Oracle tokens, live on the bubble chart

Picture a smart contract that pays out insurance if it rains in London tomorrow. The code is simple: if rain, pay. The problem is that the contract has no idea whether it rained. A blockchain can only see what's already on the blockchain, and the weather isn't.

Swap rain for the price of ETH and you have the problem behind almost every lending market and derivatives exchange in crypto, and most stablecoins. Something has to carry outside information onto the chain, and that something is an oracle. This page tracks the tokens of the networks that do that job, sized and colored by how they're trading right now. They're among the least glamorous coins in crypto and among the most load-bearing.

The oracle problem nobody has fully solved

The oracle problem is simple to state: a blockchain is trustworthy because nobody has to be trusted, but the moment it needs outside data, somebody has to be trusted to report it. A lending protocol with flawless code still fails if the price it reads is wrong.

There's no clean fix, only better and worse compromises. Most oracle networks spread the trust out. Instead of one data source, they ask many independent node operators to fetch prices from many exchanges, then take something like a median, so one bad report can't move the result. That makes lying expensive rather than impossible. It's a real improvement, and it's still trust, just distributed.

Oddly, "oracle crypto" searches catch a fair number of people looking for Oracle the database company, which has sold its own enterprise blockchain product for years. The name comes from the ancient oracles who answered questions nobody else could, which is a generous description of a price feed. Not that anyone in crypto was going to resist the reference.

What a blockchain oracle does all day

Mostly, it publishes prices. A blockchain oracle network watches markets off-chain and writes an updated price on-chain whenever the price moves past a set threshold (often around half a percent for major pairs, though it varies by feed and I've never seen a complete list I'd trust), or after a fixed time passes if nothing moves. That's the push model, and it's how Chainlink's feeds have worked since its mainnet launched in May 2019.

The newer pull model flips it around. Pyth, which started on Solana, has trading firms and exchanges sign prices constantly off-chain, and an app pulls the latest signed price on-chain only when it needs one, paying for that update itself. It's cheaper for slow-moving markets and faster for trading apps, which is why it caught on with derivatives platforms.

Whoever wins that design argument, the users are the same. Nearly every lending market behind the defi tokens on this site reads an oracle to decide when a loan is underwater.

Every perp dex uses one to set the mark price that triggers liquidations. That's the number to watch when you run a leveraged position through a crypto futures profit calculator: the exchange liquidates you on its oracle price, not on whatever your favorite exchange shows.

Oracles also check reserves for rwa tokens, confirming that the gold or treasuries behind a token actually exist.

When the price feed lies: oracle manipulation attacks

An oracle manipulation attack doesn't hack the oracle. It feeds it a fake price by moving a thin market the oracle trusts, then uses the fake price before anyone notices.

The early versions were crude. In February 2020 an attacker used flash loans against bZx to push prices on decentralized exchanges and walked away with roughly $1 million across two attacks. In April 2022 Inverse Finance lost about $15.6 million after someone pumped the price of its INV token on a single thin pool that its oracle relied on.

Mango Markets in October 2022 was the one everyone remembers. A trader pushed the price of the MNGO token up several times over on the exchanges Mango's oracle watched, then borrowed against the inflated collateral and drained about $110 million. He then argued publicly that it was a legal trading strategy (the courts have gone back and forth on that question ever since). Whatever the legal answer, the oracle did exactly what it was built to do. It reported a real price from a market that had been bent.

The lesson most protocols took: never trust a price from a market you could move with a borrowed afternoon of capital. That's why serious lending markets now use oracles that aggregate many deep markets and refuse to list tokens with thin liquidity.

Oracle tokens on this chart

These are the oracle crypto coins that usually appear as bubbles here. The column worth reading is the last one, because an oracle token is only as valuable as the money that depends on its data.

LINK is usually the biggest bubble by a wide margin, and that matches reality: Chainlink feeds secure a large majority of the value that depends on oracles, by DefiLlama's count. Holding LINK can earn staking rewards for backing node operators, and a crypto staking calculator shows what that rate compounds to, though LINK's price will matter more than the yield.

The smaller oracle tokens trade on thinner markets, which is a little ironic for a category whose whole lesson is to distrust thin markets. Earlier I said lying to an oracle is expensive. For the small tokens in this table, moving their own price is cheap, and their bubbles show it.

Reading oracle tokens on the chart

Oracle tokens rarely lead a market move. They tend to follow DeFi activity with a lag: when lending and derivatives volume grows, the networks feeding them get paid more, and the tokens eventually notice. If oracle bubbles are green while the crypto bubble view on the home page is flat, it's usually news specific to one network, a big integration or a partnership with a bank, and those moves tend to fade unless usage follows.

For LINK in particular, a bitcoin comparison over 1Y says more than the bubble does. LINK has spent long stretches tracking the market rather than its own adoption, and seeing that correlation on a chart is a useful corrective to the integration headlines. If you trade the smaller oracle tokens, run the numbers through a crypto trading profit calculator with fees and slippage included; their spreads can eat a small gain whole.

Chainlink feeds exist on most of the networks you'd find on a layer 1 crypto list, and on plenty of layer 2s too. That reach is hard for any rival to match.

Chainlink alternatives, and why most protocols still pick Chainlink

Chainlink alternatives exist and some are good. Pyth is the strongest challenger, especially for trading apps that want fast, frequent prices and are happy to pay per update. RedStone has grown quickly with newer lending markets. Chronicle, which Maker (now Sky) spun out of its own oracle work in 2023, secures much of that protocol's collateral. API3 pitches first-party data, where the company producing the data runs the node itself instead of a middleman.

UMA works differently from all of them. Its optimistic oracle accepts an answer unless someone disputes it, and disputes are settled by UMA token holders voting. That suits questions without a price feed, which is why Polymarket uses it to resolve prediction markets. It also means a large enough holder can sway an outcome, and in March 2025 a disputed Polymarket market about a Ukraine minerals deal resolved in a way many traders said was plainly wrong. That's the oracle problem again, moved from prices to facts.

I used to think the oracle market would split evenly between a few big networks. It hasn't. Chainlink's lead has mostly held because switching oracles is risky and slow for a protocol holding billions, and nobody wants to be the team that changed its price feed the week before an exploit. The alternatives win where Chainlink is slow, expensive or absent, which is a real market, just a smaller one than the bubble sizes for some of these tokens suggest.