is the wrong one.
Three candles is a tight confirmation window. A model that would have worked on the fourth bar is discarded and never drawn, so the panel never sees it either.
The panel's totals assume a perfect fill. Close of the confirming bar, no spread, no slippage, no partials. On gold the spread alone is a meaningful share of a small model.
One-bar pivots are noisy. The default Pivot Period of 1 marks a swing at almost every wiggle, which sounds like it would produce more models and in practice produces fewer, because the six-point sequence rarely survives the noise.
Nothing adapts. The ATR gate is the only thing that scales with the instrument. The 50-bar precondition window, the three-candle confirmation and the pivot period are all fixed bar counts, so they mean very different things on M5 and D1.
The panel resets on every load. Change timeframe or restart and the counts start again over whatever history the terminal gives you.
Questions you'll have
Does Swing Breakout Sequence repaint? No. The loop runs behind a new-bar check and stops one bar short of the forming candle, so points and confirmations land on closed bars and stay there. A model that expires unconfirmed is removed rather than moved.
Why do I see no models at all? Because at default settings that is normal. Six conditions have to agree, including a break of structure within the last 50 bars. Set Swing condition to none first, then widen Swing condition timelife, then lower ATR mult.
What do the BOS lines mean if there's no model? They're the precondition layer, drawn on a separate 5-bar fractal. They tell you the market broke structure there. A model may or may not follow.
What are the numbers 1 to 4 and the x? The six-point sequence, oldest to newest. The x is the swept low that becomes the stop level, and point 3 is the default target.
When does the signal actually fire? When price closes beyond point 2 within three candles of the sweep. Drawing the points is not the signal.
Is the panel percentage a win rate I can expect? No. It counts, over the bars you have loaded, whether target came before stop assuming a perfect fill. With one or two samples it is not a statistic at all.
Can an EA read the model? No. There are no indicator buffers; every element is a chart object. Use the alerts.
Best timeframe? H1 and H4, and raise Pivot Period before judging it lower down. On M15 the fast fractal spends its time on noise.
MT4? Both platforms are included, one download each. Same six conditions, same two pivot scales, same panel.
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.