Execution Intel
Why our bots win or lose trades. Not about finding opportunities — about executing them.
emoji_eventsRecent Wins
link to raw datatrending_upOpportunity Quality
stable · 12 samples (6h)filter_altExecution Funnel
show_chartWin Rate by Spread Size
At what spread threshold do trades start being profitable?
errorFailure Breakdown
alt_routePerformance by Route
| Route | Total | Win % | Net PLS |
|---|---|---|---|
| V1→V2 | 2083 | 0% | -7,592,105 |
| WPLS -[pulsex_v1]-> USDC -[pulsex_v1]-> WPLS | 678 | 0% | +0 |
| WPLS -[emp]-> WETH -[emp]-> WPLS | 607 | 0% | +0 |
| WPLS -[emp]-> DAI -[emp]-> WPLS | 568 | 0% | +0 |
| WPLS -[pulsex_v1]-> WBTC -[pulsex_v1]-> WPLS | 559 | 0% | +0 |
| piteas→9x | 483 | 91% | +801,287 |
| WPLS -[pulsex_v1]-> INC -[pulsex_v1]-> WPLS | 390 | 0% | +0 |
| WPLS -[pulsex_v1]-> USDC -[9inch]-> WPLS | 293 | 0% | -176,077 |
schedulePerformance by Hour (UTC)
blockTop Losing Tokens
Tokens with most failures — candidates for blacklisting
| Token | Fails | Sim Reverts | Thin | Wasted Gas |
|---|---|---|---|---|
| EMIT | 39786 | 39786 | 0 | 0 |
| Finvesta | 10101 | 10101 | 0 | 0 |
| SCADA | 6068 | 6068 | 0 | 0 |
| GAS | 2156 | 6 | 2150 | 8,784,938 |
| ? | 1326 | 0 | 0 | 742,343 |
| USDC | 736 | 14 | 0 | 74,916 |
| WETH | 731 | 57 | 0 | 0 |
| WBTC | 628 | 0 | 0 | 9,826 |
streamFull Execution Feed
Raw signals feed including losses, misses, prevented & paper trades. Last 200 events across all bots. Auto-refreshes every 5s.
scienceExperiment History
| Name | Parameter | Change | Duration | Result | Action |
|---|---|---|---|---|---|
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 160 → 180 | 25h | DEGRADED | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 140 → 160 | 7h | NEUTRAL | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 120 → 140 | 16h | IMPROVED | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 100 → 120 | 12h | NEUTRAL | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 160 → 180 | 5h | NEUTRAL | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 140 → 160 | 6h | NEUTRAL | - |
| Auto-MIN_PROFIT_BPS-raise | MIN_PROFIT_BPS | 80 → 100 | 5h | NEUTRAL | - |
| Test MIN_PROFIT_BPS increase | MIN_PROFIT_BPS | 10 → 80 | 0h | NEUTRAL | - |
infoAssumptions & Limitations
Win rate: Calculated as wins / total TXs sent on-chain. Does not include paper trades or simulations-only. A "win" means the TX landed and net profit (after gas) was positive.
Avg profit: Mean net PLS per winning trade from cross-agg bot. Excludes paper trades. Skewed by outlier trades (e.g., INC 148K PLS single hit).
Detection speed: From backrun detector only (has timing data). V1V2 bot runs a ~1s quote loop (Multicall3 reserves + closed-form arb), not the old 3s poll — its detection speed is effectively that ~1s loop, not measured per-trade.
Spread analysis: "THIN" entries from V1V2 log where spread was real but gas exceeded profit. Backrun paper trades contribute to spread bucket analysis but are paper-only.
Failure reasons: Parsed from V1V2 bot text log. SIM_FAIL includes both tax tokens (always revert) and timing issues (reserves moved). Not distinguished without deeper analysis.
Losing tokens: Tokens with most SIM_FAIL + THIN entries. High sim-fail count may indicate tax tokens, not necessarily bad pairs. Check the tax registry before blacklisting.
"Lost to gas wars": Percentage of failures where spread existed but gas cost exceeded gross profit. Not true gas-price competition (PulseChain has no priority gas auction like Ethereum).
Time-of-day: Combined from all bots. Low sample sizes per hour — patterns may not be statistically significant until more data accumulates.
neurologyEngine Recommendations
Auto-generated by the insights engine. Last cycle: 13:31:03 UTC
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Weight more capital toward "piteas→9x" and reduce or pause "9x→squirrel" until conditions change.
Route performance divergence: "piteas→9x" wins 98% of trades (net +658.0K PLS) while "9x→squirrel" wins only 25% (net -1,560 PLS).
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
Backrun strategy shows strong spreads. Priority: connect the backrun detector to live execution via PulseAtomicArb.
Backrun detector averages 3219ms detection time and 183bps average spread on 6308 paper trades. This is a validated signal source — the spreads are real and detectable.
targetExecution Edges
4 competitive edges derived from mempool data. Updated 27m ago (STALE)
trending_upBackrun Targets
Mid-cap tokens with regular large human swaps — ideal for backrun analysis
No backrun targets found yet
routeAggregator Gap
Routing via Piteas/9X/EmpSeal that PulseX-only bots miss
No aggregator data yet
shieldBot-Free Tokens
Active human trading, zero/low bot coverage — uncontested
No bot-free tokens found yet
scheduleOff-Hours Windows
Time windows when bot activity drops and spreads persist