ArdenAlgoNine objections to ICT — and where we agree
We build a mechanical detector for ICT/SMC setups and publish every result it produces. That makes us an unusual party to this argument: we have no guru to defend and no win rate to sell, and the research this engine grew out of already closed its first signal family with a "no edge" verdict — internal and unpublished, which is precisely why the ICT test is being run in public this time. So here is the skeptic's case, objection by objection, with our honest answer to each. Five of the nine we simply accept.
1. "ICT can't be falsified — every loss becomes 'you drew it wrong'."
We agree, for discretionary ICT. A method whose rules live in a person's judgement cannot be tested, and therefore cannot be shown to work. That is the whole reason our engine exists: every concept we use has one written definition with numeric thresholds, evaluated on closed bars, the same way every time. The consequence cuts both ways — our numbers can be wrong, publicly, in a way a guru's never can. We think that is the point.
2. "The 'institutional order flow' story is fiction."
We don't defend the story. Whether a fair value gap is a footprint of institutions or just a fast candle is a narrative question we cannot answer and do not try to. We measure one thing: after a mechanically defined event, what did price do, net of modeled costs, compared with entering at random? The mechanism folklore is not an input to that measurement.
3. "It's supply and demand with new names."
Largely true. Order blocks look like supply/demand zones; liquidity sweeps look like stop runs; a market-structure shift looks like a breakout with a filter. We use the ICT vocabulary because that is what people search for and what the companion indicator draws, but nothing in our evaluation depends on the names. If the same definitions under older names produced the same ledger, we would be equally happy.
4. "You only hear from the winners."
Yes — and it applies to us too, unless we prevent it. Survivorship is not a flaw in ICT; it is a flaw in how trading results are shared. Our answer is structural: an append-only record written at signal time, before the outcome, losses included, plus a random-entry baseline and pre-set sample thresholds below which we label results as descriptive. The first three closed trades in our ledger were all stops. They are still there.
5. "Every number a vendor publishes is an advertisement."
Partly fair, so we declare the conflict instead of pretending it away. We intend to sell a subscription. That is stated on the page where the numbers live. The mitigation is not our sincerity; it is that the data is machine-readable and can be checked by anyone, and that we committed in advance to publishing the evaluation either way. "Trust me" is an advertisement. "Check me" is the opposite.
6. "Gross results are not results."
Agreed, without reservation. A setup that is profitable before fees, slippage and funding and unprofitable after them is unprofitable. Every figure in our ledger is shown net, with the cost model documented separately, and gross beside it so the gap is visible. Most of what is called "backtesting" in this space omits this step, and that single omission explains a large share of the optimism.
7. "The man who invented it made his money teaching, not trading."
Not our argument to have. We have no view on the person and we do not test him; we test patterns that carry his vocabulary. Whether he trades, and how well, changes nothing about whether a mechanical sweep → structure-shift → gap sequence has an edge after costs on Binance perpetuals. We would rather spend the effort on that question.
8. "Screenshots prove nothing."
Correct, which is why there are none here. Not on this site, not in our indicator descriptions, not in our posts. A screenshot is a claim without a denominator. A ledger row with an id, a timestamp, an invalidation level and a later outcome is a claim with one.
9. "It's all curve-fit in hindsight."
The risk is real, and only a forward record answers it. Thresholds chosen after seeing the data will look good on that data. Ours were written down before the ledger existed and have not been changed; the ledger itself is forward-only, so there is no version of it that was fit to anything. If the forward record fails the pre-registered test, we will say so on the front page. We have recorded a "no edge" verdict before, in the internal research that preceded this engine; the difference this time is that the test and its outcome are public.
What we are not saying
We are not saying ICT works. We are saying that the objections above are mostly right about how ICT is usually sold, and that the only honest rebuttal is a record that could embarrass us. Ours is at /ledger, the definitions are at /methodology, and the test that will settle the question for our engine is described at /evaluation. Read those before reading anything we say about results — including this page.
Are we affiliated with ICT / Inner Circle Trader? No. We use the vocabulary descriptively; see the disclosure on /policies/risk.
Joining locks the launch rate before the evaluation verdict is out — and gets you two mails, ever: launch + the verdict, whichever way it goes.