Profit / loss (mark-to-market)
—
In the spirit of the 2025 experiments that put an LLM up against the markets, the question here is just one: given a verified, well-made memory, sinapsi.wiki, how well does a model read the future on markets where analysis, not execution speed, decides? We bet nothing. We take the questions from prediction markets, work out our own probability from the memory, and set it next to the price the market itself makes. The score only says which of the two read it better.
sinapsi.wiki is a verified wiki-memory system, queryable by AI too. These calls come from using that memory on a test domain, produced by an automated system with no human hand adjusting the conclusions. They are falsifiable, with source and date, and an automated system can be wrong. No real money is at stake: the score is simulated, in the style of the 2025 experiments. It is an analysis test, not financial advice and not an invitation to bet or trade. Contact: mario.brosco@salesmart.it.
€100,000 notionally placed on 12 July. The break-even line is the amount placed; above it means ahead.
For each contract: where the market price at entry, and where our estimate. The gap is the thesis. Our call runs in the direction of the gap. Sorted by divergence.