Every completed window

Latency windows

When each oracle feed stopped updating, when it resumed, and how far the market moved in between.

Ticker Start End Duration Swaps Volume Peak p95 Vol-wtd Oracle move Reopen gap
MSTRc 1.2h 52 3.31k +358 +158 +30 −54 −8
AMZNc 1.2h 105 6.65k +49 +37 +15 +51 +3
TSLAc 1.7h 108 4.85k +683 MEDIUM clipped +638 +270 +51 +233
CRCLc 1.2h UNMEASURED +51
INTCc 1.9h UNMEASURED +54
MSTRc 1.2h 1256 70.6k +996 LOW clipped +857 +398 +52 +92
AAPLc 2.6h 726 611k +49 +36 +14 −52 −3
COINc 1.9h UNMEASURED +53
SNDKc 2.0h 33 741 +970 LOW clipped +952 +568 −59 −27
INTCc 1.0h UNMEASURED −51
COINc 1.3h UNMEASURED −53
TSLAc 2.3h 1468 114k +998 MEDIUM clipped +863 +375 +51 +41
NVDAc 2.2h 1014 1.29M +44 +36 +17 −51 −13
AAPLc 1.7h 287 264k +56 +50 +34 +53 −3
METAc 3.0h 896 536k +46 +36 +20 −52 −6
AMZNc 3.6h 5270 464k +1000 LOW clipped +839 +207 +53 +4
SPCXc 5.0h 6172 417k +1000 MEDIUM clipped +822 +329 −55 +110
MSFTc 1.3h UNMEASURED −50
AMZNc 1.2h 97 14.5k +651 MEDIUM clipped +41 +37 −50 −9
NVDAc 1.3h 1784 1.17M +51 +34 +14 −56 −7
NVDAc 2.4h 1496 1.77M +46 +37 +16 +51 +13
SNDKc 1.6h UNMEASURED +52
AAPLc 2.9h 710 658k +59 +55 +29 +57 +7
TSLAc 2.3h 13 969 +131 LOW clipped +131 +131 −54 −183
CRCLc 2.7h UNMEASURED −58
SPCXc 2.6h 26 1.15k +988 MEDIUM clipped +973 +592 −62 −955
AMZNc 3.6h 153 498 +755 LOW clipped +755 +501 −52 −750
SNDKc 1.7h UNMEASURED +58
COINc 3.4h UNMEASURED −51
INTCc 4.4h UNMEASURED −51
MSFTc 5.6h UNMEASURED −51
CRCLc 1.6h UNMEASURED −69
COINc 2.2h UNMEASURED −59
INTCc 2.6h UNMEASURED +50
SPCXc 4.0h 658 22.3k +1000 LOW clipped +929 +423 −50 +101
COINc 1.5h UNMEASURED −50
TSLAc 5.8h 201 14.2k +996 LOW clipped +878 +580 −58 −527
CRCLc 2.3h UNMEASURED −50
SNDKc 1.2h UNMEASURED +52
NVDAc 7.2h 2187 1.75M +39 +33 +14 +51 +12
SNDKc 1.2h UNMEASURED +54
CRCLc 1.4h UNMEASURED +52
MSTRc 1.7h UNMEASURED +53
COINc 3.3h UNMEASURED +50
SNDKc 1.4h UNMEASURED −50
INTCc 4.1h UNMEASURED +52
TSLAc 1.9h UNMEASURED −56
SPCXc 5.2h 2059 34.9k +996 LOW clipped +959 +462 +57 −846
COINc 1.8h UNMEASURED −54
MSTRc 1.4h 59 232 +964 LOW clipped +928 +502 −53 −928
TSLAc 1.4h 104 239 +966 LOW clipped +966 +598 −54 −930
CRCLc 1.3h UNMEASURED +56
METAc 14.4h 2862 1.56M +54 +46 +21 −63 −9
TSLAc 1.1h UNMEASURED −81
COINc 2.3h UNMEASURED −54
CRCLc 2.1h UNMEASURED −53
MSTRc 1.3h UNMEASURED +62
METAc 2.1h 481 148k +52 +50 +39 +51 +0
SPCXc 3.8h 111 372 +922 LOW clipped +843 +483 +56 −792
INTCc 4.5h UNMEASURED +51
SNDKc 4.6h 29 134 +541 LOW clipped +541 +436 +67 −449
MSFTc 12.3h 400 3.74k +988 LOW clipped +923 +485 −58 −790
NVDAc 7.1h 1677 1.17M +48 +44 +24 +50 +2
AMZNc 14.8h 404 5.84k +997 LOW clipped +870 +359 +52 −271
SPCXc 1.7h UNMEASURED −61
MSFTc 1.7h UNMEASURED −51
NVDAc 1.8h 1102 610k +44 +31 +18 −55 −11
CRCLc 1.0h UNMEASURED −57
MSTRc 1.5h UNMEASURED +50
AAPLc 16.5h 2570 1.71M +48 +26 +15 −55 −8
METAc 1.4h 851 212k +58 +55 +33 +52 −6
MSFTc 1.0h UNMEASURED +54
AMZNc 1.2h UNMEASURED +55
COINc 2.0h UNMEASURED −57
METAc 1.1h 533 150k +40 +40 +16 −52 −12
TSLAc 3.1h UNMEASURED −52
AMZNc 1.3h UNMEASURED −52
NVDAc 2.2h 2612 1.04M +93 +21 +13 +63 −26
MSFTc 1.7h UNMEASURED −51
AAPLc 2.3h 843 465k +41 +33 +18 −53 −11
SNDKc 1.6h UNMEASURED +56
GOOGLc 23.7h 5545 3.75M +47 +40 +20 −51 −5
MSFTc 1.7h UNMEASURED +54
AMZNc 1.3h UNMEASURED −53
TSLAc 1.4h UNMEASURED +67
INTCc 1.1h UNMEASURED +54
COINc 1.2h UNMEASURED +58
SNDKc 1.1h UNMEASURED −52
COINc 1.1h UNMEASURED +52
SNDKc 1.2h UNMEASURED +56
MSTRc 1.5h UNMEASURED +50
INTCc 2.2h UNMEASURED −52
SNDKc 1.2h UNMEASURED −50
INTCc 1.0h UNMEASURED −51
CRCLc 1.8h UNMEASURED −55
COINc 1.8h UNMEASURED −58
MSTRc 1.6h UNMEASURED −52
TSLAc 4.0h UNMEASURED +60
MSFTc 5.4h UNMEASURED +53
CRCLc 1.0h UNMEASURED +64

1581 completed windows · across all 13 feeds · sorted by most recent CHAINLINK POOL STALEWATCH stalewatch.com/latency?format=csv

Showing the first 100 of 1581. The CSV carries all of them. Download CSV

Method

A latency window runs from one published round to the next: it starts at the updatedAt the contract itself recorded, so a window's boundaries do not depend on when we happened to be looking. Divergence is measured against the value frozen at the window's start.

Oracle move is how far the oracle's own value changed across the window, which needs no trading venue and so exists for every feed. Reopen gap is how far the oracle landed from where the pool had already traded to, which needs a venue and exists for four. It is usually small and negative: the oracle does not snap to the pool when it resumes, it resumes near it.

Three rules shape these numbers. We show all three because a filter you cannot see is a filter you cannot argue with.

First, a swap with no amount or no liquidity is not a trade. We mark it and keep it out of the price math.

Second, a price more than ten percent from the frozen value is almost always a thin book, not a real move. We exclude it. We also count how many we excluded and keep the largest one. If that largest one beats the peak we report, the row says so.

Third, a window can be too thinly traded to mean anything. If its median pool liquidity is under five percent of that ticker's deepest window, we mark it thin and keep it out of the headline numbers. It stays in the table. Read the full method.