Prediction Market Analytics & Dashboards

What this category is, and what it is not

Nothing in this category runs a market. Every number on these pages is read off somebody else's venue: Polymarket's on-chain record on Polygon and its public APIs, Kalshi's market-data API, and the trade history each venue exposes. That single fact decides most of what follows. The product you are choosing is not a data source — it is an opinion about how often to poll one, how to normalise it, and what to compute on top.

So the first question about any tool here is not what it shows but what it reads and how often. A terminal that subscribes to a one-second WebSocket and refreshes underlying quotes every thirty seconds is a different instrument from an indexer that rebuilds positions in batches, even when both print a number in the same box. Ask which venues, which endpoints, and what the refresh interval is; if the tool does not say, that is an answer too.

Who is trading a market is a different question with a different reader, and it has its own section: wallet P&L, whale feeds, trader scores and address alerts are wallet and whale trackers.

Inside this category one job is separated out by a chip: the cross-venue gap scanners, which are bought for the one thing a volume dashboard is not. ArbBets, Arbitrack, CrossOdds and Polyarbitrage each take the same question priced on two venues and compute the difference, and what they disagree on is the step before the arithmetic: whether the two contracts are in fact the same question, and who checked. The head-to-head of two of them is about that step. What none of them computes is whether the gap is a trade — fees on both legs, two settlement currencies, two close times and two resolution sources are the subject of why the same contract costs two prices, and a scanner that prints a gap has done none of that normalising for you. Predxy goes the other way and folds the venues into one volume-weighted probability per question across fifteen of them, free to read with no account. On a question two venues carry, that number is mostly whichever of the two is busier, and its matching is "reviewed by a person when uncertain" with no score or list a reader could audit. The wider cross-venue collection is the flag over the whole catalogue, most of it feeds and SDKs that merely carry two venues; the chip here is the four products a reader types "arbitrage" to find.

This is the fastest-rotting section of the catalogue

Most of the products listed below carry a terms-of-service page written or last revised in 2026, and several of those dates are from the last few months. That is not a swipe at any of them; it is the shape of the market. Products at this age change their feature set, their price and sometimes their owner between one visit and the next; one of them kept serving a static site for eight months after the commit that paused the pipeline behind it.

Two practical consequences. First, dates on this site matter more here than anywhere else: every card carries the day its pricing and availability were last checked against the vendor's own pages, and a claim more than a few months old should be re-read rather than trusted. Second, a tool in this category is a poor thing to build a workflow on without an exit — prefer the ones that expose a documented API you could re-point, and keep your own copy of anything you would miss.

The three numbers vendors use interchangeably

Volume is notional traded over a window. Liquidity is what is resting on the book right now, usually within a stated band of the mid. Open interest is capital currently at risk in unresolved positions. They answer three different questions — how much changed hands, how much you could move without walking the price, and how much conviction is still outstanding — and a market can be high on one and low on the others. A market with large all-time volume and a thin book is the most common trap: the history says liquid, the order book says otherwise, and the number you needed was the second one.

Two more that get printed as if they were obvious. Realised versus unrealised P&L is the difference between money taken off the table and a mark on an open position that may still resolve to zero. And a win rate counts resolved positions without weighting them, so an account can be right nine times out of ten and still be down.

What none of them can see

A dashboard reads one venue's public record, which bounds what it can honestly claim. It cannot see a hedge held on an exchange it does not index, an off-venue position, or the relationship between two addresses that belong to the same desk. On venues that are not on a public chain, wallet-level analytics do not exist at all — Kalshi shows aggregate market data and no per-account positions, which is why cross-venue tools here are thorough on Polymarket and thin on everything else.

And a dashboard cannot see whether a market will resolve the way its price implies. A price is what the crowd will currently pay, not a measurement of the world. Nothing in this category changes that, and the tools that compute a "fair value" or a "mispricing" are producing a model's opinion, clearly labelled as such on their own pages and labelled as such on ours.

Read before you pick one

Four things decide the choice more than the feature list does.

Who publishes it. A domain that contains a venue's name usually does not belong to that venue. The terms page names the operator; read it.

Which venues, and how deep. Polymarket-only is the default here. Cross-venue coverage is usually cross-venue market data plus Polymarket-only wallet analytics, because that is what the public record allows.

Whether it can also trade. Most of these are read-only. The terminals listed here as secondary members will place orders once you hand over an exchange credential or link a wallet. That is a different risk profile and a different set of questions — where the credential is stored, and what the size caps are.

What it costs, and what the free tier actually includes. Free browsing with paid API access is the common shape, and the open-source entries cost nothing at any tier. At least two publish a price list with no free plan behind it.