What are the stop rules?

Conditions set before any result exists: if the whole portfolio falls further behind the S&P 500 than in all but about 1 in 100 of the one-year stretches it was tested on, if the growth-inflection or momentum-leader rule trails the market by a set margin, if a trading day passes without the AI's decision being made and traded, or if the AI's decisions fall too far behind the two rules'. The last sends the AI's decisions back to the owner for review.

Why publish them in advance?

A stop rule written after the results is not a rule. Fixing the conditions first, in public, is what makes the eventual verdict credible. The rules can be made stricter later, never looser.

When are they checked?

All five are recomputed every week from the portfolio's start on 5 October 2026. Rules 4 and 5 can return a verdict at any weekly check. Rule 1 is judged every quarter from the 252nd trading day after the start, about October 2027; Rule 2 needs 150 closed growth-inflection positions, about two years; Rule 3 needs three years. Until then they read as pending.

Stop rules · checked every week

The rules that would stop the strategy.

Set before any result existed and recomputed every week. A rule can be made stricter after results come in, never looser.

Rule 1 Not enough data yet

From a year after the portfolio's start, every quarter: if over the last 252 trading days it trails the S&P 500 by more than 25 percentage points — a shortfall reached in only about 1 in 100 of the one-year stretches it was tested on — the result is published as out of range. If that happens at two reviews in a row, the owner must set a new plan.
Rule 2 Pending 0 of 150 closed

Once 150 growth-inflection positions have closed: if on average they trailed the S&P 500 over their own dates by more than 4 percentage points, the growth-inflection rule is retired and the momentum-leader rule takes its place.
Rule 3 Pending

After three years: if the momentum-leader rule has trailed the S&P 500 by more than 29 percentage points over that time, it is retired and the growth-inflection rule takes its place.
Rule 4 Within range

Every trading day the AI must make its decision and the portfolio must trade it. A day without a valid decision, or a decision that was not traded, is logged as a deviation and shown here; on such a day the portfolio holds what it has.
Rule 5 Within range

The two fixed rules keep running alongside, and both they and the AI's decisions are valued the same way: at each day's closing prices, with the same trading costs, from the same start. If the AI's portfolio falls 15% or more behind the rules' portfolio, since the start or since the owner last reviewed it, the owner decides whether the AI keeps the decisions. Every forecast the AI makes is also scored against what happened: how well it ranked the stocks, and how well its stated odds matched the outcomes.

Why set them in advance

A verdict decided by rule, not by judgement.

The portfolio is tested forward, in public. Its stop rules were fixed before any result existed, so a weak run retires a rule, or sends the AI's decisions back to the owner, on the published terms. Results are on the portfolio page; how the portfolio is decided is on the methodology page.

The portfolio How it is decided Scoreboard

Common questions

What are the stop rules?: Conditions set before any result exists: if the whole portfolio falls further behind the S&P 500 than in all but about 1 in 100 of the one-year stretches it was tested on, if the growth-inflection or momentum-leader rule trails the market by a set margin, if a trading day passes without the AI's decision being made and traded, or if the AI's decisions fall too far behind the two rules'. The last sends the AI's decisions back to the owner for review.
Why publish them in advance?: A stop rule written after the results is not a rule. Fixing the conditions first, in public, is what makes the eventual verdict credible. The rules can be made stricter later, never looser.
When are they checked?: All five are recomputed every week from the portfolio's start on 5 October 2026. Rules 4 and 5 can return a verdict at any weekly check. Rule 1 is judged every quarter from the 252nd trading day after the start, about October 2027; Rule 2 needs 150 closed growth-inflection positions, about two years; Rule 3 needs three years. Until then they read as pending.