ArbiTrack

What it is

A single browser table that lines up the same event on Polymarket, Kalshi and PredictIt, prints each venue's price in a column, and computes an "Edge" — the gap between the highest and the lowest of the three — colour-coded at 3, 5 and 8 cents. Rows filter by politics, economics, crypto, sports, climate and other, and sort by edge, 24-hour volume or Polymarket price. The table refreshes itself every three minutes against a server-side cache the FAQ says is held for up to two minutes.

It is an information tool and says so: it places no order, and the FAQ tells you to buy on the cheapest venue and sell, or buy No, on the dearest yourself. Data comes from Polymarket's Gamma API, Kalshi's Trade API v2 and PredictIt's public market-data endpoint, fetched server-side.

Matching is described as "smart algorithmic matching" and nothing more. The FAQ's worked example is "Will the Fed cut rates in September?" on one venue against "Fed rate cut — September 2025 FOMC" on another; no method, score or rule comparison is published.

The operator is not named anywhere on the site. Contact is a Formspree form, and the privacy policy names Vercel for hosting.

Availability

No account, no email and no identity check — the privacy policy lists what is not collected, and a name is on that list. The free tier needs nothing at all. Paying needs a wallet holding USDC or USDT on Polygon, and nothing else.

There are no terms of service. On 30 September 2026 /terms and /terms.html both returned 404, and the footer links to the scanner, a blog, the FAQ, a contact form and a privacy policy. Nothing restricts who may use it, and nothing sets out what you are buying, what happens if the service stops mid-pass, or which law governs. The privacy policy was last updated on 8 April 2026.

Whether you may trade on what you read is each venue's question: Polymarket's international exchange does not serve US persons, while Kalshi and PredictIt do.

Pricing

Time passes, not a subscription — the site says there is no auto-renewal and no account:

• Free — five markets, with the edge column locked.
• 24 hours — 4 USD.
• 7 days — 9 USD.
• 30 days — 19 USD.
• 365 days — 149 USD, about 12.40 a month.

Payment is in USDC or USDT on Polygon by WalletConnect, a pasted transaction hash checked against Polygonscan, or a generated deposit address that opens access when the transfer lands. Access is a signed token in your browser's local storage, valid for the length of the pass; another device or a cleared browser means verifying the payment again. There is no refund policy, because there is no policy of any kind beyond privacy.

ArbiTrack charges no trading fee because it places no order; the fee fields are null for that reason.

Markets & resolution

Whatever the three venues list that the matcher pairs. On 30 September 2026 the market list the scanner loads held 493 markets — 331 from Polymarket, 230 from PredictIt and zero from Kalshi, with a flag saying Kalshi was enabled — of which 21 were matched across two venues. On that day the three-venue scanner was a two-venue scanner, and nothing on the page said so.

The largest edges that day were all Polymarket against PredictIt on elections, and several were around 30 cents: House control after the 2026 midterms, the Kansas Senate race, a named candidate for the 2027 French presidency. Gaps that size between two liquid venues on the same question are usually one of three things — two contracts that are not the same, a price read from a stale field, or the cost of moving money between venues with very different rules — and the table gives you no way to tell which. The two venues also settle differently: PredictIt in dollars under its own rules, Polymarket in pUSD, its USDC-backed collateral token since 28 April 2026, through an on-chain oracle. Why the same contract costs two prices lists what sits in such a gap.

ArbiTrack resolves and settles nothing.

Limitations

The edge is a price difference, not a margin. No fee is subtracted on any venue, and the FAQ does not say whether the price shown is a last trade, a midpoint or an ask. On PredictIt the cost is taken on the way out — 10 percent of profit on every winning position and 5 percent of every withdrawal, per the PredictIt card — so a pair that looks like 6 cents can be under water before the money leaves. What a trade actually costs does the conversion.

The risks that remain are all yours. The site's own disclaimer names platform risk, liquidity risk and oracle resolution disputes. Add resolution wording that differs between the matched markets, capital held on two venues until each resolves, PredictIt's 30-day wait before a first withdrawal, and whether you may hold both legs at all.

The landing-page table is an illustration. The four sample rows under "updated every 3 minutes" are written into the page's HTML, not loaded, and one of them is a market on a model release dated before August 2026. Judge the product by the scanner page, not by the sample.

No book, no API, no export. One price per venue per event, nothing on depth, nothing to call from code, and nothing to download. A gap you cannot size is a gap you cannot plan a pair around.

No terms and no named operator. For a product sold by prepaid pass in stablecoins, that is the fact to weigh first.

Alternatives

CrossOdds covers Kalshi and Polymarket only, publishes its matching method, a confidence score and caveats per pair, nets out estimated fees, and costs 20 USD a month with an API. ArbBets walks full book depth across four venues from 149 USD a month. Oddpool has free dashboards over Kalshi and Polymarket with a price-spread view. If PredictIt is the reason you are here, its own card sets out the fee and position limits that decide what a gap there is worth, and matching a question across venues is the procedure for checking a pair yourself.