Online

—

Mode

—

Uptime

—

Backend

FastAPI

Database

SQLite

DB Path

—

Decisions

—

Trades

—

Server Time

—

Last Refresh

—

Heartbeat

2s

Universe refresh

60s cache

MIO research

30 min

PnL report

1h

Error watchdog

5 min

Consensus cooldown

5 min

Universe

—

Mode

—

Target/day

—

Cap/day

—

Decisions logged

—

Trades logged

—

Accuracy gate

—

SPI = 0.35 × max(0, -fz)/5
    + 0.25 × oi
    + 0.25 × div
    + 0.15 × liq

ENTER iff:
  SPI ≥ 0.65 (relaxed to 0.30 after 5 dry cycles)
  AND funding_z ≤ -1.0 (shorts paying)
  AND depth ≥ 10× position size
  AND consensus ≥ 0.45 (5-LLM Bayesian)
  AND conflict ≤ 0.75

TP = 0.008 + 0.037 × SPI
SL = -(0.005 + 0.010 × SPI)

Trailing stop:
  Activates at 60% of TP
  Trails at 60% of peak gain

base = capital × 0.95 / 3

size = base × sigmoid(consensus)
     × spi_scale(spi)
     × risk(drawdown, conflict)

BLITZ iff:
  SPI ≥ 0.85
  AND blitz_readiness > 0.60 (from MIO)
  AND dd_level < 1

Allocation: 60% of capital
Max hold: 1 hour
Cooldown: 4 hours
TP: 2.5% + 2.5% × SPI
SL: fixed -1.2%

Level 0: Normal (DD < 3%)
Level 1: Caution (DD 3-5%)
  → reduce position scale
Level 2: Defensive (DD 5-8%)
  → halt new entries
Level 3: KILL (DD > 8%)
  → close ALL, lock engine