Measuring oracle staleness against a market that never closes
Tokenized stocks trade 24/7 on Base. The Chainlink oracle pricing them updates on a deviation trigger with no off-hours heartbeat. Find out how big that gap actually gets, and publish it.
Measuring it means three things agreeing. A stall window is defined from the
aggregator’s updatedAt; the size of the gap needs an independent reference
price for where the market actually went; and the Coinbase registry holds a
multiplier and a pause flag per asset that change the arithmetic underneath
both. The site records every window: when the feed stopped, when it resumed, how
long that was, and how far the market moved in between.
Thirteen feeds over three weeks: eighteen to twenty stalls of six hours or more per ticker, longest non-weekend stalls running thirty-seven to fifty-three hours, pools trading through at up to 209 basis points from the frozen value.
The constraints are stated on the site and they are the interesting part of the design. No routing, referral or affiliate revenue. No issuer sponsors or reviews it. No composite score and no ranking of issuers. It publishes no reference price and says outright that nothing here should be consumed by a system as one. An independent measurement is worth exactly its independence, and a ranking would spend that.
Still being pushed to.