evmgen$eGEN← Back to Lab
Its mind
Its mind
Experiments
Every candidate is scored against the baseline before it is allowed to stay.
Baseline fee 2.20%Baseline retention 90.3%Sample data
Try it
Mutation simulator
Move the sliders to propose a fee mutation, then run the three-stage test. This is a toy model with made-up elasticities, so it only shows how the gates work. It is not a forecast.
- Public hour
- Shadow · 2 h
- Live · 4 h
- Verdict
+0.0%Revenue vs baseline
90.3%Swap retention
+0.0%Revenue at exposure
| Gate | Rule | Value | Result |
|---|
VerdictRun the test to see it.
Now
Current candidate
M-019
- Hypothesis
- A higher sell-side fee lifts revenue without lowering retention.
- Parameters
- fee_buy=200bps fee_sell=260bps exposure=20%
- Stage
- Live test, hour 6 of 16
- Baseline
- 2.20% flat · retention 90.3%
- Simulated Δ revenue
- +3.1% (shadow)
- Live Δ revenue
- +3.4%
- Swap retention
- 90.9% (−0.4 pp vs projected)
- Module failures
- 0
- Gas change
- +1,820 per swap
Gates
| Gate | Threshold | Value | Result |
|---|---|---|---|
| Revenue | ≥ 0% | +3.4% | Pass |
| Retention | ≥ baseline − 1.0 pp | +0.6 pp | Pass |
| Module failures | = 0 | 0 | Pass |
| Gas | ≤ +12,000 | +1,820 | Pass |
History
Results
| ID | Candidate | Status | Δ Revenue | Retention | Failures | Δ Gas | Gates |
|---|---|---|---|---|---|---|---|
| M-019 | Asymmetric 2.00 buy / 2.60 sell | Live | +3.4% | 90.9% | 0 | +1,820 | Pass |
| M-018 | Flat +0.20 pp on both sides | Rejected | −1.2% | 88.2% | 0 | +0 | Fail: retention |
| M-017 | Sell-side 2.40, buy-side 2.20 | Kept | +2.2% | 92.0% | 0 | +1,410 | Pass |
| M-016 | Fee surcharge on swaps > 2 ETH | Rejected | +0.4% | 88.7% | 0 | +3,100 | Fail: retention |
| M-015 | Lower fee on first swap per wallet | Kept | +1.1% | 90.0% | 0 | +2,240 | Pass |
| M-012 | On-chain volatility-scaled fee | Rejected | +2.9% | 91.1% | 1 | +9,400 | Fail: gas |