SMC Back to Basic looks for one specific sequence and ignores everything else. Five things have to happen, in order, before it marks anything on your MT4 or MT5 chart. On gold H1 over three weeks that produced three setups.
The sequence
Read it top to bottom. Every step has to happen after the one before it:
- BC or SC — a climax. A buying climax starts a sell model, a selling climax starts a buy model.
- AR — the automatic reaction away from the climax.
- test — price comes back and tests the climax area without exceeding it.
- HUNT — a liquidity sweep past the test, taking out the stops sitting beyond it.
- Displacement through the order block, then RTO — price returns to the order block. That return is the signal.
Miss any step and there is no model. Get all five and you get one marked setup, with the block drawn and entry, stop and two targets attached.
That's why the counts are low. If you are used to a tool that shades every order block on the chart, this will feel almost silent by comparison. The silence is the product.
The panel is a tally, not a track record
Top left the indicator counts what happened on the history currently loaded on that chart: setups confirmed, RTO signals, how many were voided or expired, how many sweeps failed, and how many reached target versus stop.
This is a description of one chart's loaded bars, not a performance statistic. The samples are tiny — the gold chart above has three settled models, the EURUSD chart below has one — and a figure computed from one or three outcomes tells you nothing that would survive another month of data. Scroll the chart and the numbers change. Switch symbol and they change again. The indicator's own footer says the same thing: levels are structural, not trading advice.
I've left the counter visible in both screenshots rather than cropping it out, because seeing a low number next to a high one on two charts is the clearest way to show what it's worth.
What the counter is genuinely good for is the voided and failed sweeps rows. Those tell you how often the structure started forming and then broke, which is a real property of the instrument and the timeframe you are on, and it is not something you can see by eyeballing the chart.
Levels
Every model carries an entry, a stop and two targets.
inp_sl_type puts the stop either beyond the sweep extreme (the default) or at the order block edge. inp_tp_type sets targets either as multiples of the stop distance — 1.25R and 2.0R by default — or as fib levels of the last swing leg.
inp_size_mode decides whether the thresholds behind all this are measured in ATR multiples or in raw points. ATR is the default and it is what lets the same settings travel between gold and a currency pair. Points mode exists for people carrying settings over from version 2.
What version 3 added
Several models are tracked at once, up to twenty-four, and a confirmed model is never overwritten by a newer candidate — which is why you can see the whole signal history on the chart instead of only the most recent one.
The results are published through indicator buffers, so an EA can read them with iCustom: the signal direction, entry, stop, TP1 and TP2. The buffer layout is deliberately identical on MT4 and MT5, so one EA can read either build without a branch.
Optional and off by default: a higher-timeframe bias filter that gates signals to the side of an EMA on the timeframe you choose, and a multi-timeframe panel that scans three timeframes for models.
inp_legacy reproduces version 2.0's behaviour exactly, for anyone who tuned against it.
| An order block indicator | A structure indicator | This | |
|---|---|---|---|
| What gets marked | Every block | Every BOS/CHoCH | Only a completed five-step sequence |
| Typical count | Dozens | Dozens | A few per month |
| Entry defined | No | No | The return into the block |
| Stop and targets | No | No | Yes, ATR-sized, two targets |
| Outcome tracking | No | No | A tally on the chart |
| EA readable | Objects only | Objects only | Buffers, same layout on both platforms |
Where it's weak
Swing detection is slow by design. The default swing window is 60 bars left and 30 right, so a swing pivot is not confirmed until thirty bars after it forms. Everything downstream inherits that. This is not a tool for reacting quickly.
Few signals means few samples. You cannot judge settings from a month of chart. That cuts both ways and it is the honest consequence of a strict sequence.
The panel can sit under a drawing. It is anchored to a fixed screen position and a large order block extending to the right can cover it. Move it with inp_panel_x / inp_panel_y.
The report resets with the chart. It counts whatever bars are loaded. Scroll back to load more and the numbers change; they are not stored anywhere.
Targets are R multiples by default. 1.25R and 2.0R are derived from the stop, not from structure. The fib mode is there if you want structural targets instead.
No trend context unless you turn it on. The higher-timeframe bias filter is off by default, so models form against the larger trend as freely as with it.
Two detection methods, and Both is the default. SMC_METHOD_BOTH runs the classic and the extended sweep definitions together, which produces more models than either alone. If you want the stricter original behaviour, pick Classic.
Questions you'll have
What sequence does it actually look for? Climax, automatic reaction, secondary test, liquidity sweep, then the return into the order block. All five, in order.
Why are there so few signals? Because all five steps have to complete. That is the design, not a fault.
Is the hit rate in the panel a performance figure? No. It counts outcomes over whatever history that one chart has loaded, and the sample is usually a handful of models. Treat it as a description of that chart, nothing more.
Which rows in the panel are actually useful? Voided and failed sweeps. They tell you how often the structure starts and breaks on that instrument and timeframe.
Where is the stop placed? Beyond the sweep extreme by default, or at the order block edge if you prefer. Sized in ATR.
Does it repaint? Swing pivots need inp_swing_right bars to confirm, so labels appear after the fact rather than moving. A confirmed model is never overwritten. A model that never gets its return is marked voided or expired rather than deleted quietly.
Can an EA read it? Yes, and this is the point of version 3: signal, entry, stop, TP1 and TP2 are buffers, with the same layout on MT4 and MT5.
Can I filter by a higher timeframe? Yes, inp_use_htf with a timeframe and an EMA period. Off by default.
What is Legacy behaviour? It reproduces version 2.0 exactly, for anyone who already tuned settings against that build.
Best timeframe? H1 and H4. The 60/30 swing window needs a timeframe where ninety bars is a meaningful stretch.
MT4? Both platforms are included, one download each. Same engine — the model logic lives in a shared file compiled into both — and the same buffer layout.
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.