Solver KPIs #316
Description
There has been much back and forth, and a couple of false starts regarding KPIs (see #82, #213)
We now have a pretty clear idea of what the absolute MVP looks like for measuring our solver.
The following KPIs are intended to be simple to measure and capture performance via the lens of what actually matters for users.
KPIs
- Measure downtime (e.g. tracking time between trades for each network and set a downtime threshold) #317
- Arb transaction efficiency (as a proxy for opportunity staleness) #318
- Track transaction revert rates (as a proxy for front-running/MEV issues) #319
These should be measured by a separate process to the solver itself. It doesn't even need to be realtime, being able to run these on CLI for a historical period would be preferable to start with.