[Jared-dz]

Summary

• Add the Kalshi crypto up/down TOB and MBP feeds to the public Kalshi page and the Edge Connect runbook.
• Document feed codes kalshi-updown-tob and kalshi-updown-mbp, group codes edge-kalshi-updown-tob (233.84.178.26, ports 37000/47000) and edge-kalshi-updown-mbp (233.84.178.34, ports 38000/48000/58000).
• Note that up/down ports are a fixed block, same as perps.

Test plan

• Confirm the Feed Addresses table matches doublezero multicast group list for the two up/down groups.
• Confirm ports match the Edge Connect registry (37000/47000 and 38000/48000/58000).
• Preview the Kalshi page and check the intro count, subscribe example, and table render.

Made with Cursor

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated
doublezero-docs	Ignored	Preview	Nov 9, 2026 4:48pm UTC

[juan-malbeclabs]

Checked against mainnet-beta and the live wire; groups, ports and feed codes are right.

• multicast group list: edge-kalshi-updown-tob is 233.84.178.26 and edge-kalshi-updown-mbp is 233.84.178.34, both activated, two publishers each.
• feed list: kalshi-updown-tob and kalshi-updown-mbp are active in all 31 metros and point at those groups.
• On the wire, both publishers send TOB on 37000/47000 and MBP on 38000/48000/58000, and two receivers in different metros see both sources on every port. The two sources use different Channel IDs on the same ports, so the existing note on tracking sequence per source IP, Channel ID and port covers it.
• The runbook's 30000:59999 iptables rule already admits these ports.

One gap inline, not blocking.

[juan-malbeclabs]

Re-approving at 3ae4b6f. The StrikeInterval (0x09, 40 B, reference) row and the note on the 15-minute roll, the Manifest Seq advance and the pending strike cover the gap from my earlier comment. The rest is unchanged from the version I checked against mainnet-beta and the wire.