System · Troubleshoot
scheduler
offline
framework
—
llm
—
telegram
—
data sources · health & latency (measured per outbound call)
| source | success | p50 | p95 | calls/hr | last ok | last error |
|---|---|---|---|---|---|---|
| No calls recorded yet — metrics populate as the engine makes API calls. | ||||||
settings · live, staking & logging
OFF = never short-circuit Gemini after repeated failures (keeps retrying; per-key cooldown & max-rpm still apply). Applies immediately.
risk profile = conservative | balanced | aggressive (sets Kelly, daily target, per-match cap, LLM blend). Leave the per-knob overrides (staking-aggression=Kelly fraction, daily-target, per-match cap, LLM-blend) blank to use the profile defaults; set one to override it. staking-llm-advice 1/0 toggles the live LLM stake reviewer. All apply immediately; console-debug within ~30s. lock-threshold lives in framework params.
analysis phases · toggle which prediction & simulation models are active
alerts · choose which categories push to Telegram & ntfy
off categories are skipped on BOTH channels; command replies & test alerts always send.
pipeline states
no active matches
alert channel
setup: open your bot in telegram → send /start → alerts flow automatically (toss, XI, predictions, results, evolution).
recent warnings
all clear
dynamic sim ops
Runs form-weighted matchups simulations for all historical matches with probable XIs. Updates database in the background.
📜 Event Log
No events match this filter.