https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/aggregated-azure-sdn-upgrade-4.13-minor-release-openshift-release-analysis-aggregator/1628279724314726400

Shows flakes like:

: ingress-to-oauth-server-used-connections mean disruption should be less than historical plus five standard deviations

Reason: We have no historical data. 0s
: cache-oauth-api-new-connections zero-disruption should not be worse

Reason: Current percentile is so low that we cannot latch, skipping (P45=0.00s successes=[1628279720183336960=0s 1628279722632810496=0s 1628279715150172160=0s 1628279718518198272=0s 1628279717679337472=0s 1628279721793949696=0s 1628279723475865600=0s 1628279716832088064=0s] failures=[1628279715989032960=2s 1628279719407390720=1s])

This does not seem correct. Stephen identified that this may be: https://github.com/openshift/ci-tools/blob/master/pkg/jobrunaggregator/jobrunaggregatoranalyzer/pass_fail.go#L369-L377

Thread: https://redhat-internal.slack.com/archives/C02K89U2EV8/p1677080600400589