Nothing here is taken on trust.
Arctix is built on a single refusal: nothing is believed because it is plausible. A thesis is a hypothesis until a window it never saw agrees with it, and a result that holds only in the data that produced it is treated as a failure, not a finding.
The operation is deliberately small and deliberately cold. One analyst sets the standard; an engine does the work of testing against it — without fatigue, without preference, and without the ability to argue with its own verdict. Of 20,612 strategies it has written, 17,130 were rejected outright. That ratio is the character of the place, not an embarrassment about it.
A strategy that works is one that survived a bar written down before it was tested. Explanations are cheap and arrive after the fact. The engine is not permitted to score its own work, and the gates are mandatory rather than weighted, so a strong result cannot be traded against a failed test.
Every figure on this site is read from the running system when the page is served. Where a number cannot be resolved from live state, no number is printed. This site has been wrong about its own thresholds before; the correction was structural rather than editorial, which is why it cannot recur.
Decisions are evaluated, not outcomes. A profitable trade from a broken process is a failure that has not arrived yet, and a rejected strategy is the mechanism working. Discipline here is a property of the machinery, not a resolution someone has to keep.
Who is accountable.
Arctix Capital is run by one person. It manages no outside capital, is not registered as an investment adviser, and deploys strategies to a paper brokerage account. "Fund", "mandate" and "track record" on this site describe a research operation and its process record — not a pooled investment vehicle, and not an offer. See disclosures.