4.4 Troubleshooting & Performance Tips

4.4.1 “No opportunities found”

Possible causes:

• Market is quiet.

  • Let it run longer; watch different pairs/times of day.

• min_profit_threshold too high.

  • Try 0.1 or even 0.05 to see more.

• One or more DEX APIs are failing.

  • Check logs / console for HTTP errors.

4.4.2 “Connection error”, HTTP 4xx/5xx, timeouts

• Your network/RPC is flaky or DEX APIs are rate-limiting.

• Add basic retry/backoff logic if needed.

• Increase sleep interval a bit if you’re hitting rate limits.

4.4.3 High CPU / RAM usage

• You set the polling interval too low with many venues.

• Reduce number of DEXs.

• Increase sleep to 2–3 seconds.

• Don’t run 5 instances on a Raspberry Pi and expect miracles.

4.4.4 Output looks insane (huge spreads, nonsense prices)

• This usually means bad or partial data from one DEX.

• Add sanity checks:

  • Drop quotes with obviously broken prices (0, negative, >10x median).

• Log raw response for the suspicious venue and fix the parser.

4.4.5 Sanity-checking before you trust it

• Pick a few opportunities from the terminal.

• Manually query the same routes via Jupiter UI/API and individual DEX UIs.

• If the engine is consistently showing spreads that don’t exist in reality:

  • Adjust normalization/decimals.

  • Tighten filters.

  • Fix any bug in the fetch/parse layer.

---

If you can’t get the engine to run cleanly after this, the problem is either your Python setup or your network. In both cases, fix your stack before touching arbitrage.