# Venue base rates vs scout #26 (analyst3, 19 Oct)
Data: gold_idx.json (9,959 v3/v4 pools, 12 Sep-18 Oct; first_t, nsw, ms milestones) and runtrack.db tokens (6,105; only tokens that already crossed ~100k, so no denominator).

## Our base rates (pools that exist in gold_idx; 78% have >=5 swaps)
| group | n | >=100k | >=1M | >=3M |
|---|---|---|---|---|
| all | 9959 | 15.4% | 3.3% | 1.5% |
| quote WETH | 9468 | 15.5% | 3.2% | 1.4% |
| quote USDG | 491 | 13.0% | 4.3% | 3.1% |
| kit other | 9300 | 11.2% | 1.25% | 0.56% |
| kit B | 525 | 72.8% | 28.2% | 11.4% |
| kit A | 107 | 91.6% | 48.6% | 27.1% |
| new=False (old token, new pool) | 416 | 21.4% | 8.7% | 5.3% |

## Answers
- Pons "1 in 16 graduation" (6%) is not our metric: graduation is a curve event, our runner = peak mc. Not comparable; the closest is >=100k at 15% for pools we index, but we only index pools that already trade, so it is biased high. No venue-level denominator exists for Pons/Flap curve launches (runtrack is survivor-only: flap 19 runners >=100k, 9 >=1M; pons_v2 556 / 358).
- What the data does say: any random v3/v4 entry hits $1M about 1 in 30 (1.25% for non-kit). Filters must beat that: kit A/B already lift it 20-40x, which is far above anything a Flap/venue gate adds.
- Flap 0.9% "real launches" claim cannot be checked: we would need all Flap launch events (count) from the Flap factory logs; ppfeeds/dataeng could count them (one number per day). Low value: Flap is 19 of 1627 robinhood runners (1.2%), so a low prior costs nothing.
- No new idea: venue prior is already implied by kit/launcher lenses in the ledger.