# A2.9: more trades at the same edge (6 Oct 2026, Bangkok)

## Read this first
- Same data and same strict rules as A2.8 (58 trades, 1.90x). Same exits: 5x target, stop at 0.6x, pull guard, 12 h. $500, 1% fee each way, pool-depth fills.
- **(a) Age 1.5 to 3 h:** pays in both halves, but the second half has only 3 trades. **Paper variant, tagged, not merged.**
- **(b) Re-entry after a stop:** pays in both halves. The buyer and army check makes it better in both. **Paper variant, tagged.**
- **(c) USDG pools:** 0 trades pass. No gain. Drop it.
- If (a) and (b) are added to the main rule: 81 trades, first half 1.85x (44), second half 1.90x (37), about +$35.4k at $500 over 16 days instead of +$26.2k.

| Version | Trades | First half | Second half | $ at $500 |
|---|---|---|---|---|
| Main rule (A2.8) | 58 | 1.91x (27) | 1.90x (31) | +26,224 |
| (a) age 1.5 to 3 h | 11 | 1.60x (8) | 1.98x (3) | +3,854 |
| (b) re-entry, price only | 20 | 1.45x (12) | 1.54x (8) | +4,844 |
| (b) re-entry, with buyer check | 12 | 1.91x (9) | 1.79x (3) | +5,283 |
| (c) USDG, full filter | 0 | | | 0 |

## (a) Coins that pass everything except age, at 1.5 to 3 h
- 28 coins hit the strict $1M gate between 1.54 h and 3 h. 11 pass top share ≤ 0.08 and army ≤ 0.009.
- Result: 3 of 11 hit 5x (one is HI, 0xee049979960d5e42b389b941ec45790ca7b6e1f7). 7 stops (one went to near zero, 0x8b462bd24331ab0593f4c604eaf9d3844fd7f549). 1 time exit.
- Without the best coin in each half: 1.14x and 0.58x. So it rests on a few hits.
- Without the army and top filters it fails: second half 0.88x (12). The filters do the work here too.
- Table: http://159.223.88.168/hx-c9ois/sw/sw-a2-9-age.csv

## (b) Re-entry after a stop-out
- **Rule tested.**
  - After a stop, watch the pool for 6 h.
  - Buy again at the first pool event where the cap is back at or above the first entry cap.
  - Buyer check at that moment, using only data up to that block:
    - new buyers since the stop ≥ 20 (an address that got tokens from the pool or a hub and never held this token before the stop);
    - army share of those buys ≤ 0.009 (strict army memory, same as A2.8).
  - Entry +5 s, same exits. One re-entry per coin.
- 34 stops. Price came back above entry within 6 h in 20 (median wait about 6 min). 12 of those pass the buyer check.
- **With the buyer check: 12 trades, 4 hit 5x:**
  - 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd;
  - 0xa2508b15ab16826c720631b9afa0a90c453ae8a4;
  - 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69;
  - 0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6.
- One passing re-entry was pulled to 0 (0x10897a25ded3b3f7d8cb9736fccf22c0ef5a414e).
- One coin the check blocked still hit 5x: 0x224a794031d54e49b46919ade2c55ff255dfdd97 (army 0.39).
- **Cut check (new buyers ≥ N, army ≤ 0.009):**

  | N | First half | Second half |
  |---|---|---|
  | 0 | 1.45x (12) | 1.16x (6) |
  | 10 | 1.91x (9) | 1.48x (4) |
  | 20 | 1.91x (9) | 1.79x (3) |
  | 50 | 1.77x (7) | 1.79x (3) |
  | 100 | 2.25x (5) | 1.79x (3) |

  Above 1x at every cut in both halves. The second half is thin (3 to 6 trades).
- **What it tells us about the stop:** many stops were wicks. Price was back above entry within 10 min in 13 of the 20. A stop that waits for a second print below 0.6x may do the same job more simply. Not tested yet.
- Table: http://159.223.88.168/hx-c9ois/sw/sw-a2-9-reentry.csv

## (c) The same gate on USDG pools
- 874 of 7,718 v3 pools quote USDG. 10 coins hit the strict $1M gate on a USDG pool first.
- **None pass the filter.** Top receiver share is 0.09 to 0.67 in all 10.
- Blind buy on all 10: first half 1.72x (7, one 7.8x coin), second half 0.18x (3). Fails.
- Table: http://159.223.88.168/hx-c9ois/sw/sw-a2-9-usdg.csv

## For the Coder (only if the Supervisor says add them)
- Tag (a) trades `age_1p5_3` and (b) trades `reentry`, so the C1 book can count them apart from the main rule.
- (b) needs token transfers for 6 h after each stop (the main rule only needs them up to the gate).

## First principles score
- (a) 6/10: real money test, both halves, but only 3 second-half trades.
- (b) 6/10: no look-ahead (everything at the re-entry block), but 3 to 6 second-half trades. Hubs are taken from addresses seen after the stop, with the same strict count rule as A2.8.
- (c) 9/10: clear no.

Version 1, 6 Oct 2026 Bangkok. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a2-9-more-trades.md