Updated July 2026

How our signals work

Every signal on this site answers one question: given the evidence, is the market's current price too low, too high, or about right? Here is exactly how we get to that answer, and what the answer does and does not mean.

A panel of AI analysts, not one model call

Each market on the board is analyzed by a panel of independent AI analysts. For a board of eight markets that is more than thirty separate analyst runs, each with its own brief and its own live research. No analyst sees another analyst's work while researching, which keeps the lenses from contaminating each other.

The panel works three lenses per market:

• Valuation and base rates. What do historical frequencies, polls, models, expert forecasts and comparable cases imply the true probability is? If a team priced at 58 cents has won this kind of match 70 percent of the time in comparable situations, that gap is the story.
• Risk. What could break the consensus? Upcoming catalysts and dates, tail scenarios, and resolution risk: the fine print of how the market actually settles, which is where careless traders get hurt.
• Track record and matchup. The history of the entities involved. Head-to-head records and tournament pedigree in sports, incumbent patterns and past behavior in politics, historical analogues in macro.

Synthesis: one honest verdict

A final synthesis pass reads all three lens reports and issues the signal: underpriced, fairly priced, or overpriced, always about the YES side at the current price. It also publishes a fair-value range and a confidence level. Two rules govern the synthesis:

• Efficient is a valid answer. Liquid markets are hard to beat. When the evidence lines up with the price, the verdict is fairly priced, and we say so rather than manufacturing an edge.
• No false precision. The fair value is a range, not a decimal point. Confidence is labeled low when evidence is thin or conflicting.

Where the numbers come from

Prices and volumes come from Polymarket's public market data at the time of analysis. Research facts come from the open web at analysis time and the strongest sources are cited on each signal page. Prices move constantly, so every signal shows the price it was built against. If the market has moved several points since, the verdict may no longer hold: check the date.

What a signal is not

A signal is a research summary, not a promise and not financial advice. An underpriced verdict means the evidence we found points to a higher probability than the market price implies. It does not mean the outcome will happen. A 70 percent favorite still loses three times out of ten, and a careful analysis can simply be wrong. Use signals as a starting point for your own thinking, size any position responsibly, and only trade where it is legal for you.