F01 // Financial Intelligence · Dealing Desk

AI trading operations for the modern dealing desk.

Real-time exposure, hedging recommendations, A-book / B-book routing intelligence and toxic-flow detection — engineered around your bridge, LPs and client book.

For retail and multi-asset brokerages running MT4, MT5, cTrader or proprietary bridges with active dealing operations.

The Problem

Why firms come to us for this.

01
Dealers fly blind between platforms

Exposure, PnL, LP flow and client behaviour live in five different screens. By the time a dealer stitches the picture, the risk has already moved.

02
A-book / B-book decisions are static

Routing rules are set once and forgotten. Client cohorts shift; markup and toxicity move with them; the book bleeds until someone reviews a spreadsheet.

03
Toxic flow is priced after the fact

Scalpers, latency-arbers and news-fraders are identified in Monday's report — after they've already run last week's book.

Anatomy

What we actually build for you.

Every engagement is bespoke — but the underlying architecture follows the same layered pattern. Owned by you, deployed on your infrastructure.

L05Dealer workstation UI (web + desktop)
L04Real-time exposure & routing engine
L03Toxic-flow & anomaly detection models
L02Bridge / LP integration layer (MT4/5, cTrader, PrimeXM, oneZero)
L01Streaming event bus (trades, quotes, fills)
Capabilities

What ships in the box.

Live exposure & PnL desk

Every symbol, every account, every LP — sub-second, aggregated the way your dealers actually think.

A/B routing intelligence

Per-client, per-symbol scoring that adapts routing in real time based on toxicity, size, latency and historical PnL.

Hedging assistant

Proactive hedge suggestions that respect your risk appetite, LP margins and cost of hedging across venues.

Toxic-flow detection

Latency arbitrage, tick scalping, news trading and mirror-flow patterns flagged the moment they emerge.

Book-shape monitoring

Continuous view of long/short balance, symbol concentration and LP dependency, with early-warning thresholds.

Dealer copilot

A chat surface trained on your dealing playbook: 'why did we bleed on gold last hour?' returns a real answer with the trades.

Example

A dealing-desk session, hour by hour.

Step 01
Session opens — book shape, overnight exposure and LP health render on the desk workstation.
Step 02
News event fires — copilot flags 34 accounts likely to trade the print in the next 90 seconds.
Step 03
Toxic-flow model routes 6 of those accounts B → A, quotes tighten on the retail cohort.
Step 04
Post-event, the copilot writes the shift note automatically: PnL, hedges, LP performance, exceptions.
Integrates with
MT4MT5cTraderPrimeXMoneZeroCentroidMatch-TraderFIX API
Talk to Us

Scoping a build for a brokerage, wealth firm or fintech?

We take a small number of new engagements each quarter. Send a two-line brief and we'll respond within 48 hours.