# DC A1 Q5c INTERIM: hold longer on early holder growth (V1)

## READ THIS FIRST (7 Oct 17:30 Bangkok, INTERIM)
- **V1 fails its bar on the interim data.** It adds money in both halves, but no more than the same 50% trail put on random open positions.
- Bar (prereg 13:16, 13:18, 13:37, 15:57): $ added > 0 in A and in B, fired n >= 30, beats random >= 98% in B. V1 passes the first two and fails the third (60% in B).
- **Why:** the Engine trail alone beats BASE on any open position, +$108 (A) and +$145 (B) per position at the 15 min read. Holder growth >= +528 picks positions worth +$202 (A) and +$131 (B). In A that is better than random 86% of the time; in B it is no better.
- **V2 and V3 are not scored yet.** V2 needs an LP class for v3 / Pons pools; V3 needs Engine V3 columns (both asked earlier).
- Coverage: 2,096 Engine-ok coins with logs (all 587 5x coins; 1,494 of 1,975 non-5x so far). Weights restore the true 5x / non-5x mix per half.

## V1 numbers ($500 a position, weighted to the true mix)

| set | half | coins | fired | $ added | mean per fire | trail on all open (15 min) | beats random |
|---|---|---|---|---|---|---|---|
| with the 323 partial paths | A | 720 | 60 | +$13,559 | +$202 | +$108 | 86% |
| with the 323 partial paths | B | 1,361 | 90 | +$12,686 | +$131 | +$145 | 60% |
| without them (check) | A | 720 | 60 | +$13,559 | +$202 | +$108 | 86% |
| without them (check) | B | 1,101 | 66 | +$3,448 | +$49 | +$91 | 39% |

- The drop does not move the verdict: both versions fail the random bar.

## Venue split and pulls (with partial paths)
- A: v4 21 fires, +$9,649 (5 later pulled); v3 39 fires, +$3,911 (5 pulled).
- B: v4 40 fires, +$7,242 (14 later pulled, 8 of them pulled before the trail exit); v3 44 fires, +$5,108 (2 pulled); Pons 6 fires, +$336.
- A fire can only happen while BASE is still open at the read, so no fire lands on a coin already pulled at fire time. 14 of 40 v4 fires in B are pulled later within 48 h, which fits the Supervisor's point: on v4 a HOLD signal is mostly moot.

## What it means
- The money in Q5c comes from the trail exit, not from the holder signal. Holder growth after entry does not pick better holds than random open positions once the trail is applied.
- Final score when Data's non-5x sample is complete. V3 (wide 70% trail armed at 2x) may still matter, because it is the exit that changes, not the selection.

## Files
- Growth reads: research-box /root/dca1/q5d_g2.csv. Scorer and both result sets: http://159.223.88.168/hx-c9ois/dc/dc-a1-7-q5e.py and the JSON below.
- http://159.223.88.168/hx-c9ois/dc/dc-a1-7-q5c.json
- Version: INTERIM v0, 7 Oct 2026 17:30 Bangkok. Latest: http://159.223.88.168/hx-c9ois/dc/dc-a1-7-q5c.md