The dashboard's Advanced tab

The dashboard — ⚡ LND Health — reads live from LND plus the operator database; it never writes. It's split into three tabs: Node (health: status, balances, watchtowers, backups, alerts), Flows (traffic: channels, sat flow, payments, routing, daily revenue) — both self-explanatory — and Advanced, the economics layer the pipeline actually acts on. That last one is worth a walkthrough.

Channel details — economics

One row per channel, in fee-pipeline order — each column is one input to the decision the engine makes every cycle:

• Local Fee — the outbound ppm currently advertised; an active inbound discount shows underneath as (inbound −N).
• Clearing — where the fee would settle with no floor: sigmoid(local ratio) × market multiplier.
• Mkt Mult — the demand-driven multiplier, green when raising, red when discounting.
• Floor — the refill-cost floor (last refill × margin). 566 ↓ of 613 means it's decaying toward a sellable price while the channel sits idle.
• Earned — trailing revenue ÷ out-volume; calibrating means too little volume yet to measure, so no cap applies.
• Refill Budget — what the planner will pay to refill: plain = full escalation, yellow ⛒ = profit-capped at earned × horizon, red stranded = refilling is a losing trade and it's parked as a capital decision.
• Rebal Cost 30d / Net Lifetime — what keeping it filled cost lately, and whether the channel has paid for itself overall.

Read across a row and you can reconstruct the engine's whole verdict — e.g. a channel earning 68 ppm with a 566 floor and an 85-ppm capped budget is one that bought liquidity dear, sells it cheap, and is being repriced.

Forwarding failures — lost-revenue watch

Routing attempts your node dropped in the last 24h — fees you could have earned but didn't. These events are live-only (LND persists them nowhere), so the Monitor service badge matters: if the htlc_monitor daemon is down you're blind, not clean. It splits empty-channel drops (recoverable — refill and the revenue comes back) from fee-too-low and expiry causes, estimates the lost fees, and names the worst channel.

Fee updates & rebalance history

• Recent Fee Updates — every broadcast with old → new ppm, the local ratio at the time, and the layer that set it: sigmoid+mkt, floor, floor ↓ (decaying), or 📌 pin.
• Rebalance History — every attempt: route, amount, fee paid with effective ppm (or the ≤Nppm budget tried on a failure), ✓/✗, and auto vs manual. A muted skipped badge (rather than a red ✗) marks a QueryRoutes early-out — a dry-run found no route within the affordable ceiling, so no payment was attempted — so it isn't miscounted as a tried-and-failed rebalance.