when markets are open, the oracle reads live CME's E-micro NASDAQ 100 futures prices from @PythNetwork's feed. the pricing mechanism converts this future price into a spot price using the cost-of-carry model. hence it trades lower than the actual futures price

when markets are closed, it switches to internal pricing. the oracle price will move towards a target price determined by the impact price difference (IPD) from the book. here is the breakdown of the formula used

the oracle doesn't jump to this target price immediately. it gradually moves using an 8-hour EMA applied to the IPD. this is different from hyperps, where the 8hr EMA of mark price IS the oracle itself.

there are several constraints during offline sessions. first, the oracle cannot move more than 5% from the last recorded price of the session. this is 1/max_leverage, which at 20x leverage means the oracle cannot move beyond your liquidation distance.

second, if the oracle goes offline for more than 48 minutes, the next update change is capped to about 9.5%. under normal operation, every 3 seconds, the price moves at a small rate towards the target determined by the orderbook.

the only question left for @sershokunin and @tradeperps team is: what is the impact notional used to calculate this price difference? how much depth in open orders does it take on either side to push the oracle during internal pricing?

tldr: this mechanism is just a filler for offline sessions, and by design does not encourage extreme scenarios. the 5% cap means you cannot get liquidated beyond what real market movement would justify when futures reopen. Hyperliquid.

@raul0ligma @tradeperps Thanks for being the first person to actually read the docs!

@sershokunin @tradeperps gud tech, gud math, gud docs.

@raul0ligma @tradeperps interesting concept, but how do we ensure oracle accuracy during volatile hours?

@cosmos_atom_ @tradeperps oracle accuracy depends on Pyth, but I assume they source it from the CME feed and depends on the exchange completely then, as in for offline hours, it's capped to 5%/

@raul0ligma @tradeperps I didnt understand any of this what's the worst case scenario in this like if there's a huge crash can these fuckers just scam wick to to straight zero

@raul0ligma @tradeperps Is this right? Price snaps back to pre-weekend level after weekend ends? Basically if during weekend price shoots up due to crypto bros gambling, we just short and make money or long if crypto bros short it? https://x.com/i/grok/share/Teo...

@raul0ligma @ThinkingUSD @tradeperps Don’t futures sometimes move more than 5% post market tho

@raul0ligma @sershokunin @tradeperps Very interesting

@raul0ligma @tradeperps When the market triggers a circuit breaker, what about $XYZ?

@raul0ligma @tradeperps Pythliquid.

@raul0ligma @tradeperps Hyperliquid

@raul0ligma @tradeperps never understood the science about this, but now that i get it, it's actually fascinating Hyperliquid

@raul0ligma @tradeperps thank you! was just wondering how that market close would affect pricing @tradeperps assets

@raul0ligma @tradeperps Gud tweet

@raul0ligma @sershokunin @tradeperps max leverage on market close time

@raul0ligma @zygomeb @tradeperps Fast take: $XYZ100 on @tradeperps heating up. Trade focus: funding skew, OI velocity, slippage across HIP3 DEXs 🟠 · numbers tell the story 📊. setup hits different. End of thread.