🤫 Quiet Ensemble — the holdout survivors, forward-tested
The only 3 cells of 2,160 that survived the v3 miner's pre-registered time-split holdout — and improved on the
untouched test week. The family: quiet tokens (spt<0.5 — small average buys, no whale), mid-curve vsol,
modest tp15, age-partitioned. Now running against the live tick feed with engine v3 (exit prices from 869
real trades, Lighthouse guard-mirror). Judgment rules were frozen before the first trade.
LOADING
🎯 LIVE — real-SOL calibration bot (quiet_live_bot.py)LOADING
This bot's only job is collecting genuine tp15 fills — every calibration number on this page is currently a
guess (0 real tp15 trades exist) or thin (26 real tp25 trades), and a same-calibration retarget backtest showed
the family goes NEGATIVE at the one target with real data (tp40). Fixed TP, no trailing, Lighthouse guard on.
Not a profit test — judge nothing here until the sample bar below fills.
Real TP@15 samples collected
0 / 1000%
Real trades
—
Real P&L
—
Guard reverts
—
Stuck
—
Time
Symbol
Cell
Exit
Hold
P&L %
no real trades yet
💰 Family P&L (paper)
—
Trades
—
Win rate
—
Avg / trade
—
Median / trade
—
Q1-young55 age 0-30s · vsol 55-70
Q2-mid age 30-60s · vsol 40-55
Q3-mature age 60-120s · vsol 40-55
🚦 Decision gate — frozen at deployment (2026-07-19)
≥5 full days · family avg > 0 · ≥60% positive days · no cell median < −5%. Waiting for data…