Tenki

Tenki is an analytics tool. Tenki scans Kalshi and Polymarket contracts, flags those that look mispriced, and sends three to seven evidence-backed picks a day. It does not trade for you. A public graded track record is planned, and the service is waitlist-only for now.

Tenki is a research service for prediction market traders. It tracks the contracts it covers on Kalshi and Polymarket, compares where each market is trading with what the available evidence supports, and surfaces the ones where the two appear to diverge. The product is positioned as analysis rather than brokerage: it does not place trades, and the user decides whether and how large to size a position on the exchange itself. Each day Tenki delivers a short list of three to seven picks. A pick lays out the market, the evidence behind the read, the size of the gap the team believes exists, and the conditions that would make the read wrong. That last element matters for traders who want to challenge a thesis rather than accept it. The reasoning trail is meant to be checked, so a user with a data background can pull on any thread and form an independent view. Both exchanges appear in a single feed, which removes the need to compare across separate tabs. The audience is broad but specific: part-time traders without hours to spend hunting for edge, regular Polymarket and Kalshi users who want a second opinion before committing money, analytically minded people who prefer a starting point to a black box, and cautious newcomers who would rather study a short curated list than a stream of every open market. Tenki stresses transparency about performance. Once live, every pick is meant to be settled, graded, dated and published, with misses left up beside hits. The site is explicit that no one can guarantee winners and that the output is research, not gambling advice or risk guidance. At the time of the source page, the service was not open to the public. Interested users join a waitlist through a short questionnaire and are promised first access plus a two-week free trial at launch. Among alternatives, it sits between raw market browsing and self-directed research: less hands-on than building a model yourself, but more evidence-led than a simple signal feed.