You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Confidence Intervals of Pyth Network's Prices Are Ignored
Summary
The prices fetched by the Pyth network come with a degree of uncertainty which is expressed as a confidence interval around the given price values. Considering a provided price p, its confidence interval σ is roughly the standard deviation of the price's probability distribution. The [official documentation of the Pyth Price Feeds](https://docs.pyth.network/documentation/pythnet-price-feeds/best-practices#confidence-intervals) recommends some ways in which this confidence interval can be utilized for enhanced security. For example, the protocol can compute the value σ / p to decide the level of the price's uncertainty and disallow user interaction with the system in case this value exceeds some threshold.
Currently, the protocol [completely ignores the confidence interval provided by the price feed. Consider utilizing the confidence interval provided by the Pyth price feed as recommended in the official documentation. This would help mitigate the possibility of users taking advantage of invalid prices.
The text was updated successfully, but these errors were encountered:
Mean Brown Barbel
Medium
Confidence Intervals of Pyth Network's Prices Are Ignored
Summary
The prices fetched by the Pyth network come with a degree of uncertainty which is expressed as a confidence interval around the given price values. Considering a provided price
p
, its confidence intervalσ
is roughly the standard deviation of the price's probability distribution. The [official documentation of the Pyth Price Feeds](https://docs.pyth.network/documentation/pythnet-price-feeds/best-practices#confidence-intervals) recommends some ways in which this confidence interval can be utilized for enhanced security. For example, the protocol can compute the valueσ / p
to decide the level of the price's uncertainty and disallow user interaction with the system in case this value exceeds some threshold.POC
Currently, the protocol [completely ignores the confidence interval provided by the price feed. Consider utilizing the confidence interval provided by the Pyth price feed as recommended in the official documentation. This would help mitigate the possibility of users taking advantage of invalid prices.
The text was updated successfully, but these errors were encountered: