Bot Titles
PLS-LIQ-KEEPER-V1Performance So Far
PLS-LIQ-KEEPER-V1Recent Fire Attempts
PLS-LIQ-KEEPER-V1Token Reverters (24h)
PLS-LIQ-KEEPER-V1Biggest Block
PLS-LIQ-KEEPER-V1Biggest block (2026-07-04): funding + gas-bid calibration — both operator decisions, not code. The wallet (0xbAb8…cd88) holds 0 PLS, so the keeper cannot act at all. Once funded, the measured field matters: winners of Liquid Loans closings paid a MEDIAN ~297× base fee (max ~5306×) to get included, and our LIQ_GAS_MULT=11 loses most contested events. The economics make this easy to fix — a 297× bid costs ~$0.43 and the historical max ~$7.7, against an average ~$300 reward per closing (200 USDL + 0.5% of collateral) — but raising the bid and the gas floor (one competitive attempt needs ~150K PLS; the current 20K floor cannot cover one shot) scales real spend, so it is gated on the operator. Secondary constraint: event frequency is low (~0.7/day Liquid Loans, ~0.1/day EARN) and closings only occur when the collateral asset falls enough to push a position under the 110% floor — long idle stretches are the correct behavior, not a fault.
Fantasy
PLS-LIQ-KEEPER-V1If we had some way to instantly remove this bot's biggest constraint — capital, latency, signal coverage, or counterparty risk — what would the world look like? (Pending creative assessment.)
Suggestions for Improvement
PLS-LIQ-KEEPER-V1- 01HIGHFund the dedicated gas wallet (0xbAb8…cd88) — ~1M PLS (~$6 at current prices) buys several competitive attempts. Zero principal at risk beyond gas: estimateGas gates every send, and the 3-revert breaker caps lost-race burn.
- 02HIGHRecalibrate the gas bid off the competition probe's measured winning curve instead of the current LIQ_GAS_MULT=11: median winner paid ~297× base fee. ~600× (2× the median winner) costs under $1 per attempt against a ~$300 average reward. Raise LIQ_MIN_GAS_PLS to ~200K alongside (one 600× attempt needs ~150K PLS; the 20K floor cannot cover one shot).
Conclusion
PLS-LIQ-KEEPER-V1Verdict (2026-07-04, build + same-day review): correctly built, correctly idle — the remaining moves are operator calls. Both protocols' addresses and liquidate selectors verified on-chain twice (build + review); every send is gated by the protocol's own estimateGas, so a position can only be closed when Liquid Loans / EARN themselves agree it is under the 110% floor, and a false trigger costs one eth_call. The field was measured honestly: 47 closings/8 winners on Liquid Loans and 7/3 on EARN over ~70 days, with winners paying a median ~297× base fee — contested, not empty. The keeper runs a probe that logs each closing's winning gas so the bid can be calibrated from data. Idle-at-deploy is correct behavior (riskiest positions sit at ~122% and ~222% vs the 110% floor); it earns only in PLS/PLSX drawdowns. The passive sibling (PLS-SP-EARN-V1) captures the same event value without entering the gas race — fund that first if choosing one.