- Polyseer writes a long research report on one Kalshi or Polymarket market: a probability with a range and confidence level, compared with the market price, plus evidence for and against, base rates, catalysts and risks.
- The hosted site offers "$10 free" on sign-in with Valyu and bills Valyu credits. The project's README lists $0.10 to $15 per run, depending on research depth.
- Kalshi is supported directly: it reads the market from Kalshi's API.
- The code is public on GitHub. The README says MIT, but the repository has no LICENSE file, so the terms of reuse are less clear than that suggests. Last checked October 15, 2026.
Disclosure: PredictionsMarket.com earns referral commissions from Kalshi when readers sign up through our links (see our affiliate disclosure). Polyseer is not a partner and did not pay for or review this page. Not financial advice. Trading involves risk.
This Polyseer review covers what the reports contain, how the tool works under the hood, what a run costs, how it handles Kalshi, what can be verified and who it suits. It is based on the hosted site at polyseer.xyz and a full read of its public code repository, both on October 15, 2026. We did not sign in or run a report. For other tools, see our guide to the best AI tools for Kalshi.
Polyseer review: what a report contains
You paste a Polymarket or Kalshi market link. Polyseer reads the live market (question, prices, close date and resolution rules) and hands it to Valyu DeepResearch, an outside research agent that searches news, primary sources, data releases and related markets. According to the README, the finished report includes:
- A probability for the outcome, with a plausible range and a confidence level, set against the market price.
- A suggested position, or none when the gap to the market price sits inside the tool's own uncertainty.
- Evidence for and against, ranked by weight, each with its source.
- The base rate, upcoming catalysts, risks, and what would change the forecast.
- A PDF report, a CSV file of every factor affecting the market, and an email when the research finishes.
The hosted homepage lists trending markets from both venues with a research button on each. On the day we checked it showed 8 Polymarket and 4 Kalshi markets.
How it works
The current README says Polyseer is "a thin Next.js app around one external research job" and that "there is no in-process LLM pipeline". The research runs inside Valyu's service, and the README does not name the language model used.
The research instructions are in the repository, in src/lib/research/prompt.ts. They tell the agent to start from a reference class and its base rate before reading the news, to look actively for the strongest evidence against its own emerging view, to give a range rather than a single number, and to suggest no position when the gap to the market price is inside its uncertainty band.
That is a change from earlier versions. Until September 14, 2026, the README described a pipeline of separate planner, researcher, critic, analyst and reporter agents, with research run separately for and against. That design is still in the repository's history, at commit 7843c46 and earlier. The for-and-against idea now lives as an instruction inside one research prompt.
Polyseer cost and free credit
There are two ways to use it. You can run the code yourself with your own Valyu API key, and research is billed to that key. Or you can use the hosted site, where you sign in with a Valyu account and runs draw on that account's credits.
| Effort | Typical time | Cost per run |
|---|---|---|
| Fast (default) | About 5 minutes | $0.10 |
| Standard | 10 to 20 minutes | $0.50 |
| Deep | 30 to 60 minutes | $2.50 |
| Max | Up to 2 hours | $15.00 |
The README adds $0.10 per run for the second file deliverable. The hosted site has no pricing page, and its homepage shows the default effort as "Fast" without a price. Its terms say credit costs vary with analysis complexity and third-party API use, so the README table is the best guide available, not a price list from the hosted service.
Free credit
The hosted homepage shows "$10 free" next to the Valyu sign-in button. The site does not explain the terms of that credit, and we did not sign in to find out. Running the code yourself costs nothing from Polyseer, but Valyu usage is still billed.
Kalshi support
Kalshi is supported directly. The README's architecture section says Polyseer fetches Kalshi markets through Kalshi's trade API and Polymarket markets through Polymarket's own APIs, and the hosted homepage lists Kalshi markets alongside Polymarket ones. Our Kalshi review explains how Kalshi markets resolve, which matters when you read any report's view on the rules.
What can and cannot be verified
- The method is open to inspection. Because the code and prompt are public, anyone can read how a report is requested and what the agent is told to do. That is rare among the tools we reviewed.
- No accuracy record. The README describes its probability as calibrated but links to no accuracy record, backtest or scored history. We found none in the repository or linked from the hosted homepage; research history needs sign-in.
- The licence is unclear. The README says the project is licensed under MIT and points to a LICENSE file, but no LICENSE file exists in the repository at its latest commit or anywhere in its history, and the package file has no licence field. The code is public; whether you may reuse it on MIT terms is less clear than a standard MIT project.
- Popularity. We could not read the repository's GitHub star count from our environment, so we do not quote one.
The README carries the disclaimer that Polyseer is "NOT FINANCIAL ADVICE" and provides analysis for entertainment and research purposes.
Who runs Polyseer
The hosted site's terms, effective December 8, 2025, say Polyseer is a service operated by Valyu.ai Ltd, a company registered in England and Wales. The research itself runs on Valyu DeepResearch. The repository belongs to the GitHub user yorkeccak, and its latest commit when we checked was dated October 14, 2026. We found no affiliate program.
Who it suits
Polyseer suits someone who wants a long, sourced research report on a single market, with the case on both sides set out, and who values being able to read exactly how the report was produced. Self-hosting suits developers who already have a Valyu key. It suits people less well if they want a quick read on many markets, a published accuracy record, or a fixed monthly price; and the deeper research levels can take up to two hours per market.
Use any report as one input. Read the market's rules on the venue before trading, and check costs with our fee calculator. Live prices are on our markets page.
Frequently asked questions
Is Polyseer free?
The hosted site shows "$10 free" when you sign in with Valyu, without spelling out the terms. After that, runs bill Valyu credits; the README lists $0.10 to $15 per run depending on depth. Running the public code yourself costs nothing from Polyseer, but Valyu usage is billed to your own key.
Does Polyseer work with Kalshi?
Yes. Paste a Kalshi market link and it reads the market from Kalshi's trade API.
Is Polyseer open source?
The code is public on GitHub. The README says MIT, but the repository has never contained a LICENSE file, so the reuse terms are not as clear as for a standard MIT project.
Who makes Polyseer?
The hosted service is operated by Valyu.ai Ltd, registered in England and Wales, according to its terms.