Price Oracle

From Scope GitHub - hubbleprotocol/scope: 🎯 Scope is a Solana price oracle aggregator - a price oracle aggregator

pub enum OracleType {
    Pyth = 0,
    SwitchboardV1 = 1,
    SwitchboardV2 = 2,
    /// Deprecated (formerly YiToken)
    // Do not remove - breaks the typescript idl codegen
    DeprecatedPlaceholder = 3,
    /// Solend tokens
    CToken = 4,
    /// SPL Stake Pool token (like scnSol)
    SplStake = 5,
    /// KTokens from Kamino
    KToken = 6,
    /// Pyth Exponentially-Weighted Moving Average
    PythEMA = 7,
    /// MSOL Stake Pool token
    MsolStake = 8,
}

 

Solend - CToken

Solend 基于 solana-program-library/token-lending 开发。

CToken state account 实现了 `Pack` (并不是用的 anchor) https://github.com/solendprotocol/solana-program-library/blob/86746738e3c8289db6c18d2a0f2f6b421c74a2f1/token-lending/sdk/src/state/reserve.rs#L1128

获取CToken价格时，应首先手动刷新利率累积

reserve.accrue_interest(clock.slot).is_ok()

 

pyth

sdk: GitHub - pyth-network/pyth-sdk-rs: Pyth SDK in Rust

load price info account: pyth_sdk_solana::state::load_price_account 会检查 data 的 长度>= 0xCF0 bytes / magic / version / type.

返回 price struct：

Price {
      price: self.price,
      conf:  self.conf,
      expo:  self.expo,
  }

conf https://docs.pyth.network/documentation/solana-price-feeds/best-practices#confidence-intervals 置信区间，他不是百分比，是标准差的值大小。正常情况下应低于价格的1%

Pyth represents these possibly-different prices by giving its users a probability distribution over price instead of just a single price. Pyth models the price according to a Laplace distribution centered on the Pyth aggregate price with a standard deviation equal to the confidence interval (the scale parameter b of the Laplace distribution is equal to the standard deviation divided by the square root of 2). The Laplace distribution contains ~95% of the probability mass within ~2.12 standard deviations (~3 times the scale parameter). If markets are behaving normally, then the confidence interval will be tight -- typically much less than 1% of the price -- and the Laplace distribution will be highly peaked. However, at unusual times, the confidence interval can widen out dramatically.

pyth-ema

https://docs.pyth.network/documentation/how-pyth-works/ema-price-aggregation

for 指数加权平均价格 exponentially-weighted moving average (EMA) price ，其实就是 TWAP 的一种算法变体。因为 solana 的网络特性，pyth 会高频地推送price，这就导致传统的 TWAP 算法在数据采样上会浪费大量资源在时间窗口早期的数据上。 pyth 采用 EMA 算法使临近的数据在 TWAP 采样中有更高的权重。