# A2.37 Late bloomer door (Analyst 2)

7 Oct 2026, 10:19 Bangkok (server clock). Pre-registered 10:15 in the status page, before any result.

## Answer
**KILL. Late $1M crossings do not run. Not one of 20 reached 5x.**

- **What I tested:** every token in my archive whose FIRST $1M sellable crossing came more than 3 h after its first swap.
  - That is 20 tokens out of 165 $1M crossings (944 pools). They could never pass our live age gate.
  - 15 crossed in half A, 5 in half B.
- **Live exits at $1k, real tick depth for all 20:** mean 0.61x.
  - Half A: -$293 a day.
  - Half B: -$456 a day.
  - Beats random entry on the same coins only 32% of the time in half B.
  - 0 of 20 hit 5x. The best paper peak after entry was 2.91x (0x40bd44f364bce2b3f842462041cdd9356aedf0c6).
- **No cut rescues it, even in half A (fitted in-sample):**
  - The best single cell is age 36.9 h or more: -$66 a day on 8 trades.
  - The best pair (age 36.9 h+ AND army 0.066+) is -$5 a day on 5 trades.
  - The frozen rule takes 0 trades in half B, so it can't pass.
- **How many went on to $3M+:** 1 of the 15 real ones: 0xd0ca71118ca21d674f9018a53026a38a0ab54dbf, which crossed already at $5.0M and peaked at $6.5M. The trade still came back at 0.40x.
- **5 rows are odd-supply or broken caps** ($58k to $1.8B "caps" from tokens whose cap math does not hold):
  - 0x3cae51882495b2a2dce68b062e5143245d2b7f5d;
  - 0x61a10e8556bed032ea176330e7f17d6a12a10000;
  - 0xd02e26f106bab3a93234feeaba74bf314dbd4c75;
  - 0xc82fb80ab69f68ca9f885f43f2b0cde0b499102c;
  - 0x2627cbc6880fcc356a49ce70aee59e18baa46224.
  - Without them: 15 trades, mean 0.62x (A) and 0.45x (B). Still a KILL.

## Why (mechanism)
- **The live gate's 3 h age cap is doing real work.** A coin that needs more than 3 h to find $20k of buyers at $1M is being bought slowly.
- **Without a fresh wave of new buyers, there is no fuel for 5x.** 9 of 20 hit the 0.3x stop; 7 of those first popped 1.14x to 2.13x.
- **Kit A coins don't save it.** All 3 that crossed late (0xc8885f68acb213b144fda1835d61ef05d88ad687, 0x1083634f9ff16e21240da237820a255c50098102, 0xb64fd97d3ce8c487bd3613fcda691eb874f37ace) stopped out at 0.26x to 0.30x.
- **This matches A2.25:** looser age gates (6 h, 12 h) already failed.

## Limits
- My archive only holds tokens whose pool opened within 72 h of birth. It ends 6 Oct 10:58 Bangkok.
- PP's e1-runners.csv (every RH pool ever at $3M+) is not published yet (404).
  - If it shows many $3M+ coins outside my archive that crossed $1M late, I re-run this exact frozen test on them.
  - That would take about 20 minutes once the file exists.
- Pricing used real tick depth for all 20 pools, every live friction, and a separate book (no shared slots).
  - The combined worst day with live goes from -$3,161 to -$3,679 in half B if every late crossing were taken.

## Files
- http://159.223.88.168/hx-c9ois/sw/sw-a2-37-late.csv (all 20 events: features at the crossing, exit, x, peak)
- http://159.223.88.168/hx-c9ois/sw/sw-a2-37-cells.csv (half A rule search)