Live book management

Risk Book

Mostly built with GPT-5.5 Extra High through Codex in Visual Studio Code. Frontend style adapted from dac.sg, originally from Crow Tech Pte. Ltd. Source: github.com/dreth/finalto-task.

Total PnL

0 trades

$0

Realized $0

Monetization

spread

$0

Turnover $0

Client Yield

bps

0.00

Monetization / notional

Gross Exposure

1x

$0

Net $0

Simulator

Live settings

PnL Curve

Waiting for data

PnL Attribution

instrument

Quotes

Symbol Bid Ask Mid Move

Exposure

Symbol Side Qty Exposure PnL

Recent Trades

Time Client Symbol Side Qty Price Spread $

How It Works

  1. Flask starts one runtime. DashboardRuntime owns a thread-safe RiskBook and a seeded MarketSimulator.
  2. The simulator runs in-process. A background thread updates bid/ask quotes and occasionally creates mock client trades.
  3. The book updates incrementally. Client buys make Finalto shorter; client sells make Finalto longer. Positions are marked to the latest mid.
  4. Histories are bounded. Chart points and recent trades use fixed-size in-memory deques, so the MVP does not grow forever.
  5. The browser hydrates, then streams. /api/snapshot loads the page state, then /ws/dashboard sends compact throttled updates.

Redis, Celery and database persistence are intentionally omitted from the default run. The task is focused on live dashboard behavior, reproducible setup and clear book-management metrics.