A downtrend makes lower highs and lower lows. Then it pulls back, sweeps the liquidity sitting above the last couple of highs, taps an order block, and carries on down. Everybody who trades this knows the shape.

The problem is that the shape is easy to see afterwards and very hard to see at the right edge of the chart, because in the moment the pullback looks exactly like a reversal. This draws the whole eight-point chain as it forms, names each point, and marks the one zone that matters.

Gold H1. Points B through H, the order block extended to the right, and the entry with its stop and target once price returned to the zone.

The eight points

Left to right, for a sell. The buy model is the exact mirror.

A  B  C  D  E  F  G  H
H  L  H  L  H  L  H  L

Highs must satisfy A > C, and the lows must satisfy B > D > F > H — a proper sequence of lower highs and lower lows. Price then rallies to a point the model calls I, above both E and G so that it sweeps the liquidity resting there, but still below C, where it meets the order block.

That last condition is the whole trade. A rally that clears E and G has taken out the obvious stops. A rally that also clears C has broken the trend and the setup is void. The gap between those two prices is where the model lives.

Two ways to find the structure, and why the default is the second one

Pivot mode BOS chain mode (default)
A swing is a bar higher than the N bars each side a point price retraced from by at least n × ATR
Scale measured in bars volatility
B > D > F > H is tested on price a consequence of three closes below structure
A lower wick that closes back inside can create a swing is not a break
Same setting across symbols and timeframes no yes

Pivot mode is the version almost every free structure tool implements, and it has a specific failure: the scale is fixed in bars. Set the pivot length to 8 and gold in a quiet Asian session and gold on an NFP release get measured with the same ruler. One of the two is always wrong.

BOS chain mode does not search for the points at all. It watches for three consecutive bearish breaks of structure and reads the points out of the chain:

BOS 1  ->  low broken = B ,  high before it    = A
BOS 2  ->  low broken = D ,  high between B,D  = C
BOS 3  ->  low broken = F ,  high between D,F  = E
then   ->  new high   = G ,  current low       = H

Both modes ship, and the dropdown is one click, so you can put the same chart through both and see where they disagree. They usually agree on a clean impulse and disagree in a range, which tells you something worth knowing about the range.

Non-repainting, and what that actually means here

Every point is confirmed on a closed bar and never moves afterwards. A break of structure is a close beyond the level, not a wick, so a spike that closes back inside does not create a point and does not create a break. The order block is taken from bars that have already closed. The stop and target are computed from the ATR of the entry bar, not the current one.

The one thing that is genuinely live is the alert in TOUCH mode, which watches the current bar tick by tick on purpose, because that is what a touch is.

What this rules out: a chart you look at in the evening will show the same eight points it showed at the time. What it does not rule out: point H is only confirmed once price has retraced from it, so the model appears a few bars after the low itself. That delay is the price of not repainting.

The order block, and when it dies

For a sell it is the last bullish candle before the drop away from C, and the zone spans that candle's full range, high to low.

An order block is liquidity, and liquidity is worth something only while price has not traded through it. Through is measured on the close: a sell block dies on a close above its top, measured from the bar after C. A wick that pierces the zone and closes back inside leaves it alive, which is the case that matters, because that wick is usually the sweep you were waiting for.

If the block dies before price ever returns, the whole pattern is dropped rather than left on the chart looking like a plan.

Gold M15. A complete structure with the order block still waiting - price has not come back to it yet, so there is no entry drawn and no entry implied.

Stop and target

SELL:  SL = OB top    + n × ATR      TP = H + n × ATR
BUY :  SL = OB bottom - n × ATR      TP = H - n × ATR

The two offsets point in opposite directions and that is deliberate. The stop is pushed outside the order block, away from the entry, so the sweep that takes the zone does not take you with it. The target is pulled inside point H, towards the entry, so you are out before price has to break the prior extreme to pay you.

If that leaves the target on the wrong side of the entry, the setup has no room and is discarded rather than shown with a negative reward.

Both are drawn as boxes running right from the entry bar: entry to stop in orange, entry to target in cyan.

Gold H4, a sell setup. Labels run A through H in red for the sell side, the order block sits at C, and the target is pulled inside H rather than parked on it.

The panel

SMC Cont. BOS 2.0xATR
Buy  T/W/L    4 / 1 / 3
Buy win %        25.0%
Sell T/W/L    2 / 1 / 1
Sell win %       50.0%

Total, won, lost, for each direction, over the bars it scanned. It counts a pattern as won when price reached the target before the stop and lost when it did not, using the levels the model itself drew.

Read it as a density and behaviour gauge for the chart in front of you, not as a performance record. It knows nothing about spread, slippage, or the fact that you would not have taken every one of them. A run of 0 / 0 / 0 is telling you this symbol and timeframe are not producing the pattern right now, which is genuinely useful and is most of what I use it for.

Alerts

Two modes, at most one per bar and per direction.

TOUCH fires the moment price trades into the order block, checked every tick. CLOSE fires on bar close only, when the bar reached into the zone but did not close beyond it — a slower, cleaner signal that costs you the first part of the move.

The first calculation pass runs silently. Loading the indicator on a chart with two months of history does not replay two months of alerts at you.

What it can't do

It does not draw fair value gaps, breakers, sessions, killzones, or higher-timeframe zones. It draws one pattern. If you want the full point-of-interest engine, that is ICT Pro, and your subscription already covers it.

Point I appears only after the retracement has topped out, which means at the moment you are deciding whether to take the entry, I does not exist yet. The entry is placed at the near edge of the order block for exactly that reason. You are trading the zone, not the confirmed sweep.

The BOS chain gives you a clean invalidation — a close beyond A discards the pattern — but nothing stops price closing beyond A one bar after you are in.

You won't find a win rate table in this listing, and I would treat any indicator listing that has one with some care. The panel's win column is arithmetic on the levels the tool itself drew, over whatever history your chart happens to hold. It is not a backtest and it is not a forecast.

The MT4 build has no Telegram alerts. Terminal alert, mobile push and email all work on both platforms; the Telegram path exists only in the MT5 build, because the shared MT4 library this is built on does not carry it. That is the one real difference between the two files.

Questions you'll have

Which mode should I use? BOS chain, and leave it there. Pivot mode is included so you can see what the bar-counting approach does differently on your own chart, not because it is a serious alternative.

Why is my chart empty? Because the pattern is not there. Three consecutive breaks of structure followed by a sweep into an order block is a specific thing, and on a ranging symbol it can be absent for weeks. Check the panel: 0 / 0 / 0 is an answer.

What timeframe? H1 and H4 are where I look. M15 works and produces more setups; below M15 the ATR-confirmed swings start tracking noise.

MT4 or MT5? Both, and your subscription covers both. Same eight points, same order block rule, same stop and target arithmetic — see the note above about Telegram.

Does it place trades? No. It draws.

Support

Message me through the site with a screenshot that includes the panel, plus the symbol and timeframe. That is usually enough to work out what you are looking at. Updates are included for as long as your subscription is running.

Analysis tool. It places no trades and makes no profitability claim. Past price behaviour is not a guide to future price behaviour.