**This corrects, not just narrows, the P2 adversarial review's own Finding 8.** That review's day-level
extension (`P2_fleet_economics/98_REVIEWS/ADVERSARIAL_REVIEW.md` Finding 8) compared a sender's total
*same-day buy cost* (0.501-0.609 x face value of shares sent) against the *face value* of shares sent to
C, concluding the group's margin is clearly positive because cost < face value. That comparison silently
assumed shares sent = $1 realizable value each. Correctly comparing cost to *realized* payout (face value
x measured win share) for the same six sender-days:

| | value |
|---|---:|
| Total buy cost, all trades, 6 sender-days (`15_DAY_LEVEL.csv` + extension) | $14,193.46 |
| Total face value of shares sent, same 6 sender-days | $26,659.98 |
| Cost / face | 0.5324 |
| Implied real payout at measured win rate (58.0% legs / 60.3% $) | $15,462.79 – $16,075.97 |
| **Real margin** | **+$1,269.32 to +$1,882.50** |
| **Margin as % of cost** | **+8.9% to +13.3%** |
| Margin as % of face value | +4.8% to +7.1% |

(`adv_raw` inline Python check, numbers reproduced above; Decimal arithmetic throughout.)