Backtest and automate ICT and SMC concepts

Why are ICT and SMC concepts so hard to test?

ICT and SMC are taught as frameworks for reading a chart, and most of that teaching is deliberately discretionary. “Wait for price to return to the order block, with displacement, in the right session, if structure agrees” is a sentence a human can act on and a computer cannot. Each clause hides a judgement: which block, how much displacement, whose definition of structure.

That is why so many traders spend years on these concepts without ever knowing whether their version works. Without rules, there is nothing to measure, so every losing week can be explained away as a mis-read and every winning week as confirmation. The point of this page is narrower and more useful: not to decide whether ICT or SMC is correct, but to let you pin down your interpretation precisely enough that historical data can answer back.

Turning concepts into rules a backtest can evaluate

The builder works by composition: you place detector blocks, configure their parameters, and the result is a strategy that evaluates identically every time it runs. The detectors cover the structural vocabulary these frameworks use:

• Order blocks — a defined zone from a specified candle, with explicit rules for what counts as a valid return to it.
• Liquidity levels and sweeps — prior highs and lows, and what qualifies as having taken them.
• Structure breaks — break of structure and market structure shift, measured against swing points you define rather than drawn by hand.
• Entry zones anchored to a higher timeframe — a zone identified on a slower chart, with entries taken on a faster one.
• Time as a filter — the True Day Open filter described below is one. Other time windows are among the easiest rules to test, because time is unambiguous, so write the window down and compare results with and without it.

Parameters that are usually left to feel become explicit numbers here. That is the hard part and also the valuable part: once a rule has a number, the number can be wrong, and a backtest can tell you so.

SMT divergence: when correlated markets stop agreeing

SMT divergence looks at two markets that normally move together and asks whether they are still confirming each other at a swing point. If one makes a higher high while the other fails to, the move lacks agreement. Traders read that disagreement as a sign that liquidity is being taken rather than a trend extending.

AKQX detects this as a swing-style comparison across a configurable pair of correlated markets, so the divergence is either present by your definition or it is not. Entry pair selection decides whether both markets are traded independently or only one is, which is the kind of choice that quietly changes a result far more than people expect.

PSP and True Day Open: two AKQX-specific definitions

These two terms have no single agreed public definition, so here is exactly what AKQX means by them.

PSP

A PSP in AKQX is a divergence between two correlated pairs where their candles covering the same period close in opposite directions. The detector marks that candle on the timeframe you choose, and the entry logic can then wait for a body-close breakout of the PSP candle's high or low on a faster timeframe. The marked candle expires after a configurable number of bars, so a setup that is not taken does not sit around indefinitely waiting to be right.

True Day Open (TDO)

The TDO is the opening price of the 00:00 UTC candle for that day, used as a daily directional anchor. A TDO filter can then restrict a strategy to one side of the market: for instance, allowing only sell setups while price trades above the TDO line and only buy setups while it trades below. Because it is a single, unambiguous price, it is one of the few ICT-adjacent concepts that is trivially testable — which makes it a good first filter to measure.

Multi-timeframe confirmation and live anchor tracking

Most of these frameworks are inherently multi-timeframe: context from a slower chart, execution on a faster one. The builder supports that directly — a detector can run on a 30-minute candle while the entry trigger watches a 3-minute candle, with the slower condition gating the faster one.

Live anchor tracking keeps the higher-timeframe reference current as the session develops, rather than freezing it at the moment the strategy was deployed. Each deployed algo also reports a state you can read at a glance: running, waiting for a setup, or paused. A strategy that has sat in “waiting for setup” for days is telling you something useful about how often your conditions actually occur.

Fee-aware stops, and why tight stops fail these setups

ICT and SMC entries often sit close to a structural level, which produces a tight stop, which looks attractive on a reward-to-risk calculation and then quietly fails in practice. The reason is cost: when the stop distance is small, trading fees and slippage become large relative to the risk being taken, and a strategy that is profitable on paper is unprofitable after costs.

AKQX applies a minimum stop distance and a fee-to-risk breaker for exactly this case, and position sizing is fee-aware rather than cost-blind. These guards will sometimes refuse a setup you wanted to take. That refusal is the feature.

Backtest, paper trade, then your own decision

Once your rules are written down, the path is the same as every other strategy on the platform. The backtest scores them out of 350 points across profitability, risk, consistency and robustness, and auto-rejects results that fail its thresholds — negative expectancy, profit factor below 1.0, drawdown over 30 percent, or a weak overfitting, walk-forward or Monte Carlo score. Overfitting matters especially here, because frameworks with this many parameters are unusually easy to tune into a beautiful, meaningless curve.

Paper trading comes next, on live prices with virtual capital, and the drift between paper and backtest tells you how much the simulation flattered you. Real execution, if you ever choose it, happens only on crypto and only through your own exchange account with your own API keys. AKQX holds no funds and makes no recommendation at any point in that sequence. If you want to see the volume behind a liquidity sweep while you are forming a rule, the footprint chart is the place to look.

Questions about ict and smc algo builder

Related pages