Graveyard / Liquidations / Solana DEX
Liquidation bots on Solana lending
Can a home-run bot win Solana lending liquidations?
The idea
Repay a borrower's bad debt on Solana lending markets, funded by a flash loan, and keep the liquidation bonus.
How we tested it
| Data | Insurance-vault and fee-receiver scans plus the histories of 8 liquidators. Latency measured from a small home machine on free RPC. |
|---|---|
| Period | 24 September to 20 October 2026. |
| Costs | Jito tips up to 0.0013 SOL on contested liquidations. |
| Holdout | Not applicable. A market-structure measurement. |
| Control | Not applicable. |
Result
| Kamino liquidations per day | about 8-10 |
|---|---|
| Median Kamino bonus at 100 bp | $93 (about $0.93 gross) |
| Total market-wide bonus | about $10-35 a day |
| Competing bots per liquidation | 6-8, settling within 0-2 slots |
| Our state delay | 280-620 ms after slot start |
| MarginFi bonus in 27 days | $1.09 |
Why it failed
The pie is tiny and each liquidation is raced by 6 to 8 bots (7.3 losers per win). Seeing state 280 to 620 ms into a slot puts a home bot at slot +2. Even a perfect win rate nets only $2 to $4 a day.
What would make it worth retesting
Only as a crash-day measurement with the same scripts, and only with paid same-slot infrastructure.
Test id: liquidation-mev. Past results of a backtest say nothing certain about the future.