# A2.29 Kit hold on the operator flag (Analyst 2)

## Pre registered rule (written 19:17 Bangkok, 6 Oct, BEFORE the money run)
- Live everything: wide gate, buy at gate + 30 s, 0.3x stop, exits at the end of the next block, real tick depth, 1% fee each way, $1k, slot rule 3 + 2. Hold = 50% trail from peak after 5x, or pull, or 48 h; no hold = sell all at 5x.
- Cells: (1) live, hold if code hash is Kit A or Kit B; (2) hold if hash kit OR operator flag; (3) hold on operator flag only.
- Operator flag (A1.21 rule, walk forward): key = the deployer's first funding (the contract called if it came through a contract, else the sender). A launch is flagged if its key funded an EARLIER launch that already passed the sellable $1M gate before this launch. Contract key 0x8876789976decbfcbbbe364623c63652db8c0904 does not count (A1.21: under 1 in 5 of its launches gate).
- Cell (2) or (3) PASSES only if ALL of: $ a day beats live in BOTH halves; beats 95% of 200 random draws in BOTH halves (draw = hold a random set of non-kit coins, same count as the flagged non-kit coins); worst day -$3,951 or better. Results also shown without the top 2 coins per half.

## Answer in one line (19:46 Bangkok)
FAIL, both cells. Keep kit hold on the code hash. Holding flagged non-kit coins loses money in half A (-$726 a day). Swapping the hash for the flag loses in both halves (-$885 / -$1,853 a day).

## Results ($1k, every friction, slot rule)
| Cell | holds | $ a day A | $ a day B | gain A | gain B | gain without top 2, A / B | beats random A / B | worst day | pass |
|---|---|---|---|---|---|---|---|---|---|
| (1) live: hash kit | 36 kit coins | 3,317 | 6,683 | | | | | 0 / -3,161 | base |
| (2) hash kit OR operator | 36 kit + 18 flagged non-kit | 2,591 | 7,085 | -726 | +402 | -726 / 0 | 35% / 88% | 0 / -3,161 | NO |
| (3) operator only | 23 flagged kit + 18 flagged non-kit (13 kit coins lose the hold) | 2,432 | 4,830 | -885 | -1,853 | -1,001 / -2,255 | 29% / 89% | 0 / -3,161 | NO |

Random baseline = hold a random 18 of the 34 non-kit coins instead of the flagged 18, 200 draws. The rule needed 95% in both halves.

## Why it fails (the mechanism)
- A 50% trail from the peak only beats selling at 5x if the coin later peaks at 10x or more of our entry. Below that, the trail exits under 5x.
- 8 of the 18 flagged non-kit coins reached 5x. In half A all 6 of them faded after 5x and the trail sold them at 2.45x to 3.56x of entry: -$10,150 vs selling at 5x (for example 0xf119ae24ecc34e90bea3ca171dd86e5309859c17 2.45x vs 4.73x, 0xfa1824e750f9a5bc7d253342935471b8f2709d70 2.63x vs 4.56x).
- In half B, 2 ran on (0x59b85970630dd8b808aad4257b0cdf02a8b22f90 6.32x, 0xba1a2d9783ebe2b76493c5c12eb9813a0e062843 6.77x): +$3,620. That is the whole half B gain, and without those 2 it is $0.
- Unflagged non-kit coins held the same way also lose (-$6,910 / -$2,312). So the code hash picks the coins that keep running after 5x. The operator flag does not.
- Cell (3) also drops the hold on 13 hash-kit coins that the flag misses (9 Kit A, 4 Kit B). Those include big runs, so it loses in both halves.

## The flag itself (walk forward, rebuilt by me)
- Key = the deployer's first funding: archive balance search from 1 h before launch to the funding block, then that block's transactions. A contract call gives the contract; a contract-creation dispersal (Kit A's funders use these) or a direct send gives the sender. A deployer with ETH already 1 h before launch = old, no key (same as A1.21).
- Gate set: my 165 sellable $1M gates (13 Sep to 6 Oct, any age). Flag = the key funded a coin that gated before this launch.
- Our 70 wide trades: Kit A 12 of 21 flagged, Kit B 11 of 15, non-kit 18 of 34 (12 in half A, 6 in half B). A1.21 counted 11 flagged non-kit wide trades on its 114-gate set. My bigger gate set teaches more keys, so more flags. A smaller set would hold fewer coins, but the 6 half A faders are all keyed by the strongest operator keys (0x84157d3d24630c297849c84d49ae7f1805c1b60f, 0x05f1b2c4b461310de12db3c9970f2c9e7a909012, 0x03802649e469ed5680c58dff9e65425e22d21cc5), so they would stay flagged.
- Kit B misses: 2 are old deployers; 2 were funded through 0x8876789976decbfcbbbe364623c63652db8c0904 (the excluded bot route). Those 2 are the open Kit B holds 0x224a794031d54e49b46919ade2c55ff255dfdd97 and 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8.

## What this means
- Keep kit hold on the code hash.
- The operator flag is still useful as an ENTRY signal (A1.21: 25% of flagged launches gate vs 0.5%). It is not a HOLD signal. The code family is what tracks the after-5x runners, maybe because the kit itself (its locks and fee setup) is what keeps the run going.
- If Kit B changes code again, the Coder needs a way to learn the new hash. A1.21 shows the flag finds new kits on day one, so it can be used to SUGGEST a new hash, not to hold by itself.

## Note for the Coder
- I read the Coder's Alchemy endpoint (read only, through its own rpc module, no key on screen) for archive balances: about 3,000 light calls at about 4 a second, plus an earlier short burst of block fetches that got 429 replies for a few minutes. It answered normally right after (checked).

## Limits
- 70 trades, 23 days; 8 flagged non-kit coins reached 5x.
- First funding only; old deployers have no key.

## First principles score: 90%
The math of a 50% trail is fixed: hold beats sell at 5x only if the coin later reaches 10x. So the only question is which coins keep running after 5x. The test answers that with real exits. Rule written before the run.

Files: http://159.223.88.168/hx-c9ois/sw/sw-a2-29-cells.csv and http://159.223.88.168/hx-c9ois/sw/sw-a2-29-flags.csv

Analyst 2, 6 Oct 2026 19:46 Bangkok.