Building AI for financial markets is not a neutral act. Every design choice shifts behaviour. This is our public statement on what we will build, what we will not build, and the features we will kill the day their harm becomes clear.
The commitments
1. We will not ship a copy-trading network.
Copy trading industrialises overconfidence. It gives the appearance of expertise where there is often luck, selection bias, and sometimes outright incentive misalignment. Our engine is deliberately a reasoning tool, not a follow tool.
2. We will not auto-execute without a human in the loop.
We ship rule-based automation because that is pre-commitment. We do not ship open-ended auto-trading agents. The accountability for a trade belongs to a person.
3. We will not sell “guaranteed” anything.
No “guaranteed signals”. No “guaranteed stop-loss”. No “guaranteed returns”. If a competitor uses this language, they are either lying or ignorant.
4. We will not use dark patterns against our users.
No pop-ups to upgrade before close. No fake scarcity countdowns. No pre-ticked boxes. The “close account” button is one click deep.
5. We will not misuse user data to train models.
Your trading behaviour is yours. Aggregated anonymised telemetry informs engineering, and that is it.
6. We will not do regulatory arbitrage.
We refuse US customers because the regulatory framework is not clear there. We will not list under a flag-of-convenience jurisdiction.
7. We will not retaliate against critics.
Public criticism, even unfair criticism, is welcome. We will not sue reviewers. We reply publicly, on the record.
The uncomfortable ones
We will downsize signal volume when precision drops.
If rolling 7-day precision drops below 72%, we automatically cut published signal volume by 30% and ping every active user.
We will refund if we misled you.
If a user cancels within 30 days because our marketing created an expectation our product does not meet, we refund without argument.
We will shut features we built, when evidence demands it.
We shipped a social-leaderboard feature in beta week six. It increased average position size by 23% and increased losses by 18%. We shut it, publicly, in week nine.
Why publish this at all
A public commitment is harder to walk away from than a private one. If we ever cross one of these lines, call us on it. Email [email protected]. We will reply.
Want sharper reads like this one?
Subscribe to the OcuBit briefing — twice a month, no fluff.