Grok Super Bot
Waiting for a saved paired comparison.
Recorded experiment / simulated money
One month. Two Super Bots.
A comparison we can inspect.
Grok and OpenAI each have a paired paper comparison: one portfolio follows the shared rulebooks, while the other also uses the trial’s evidence-based selection and sizing. The clock begins at explicit activation.
Loading the latest saved report…
The trial window
Waiting for verified activation and end times.
Thirty consecutive days from actual activation. Pausing does not restart the clock or restore the API allowance. When the trial pauses or ends, existing positions continue through their normal exits and saved history remains available.
One shared API allowance
Extra trial calls for both bots share a cap of $5 per New York trading day and $150 for the whole trial. Existing system costs are separate.
- Reserved this New York day
- —
- Reserved for the whole trial
- —
- Cost estimated from reported usage
- —
Missing provider usage reports remain unknown.
A reservation covers the bounded maximum cost before a call. It stays charged to the allowance after failures, missing usage or a smaller bill. Usage-based cost estimates are not provider invoices. API costs are shown separately from paper trading P&L.
Paired paper portfolios
This comparison isolates only the trial’s code selection and sizing. Both sides share evolving AI rulebooks. A live-mirror source receives extra trial research only with separately recorded owner consent. Experimental position sizing remains paper-only. It is not a randomized trial and does not measure the total benefit of AI.
OpenAI Super Bot
Waiting for a saved paired comparison.
What these numbers can tell us
Gross paper equity includes open positions. Realized P&L after modeled trading costs includes recorded trial exits; a dash means that figure is unavailable. Positions carried into the trial can contribute to both portfolios’ results. Previous closed trades are excluded from the trial’s exit count.
“Accepted” describes a saved review passing the trial’s checks. It does not establish that a hypothesis changed a trade or improved returns. The latest recorded review may be rejected or fail at the provider; that outcome is shown too.
The portfolios use the same starting equity and shared market inputs, but their later holdings can diverge. A positive difference over a short, correlated period is descriptive evidence, not proof of profitable learning or future performance.