The @Pyth Network whitepaper proposes a paradigmatic shift in the landscape of oracles, as it essentially shifts away the third-party data model to a first-party model of data. The design option is not just an architectural style of choice, but a vital security and transparency innovation that reinvents the integrity of data feeds within the multi-chain decentralized finance (DeFi) landscape. Sourcing data directly to the institutional entities that own it, Pyth significantly reduces the risk of manipulation and provides an unprecedented amount of real time market transparency due to its unique application of confidence intervals.
The Excellence of the First-Party Data Sourcing.
Oracle networks Traditional oracle network (third-party models) are based on a layer of decentralized node operators, which access data through public application programming interfaces (APIs) on the centralized exchanges or data aggregators. This presents a number of trust assumptions and possible vulnerabilities:
Trusting the Middleman: DeFi protocols have to trust the third-party node operators, that they will faithfully execute their off-chain work without collusion and errors before publishing the result on-chain.
Delayed Data: Data being reported is not always the most recent price and is instead a price that is present on a public API, introducing a latency between the true market price and the on-chain reported price.
Source Reliability: The user is a passive consumer of the data and has no control over its source, rather than the third-party node, which they must trust to make their choices.
Pyth Network addresses these problems by establishing a network of first-party publishers - large world exchanges, proprietary trading companies and professional market makers.
Data Producers as Publishers: These companies are the publishers and creators of proprietary price data of high-fidelity. They witness the fastest and most granular data flows since they are actually engaged in carrying out trades and creating markets.
Aligned Incentives: Publishers will have an incentive to offer their most accurate and high-frequency data since they will be rewarded with PYTH tokens and have a reputation on the network that is publicly visible. Most importantly their incentives are consistent with market accuracy where faulty data is severely penalized by cutting their staked tokens.
Reduced Manipulation Risk: As Pyth aggregates has feeds in excess of a hundred different institutional sources in various venues (equities, crypto, FX), it becomes exponentially difficult to control the aggregated price by any single participant. To attack the Pyth feed an attacker would need to simultaneously manipulate the prices of a dozen or more independent, highly liquid trading venues, which is practically and economically unfeasible to accomplish.
Removing the third-party intermediary layer, Pyth not only makes the data faster, but inherently more reliable and more reflective of the actual market-wide price of an asset.
Openness to Uncertainty: The Confidence Interval.
The most important innovation is perhaps the introduction of a confidence interval being provided together with each price update as seen in the Pyth Whitepaper. Pith offers a price and a price uncertainty (e.g., BTC/USD = 68,000 + -50) as opposed to other oracles that only give a single price (e.g., BTC/USD = 68,000 = 68,000 ).
The Confidence Interval Reveals What.
Stability The confidence interval (sigma) is a statistical metric that is obtained in the aggregation procedure. It is used to measure the level of dispersion in prices recorded by all the reporting publishers of a particular asset.
High Uncertainty (Wide ), The wide confidence interval shows that the publisher prices are very different. This normally happens at a time of:
Extreme Volatility: Rapid and abrupt market events in which the prices of securities diverge between exchanges, temporarily.
Low Liquidity: In low liquid markets, small trades may result in wide bid/ask spreads, and the reported prices may change dramatically.
There is a tight confidence interval meaning that there is high consensuality among publishers which means that the market is liquid and stable.
Smart Contract Risk Management empowerment.
Confidence interval is an effective on-chain risk management tool, which third-party oracle models totally lack. The confidence interval can be programmed into smart contracts that are then able to read and respond to it with sophisticated and safer decision making:
Liquidation Protection: A lending protocol is able to modify its liquidation threshold. In the situation whereby the confidence interval of a collateral asset abruptly expands (high uncertainty), the protocol may be implemented to halt liquidations or raise the margin requirement to avoid false liquidations due to temporary and non-malicious price divergence.
Fee Adjustment: A decentralized exchange (DEX) is able to dynamically adjust the trading fees. When the confidence interval of a token pair expands, the DEX may charge a larger swap fee to cover the increased market risk which liquidity providers are taking.
Outlier Flagging: The protocols can give a basic safety check: when the confidence interval (sigma) is too big in comparison with the price (P), then the contract may temporarily discard the update and default to the previous stable price.
Pyth allows DeFi developers to create resilient and adaptive financial applications capable of responding to the turbulence of the real world market with institutional grade risk awareness by offering a probabilistic truth: the most probable price, and an understanding of how likely it is. This open and transparent layer of data is the building block of the mission of Pyth to become the safe and reliable price layer of the world of financial markets.