Indexer checkpoint

Under exponential rebates, an indexer with a stake ratio of 10 on a subgraph keeps roughly:

The curve is effectively flat past a ratio of about 8, where over 99% is already returned. And no part of it depends on other indexers, which was the whole point of replacing Cobb-Douglas.

Your allocation is open, the deployment is synced, and it earns no indexing rewards. Most likely reason?

Indexing rewards require signal on the subgraph. Under Horizon you no longer close to collect, and a technically perfect allocation to an unsignalled subgraph earns nothing while looking entirely healthy.

max_amount_willing_to_lose_grt set very high means:

It bounds unaggregated receipt value. High means fewer aggregations, cheaper to operate, and more money in flight when something breaks. The name is literal.

Which is NOT grounds for slashing?

Downtime costs you revenue, which is the network's way of handling it. Slashing is for false claims, not for poor service, and the separation is deliberate.

Horizon permits indefinite allocations. What does the shipped indexer stack do at launch?

Protocol capability and shipped behaviour are different layers, and the docs state both on different pages. An operator who reads only the capability will be surprised by their own agent.

Which failure is invisible on a standard systems dashboard?

Overpricing looks exactly like low demand from your side. Every economic failure mode in this path reports green on a machine dashboard, which is why the advice is to monitor the money rather than only the machine.