# Launch runners, round 3: what runners look like at second 0, 5, 10, 15, 20 and 30 (5 Oct 2026, Bangkok)

## Read this first (for the launch session)
- This was written by a separate session that only reads your data. Nothing live was changed. Everything here is a proposal, with receipts.
- Ben's question: what do coins that go to $500k, $1M or multi-millions look like at second 0, 5, 10, 15 and so on?
- **Runner** here means: reached a sellable $500k AND was still 10x away from that peak at the moment we could buy. A coin that was already $500k when we could buy doesn't count.
- Data:
  - 14,134 launches with trades, 21 Sep to 3 Oct.
  - Per-second chain export: http://159.223.88.168/hx-c9ois/brx/persec.csv.gz (seconds counted from the pool's creation block; every buy, sell, buyer wallet, ETH amount and price).
  - Outcomes from mcap.db (sellable cap).

## 1. Coins that are "already at $500k" were already there at second 0
- 112 coins were worth $300k+ at 30 s.
- 88% of them were already $300k+ at second 0, in the launch block itself.
- Their typical launch cap was $4.5k. A median of 9.9 ETH was bought in the same block that created the pool (55% had 5 ETH or more). That is the creator buying at the moment of launch.
- Only 4% of them went another 10x after 30 s.
- So this group can't be bought earlier, and it isn't a runner pattern. Skip it.

## 2. Second 0: the launch settings are the main pattern
Runner rate (10x from the 5 s price and $500k+). Base rate is 2.0%.

| Launch settings | % of launches | Runner rate | % of all runners |
|---|---|---|---|
| fee 0.01%, 10 ETH start (the pond) | 9.8% | 11.4% (early half 9.9%, late half 13.6%) | 56% |
| same, native ETH pool | 1.1% | 18.3% | 10% |
| same, WETH pool | 8.8% | 10.6% | 46% |
| fee 1% | 42.8% | 1.2% | 25% |
| fee 0.25% or 0.05% | 34.2% | 0.4% | 6% |
| fee 0 (hook pools) | 2.8% | 2.3% | 3% |

- Launch time (BKK), for the low-fee group: 00 to 07 gives 11.0% and 12 to 18 gives 11.0%; 07 to 12 gives 8.1% and 18 to 24 gives 7.7%.

## 3. Seconds 1 to 30, inside the pond (pond rate 11.4%)
| When | Pattern | % of pond | Runner rate |
|---|---|---|---|
| 5 s | 5+ different buyers already | 3% | 24% (early 33%, late 22%) |
| 15 s | 2+ ETH bought | 8% | 19% |
| 15 s | price up 50%+ from launch | 7% | 17% |
| 15 s | 13+ different buyers | 12% | 15% |
| 15 s | no trades yet | 11% | 6% |
| 30 s | 4+ sellers | 14% | 3% |

- After 20 s the flow stops telling runners apart. At 30 s, buyer count, ETH in and price change are all flat (about 9% vs 9%).
- One big buyer who isn't alone (70 to 90% of the ETH) gives 7 to 8% at 15 to 20 s, but only in low-fee pools.

## 4. Waiting costs more than it tells you
| Seconds after launch | % of runners still 10x away | Runner's median multiple left | Model's top 5% hit rate (test days) |
|---|---|---|---|
| 0 | 87% | 16x | 14.5% |
| 5 | 72% | 15x | 12.4% |
| 10 | 63% | 14x | 13.8% |
| 15 | 56% | 12x | 14.2% |
| 20 | 50% | 10x | 12.1% |
| 30 | 42% | 8x | 14.2% |

- Model: gradient boosting on launch settings plus all flow up to that second. Trained on the first 60% of days, tested on the last 40%.
- The best picks are no more accurate at 30 s than at second 0, but by then they are worth half as much.

## Proposals
1. **Pond book at the earliest possible second.** Buy every launch with fee 0.01% and a 10 ETH start as soon as it is seen. Native ETH pools first.
2. **Second check at 15 s, inside the pond:**
   - Add when 2+ ETH has come in, or 13+ different buyers, or the price is up 50%+.
   - Cut when there have been no trades at all.
3. **Never chase a launch-block jump** (5+ ETH bought in the launch block, cap already $300k+).
4. **Speed is part of the edge.** Every 5 s waited inside the pond removes about 5 to 10% of the runners.
5. **Skip 0.25% and 0.05% fee launches** for the runner lane (0.4%).

## Data warnings
- mcap.db often has only about 7 min per coin, so runners that peak later are undercounted.
- Only 13 days of data. The native ETH pond only appears in the late half (153 launches).
- The buyer is the final token recipient in the transaction, which can be a router for some trades.

Scripts: coin brain session /root/brain/r3 (build.py, it2 to it5.py); export script /opt/brain/review/exp_persec.py (read only).

Version 1, 5 Oct 2026. Latest: http://159.223.88.168/hx-c9ois/brx/launch-runners-round3.md