The Inner Circle Trader’s concepts, and his 2022 model tested on Nasdaq

One idea runs through all of it. Price, he says, is delivered by an algorithm (the “interbank price delivery algorithm”, IPDA) that travels between two kinds of places: liquidity, the stop orders resting above old highs and below old lows, and imbalances, price ranges that were crossed in one direction only. Before a real move it first runs the stops on the other side, then moves away fast in the true direction. Every model he teaches is a version of this chain:

The 2022 Mentorship (41 free episodes) is where he is most mechanical. He teaches it on Nasdaq and S&P futures, says there’s absolutely nothing else that you need to know about it and that every single trading day this pattern forms. Here is each rule as he states it, and how I coded it. A sell is shown; a buy is the mirror image.

The pattern is common, as he says: an order on of mornings ( of ). Only of filled trades win, because the target is usually far away (median the stop). The total is , per setup, with a 95% range from to . The first period (to Sep 2024) lost per setup and the later one made ; with these swings neither is distinguishable from zero.

To see whether his choice of direction matters, I replayed the same moments with the same stop and target sizes 20 times, picking the side with a coin flip. The grey band is the range of those 20 runs.

Each row changes one thing and shows R per setup with its 95% range. Two of his ingredients do something. Requiring the stop hunt first is worth about per setup over taking any shift with a gap. His choice of side beats of 20 coin-flip runs. Neither is big enough to carry the costs on its own.

The bias is what he says matters most, and he never defines it mechanically. The proxies I could code add little. Following the previous weekly candle looks best (), but it mostly picks longs in a market that nearly doubled, its range still includes zero, and it is the best of 17 tradable versions tried, which is how luck usually looks. Even a perfect bias, knowing each morning’s direction in advance, gets only per setup out of these entries.

On the 1-minute chart the stops are small: median against a typical spread of . Before costs the base makes per setup (95% range to ); the CFD spread takes that to , and futures costs leave . The smallest stops do worst, as cost eats a bigger share of them. Moving to a 5-minute chart gives bigger stops and lower costs per R, but no more edge ( per setup).

He says the 2022 model is for index futures, and that his forex ideas (Asian range and the like) don’t apply to indices. I still ran the same rules on nine forex pairs and gold, in the New York and London windows, on the 1-minute chart and on the 5-minute chart, since his forex lessons use bigger charts and stops.

The model as taught has no reliable edge on his own market. At futures costs it is about break-even, per setup, and the 95% range runs from a clear loss () to a clear win (). On a retail CFD it loses slightly. Years with 50 or more setups range from to per setup.

He is right about some mechanics. Stop hunts before reversals are real enough that requiring one improves the result, and his time windows matter: the same rules lose per setup in the London window and in the afternoon.

The part that would have to carry the edge is the part he doesn’t define. Everything that is mechanical here nets out near zero; whatever extra a skilled ICT trader gets has to come from the discretionary read of bias and “obvious” levels, and even a perfect daily bias adds only about a quarter of an R per setup to these entries. Without a track record, there is no evidence that the read exists.

What to take from him: the vocabulary (liquidity above and below obvious levels, gaps, displacement), the time windows, and the risk rules. Treat the models as a way to describe charts, not as a tested system.

• Data: Orbi’s US100 CFD on 1-minute bid/ask bars with the real per-bar spread, not NQ futures ticks. The export has no Mar–Dec 2022 and Mar–Dec 2023. Within a minute the order of high and low is unknown; scoring it the optimistic way moves the result by less than 0.01R.
• Discretion: his “obvious” swings and “energetic” displacement became a 3-candle swing and the gap itself. A trader skipping ugly days could do better or worse.
• Scope: one trade a morning, no order blocks, no SMT, no trailing stops. Only the 2022 model was tested; OTE, the Judas swing, the Silver Bullet and the 2023–2026 models are not.
• Transcripts: 239 of 721 videos so far (the 2016–17 core content, Charter models, the 2022 Mentorship and the forex series). The rest are still downloading.

Code and numbers: ICT/ict_2022_model.py, ICT/ict_2022_study.py, ICT/results/ict_2022_study.json in the BACKTESTTRADING repo.