Research / 03
Research
How the work is done is public. What the work found is not.
Variants evaluated
30
aggregate, all families
Rejected
26
a rejected idea is a result
Families closed
—
closed on evidence, with the reason recorded
Method, in outline
- Preregister the test. The falsification bar is fixed before data is touched, so the bar cannot be moved after seeing the outcome.
- Measure the data route as carefully as the idea. If the inputs cannot be trusted — stale, unnamed, or a field silently unparsed — the question is set aside rather than answered badly.
- Test against a control. A known-positive case must reproduce, or the instrument is treated as broken before any result is reported.
- Hold out. Validation happens on data the idea was not formed on.
- Report the failure. Rejections, inconclusive runs and self-inflicted measurement errors are recorded permanently and are not quietly retired.
What is deliberately not published, and why
Strategy logic, entry and exit rules, formulas, settings, candidate securities, candidate lists, execution details and per-strategy results stay private. Publishing them is not transparency; it is handing over the research. What can be stated honestly without them is stated: aggregate counts, evidence class, method, and the discipline of the process.
On evidence classes
Every figure carries one. Backtests are historical simulations and are labelled as such. A figure is never relabelled or promoted by describing it differently. Where a result was inconclusive, it says inconclusive rather than being dropped or rounded toward the hoped-for answer.
Current stage
What is true today
- No public token exists.
- No investor capital has been raised or is being solicited.
- No live offering exists.
- No live trading track record is presented here.
- Nothing on this site is an offer, a solicitation, or investment advice.