What it is
DefiLlama is the open analytics layer that much of the crypto industry uses as a reference: on-chain TVL, volumes, fees, revenue, yields, stablecoins and fundraising, built from open-source adapters rather than vendor feeds. Its own framing is that you can analyse thousands of protocols across hundreds of chains, and prediction markets sit inside that as a first-class category page rather than a footnote.
Three surfaces matter to a reader of this directory. The prediction-market category page, which ranks venues against each other. Individual protocol pages, which carry the breakdowns and the history. And the API — a free public one plus a paid one — which is how most of the numbers elsewhere on the internet are actually fetched.
Why a prediction-market reader cares
Almost every other tool in this directory answers one question about one venue, or answers a cross-venue question about prices. DefiLlama answers the questions that sit underneath both: how much money is actually locked in this venue, how much is changing hands over the last day, week and month, how much users are paying in fees, and how much of that the protocol keeps. That is the difference between a market that looks busy and a market that is worth building on.
Its coverage of this sector is also wider than most people assume. It tracks venues that other on-chain tools ignore, including centralized ones, and it holds history rather than only a current snapshot, so a venue's fee take can be read across quarters rather than inferred from a single day.
The prediction-market page, and what it measures
defillama.com/protocols/prediction-market is a single ranking table with chain filters and a toggle for whether liquid staking and double-counted value are included in the totals. Each row carries TVL, prediction volume, fees and revenue over 7-day, 30-day and 24-hour windows, plus a market-cap-to-TVL ratio where one exists; the whole table exports as CSV or PNG. Parent protocols expand to their child venues rather than being double-counted as separate rows, and each venue links through to its own page.
On the review date the table listed roughly 120 tracked venues, with the largest 7-day prediction volume coming from a venue that has no TVL at all — which is the next point.
Off-chain venues are in scope, so TVL is not the whole story
DefiLlama's chain breakdowns include a line called Off Chain, and it is not a rounding artefact: it is where centralized and hybrid venues are tracked. Kalshi's page shows no TVL, because a regulated exchange holds no user coins in smart contracts, while still showing tens of billions in monthly notional volume and an open-interest figure in the billions. Polymarket splits the same way, with a Polygon leg and an off-chain leg across fees, revenue and volume.
The practical consequence is that ranking prediction venues by TVL — the default table order — measures something close to the opposite of what a trader cares about. A category leader by volume can sit at the bottom of a TVL column, and the venue with the deepest liquidity can have no contracts to lock anything in. Read the volume, fee and revenue columns before the TVL one.
Every number links to the code that produces it
This is the property that separates DefiLlama from a data vendor. Each protocol page carries a methodology block, and each metric in it links to the adapter file behind it on GitHub — the fee formula, the contract addresses, the accounting decisions, all readable. Polymarket's taker fee line, for instance, documents the formula used and the dates its parameters changed, code included; Kalshi's volume line points to a single adapter file that defines what counts as a trade.
Those repositories are actively maintained in public: the TVL adapters repo alone carries tens of thousands of commits, with merged contributions landing hours before this page was written. Coverage therefore grows by pull request, and the methodology is auditable rather than asserted — which also means a gap in coverage is usually the absence of an adapter rather than a statement about a venue.
It measures venues, not markets
Worth stating plainly, because it is the most common misunderstanding: DefiLlama tracks protocols, not individual contracts. You will not find a specific election market's odds, an order book, or the price of a single contract here. For that you need the venue's own API — Polymarket's CLOB and Gamma interfaces, or Kalshi's exchange API — or a cross-venue price feed such as OddsPapi. DefiLlama is the layer that tells you whether a venue is worth pulling from in the first place.
Coverage wider than DeFi
The same site tracks stablecoins, yield pools with a verified-APY view of what a holder actually received, token unlocks, bridges, ETF flows, digital asset treasuries, equities and pre-IPO valuations. For someone building a prediction-market product, the immediately useful neighbours are the fee and revenue history, the funding-round records that show which venues are still being financed, the hacks database that records incidents at prediction venues as they happen, and the unlocks calendar for tokens whose supply is about to change.
Where it sits in this directory
Think of the directory in layers. OddsPapi supplies cross-venue prices so one venue's quote can be judged against a sharp reference. Venue APIs — Kalshi's, Polymarket's — supply the market-level detail and execution. DefiLlama sits above the venues as the measurement layer: what each one holds, trades, earns and has raised over time. It is the tool you use to decide which venues deserve your attention, not the tool you use to trade them.
Figures that move under you
Every headline number on DefiLlama is live, and the two category pages read minutes apart during this review already disagreed on category TVL and 7-day volume — which is normal for data that refreshes continuously, not a fault, but a reason never to quote a DefiLlama total in copy that will be read tomorrow. The coverage figures have the same property: the site markets thousands of protocols across hundreds of chains, while the prediction-market category page itself counted around 120 venues with its own TVL total. And two of DefiLlama's published figures do not agree with each other at all — see the method notes below.





































