The AI Brain Behind
Your Dealing Desk

Two engines working together: algorithmic detection classifies experienced traders into 17 categories, while ML predicts risk for new accounts from the very first trades.

0Algorithmic Categories
0Behavioral Metrics
Day 1ML Predictions
10M+Orders / Second
Get Demo Access
Try 2 Weeks Free * * with automatic seamless onboarding
The Problem

Every Misrouted Trader Costs You Money

A toxic trader on B-book bleeds your P&L. A consistent loser on A-book wastes LP costs. Manual classification is too slow, too crude, and too late.

⏱ 2–4 weeks

The New Trader Blind Spot

Every new account sits on B-book while you wait for data. A latency arbitrageur costs you $3,000+ in that window alone. A signal-follower copies profitable flow — and you internalize every loss.

📈 $400K+/mo

The Misclassification Tax

200 consistent losers on A-book instead of B-book: that's hundreds of thousands in foregone internalization profit. Without granular classification, this revenue leak is completely invisible.

💣 1 event

The Martingale Time Bomb

Grid traders look profitable for months — until the grid collapses. One margin call wipes out weeks of B-book income. Without volume progression monitoring, you never see it coming.

Dexa solves all three. Automatically.

Command Center

Your Entire Book. One Screen.

Stop switching between MT5 admin, spreadsheets, and bridge dashboards. Every KPI, every classification, every trade — live, drag-and-drop, configurable from 4 hours to one year.

No more morning scramble through multiple tools. A whale opens a $2M position while you're at lunch — you see it instantly. Exposure breaches are discovered in seconds, not hours. Pre-shift the Martingale trader to A-book before the grid collapses — not after.

Real-time Exposure Dexa Dashboard
Dual AI Engine

Two Engines. Zero Blind Spots.

The industry forces you to choose: wait weeks for data, or guess. Dexa runs two engines in parallel — algorithmic detection for traders with history, and ML prediction for new accounts from the very first trades.

When data is sufficient
⚙

Algorithmic Detection

17 categories • 79 metrics • deterministic

Rule-based engine with 16 specialized detectors evaluated in priority order. Each detector analyzes a specific behavioral pattern — from Martingale volume progression to latency arbitrage timing. Deterministic, explainable, auditable.

  • 79 metrics across 4 computation levels (SQL + C#)
  • Win rate, profit factor, Sharpe ratio, deal duration
  • Order flow: cancel rate, reject rate, pending order ratio
  • Martingale series, SL/TP usage, news trading correlation
  • First match wins — transparent reasoning for every decision
From the very first trades
🧠

ML Early Prediction

XGBoost • gradient boosting • probabilistic

Machine learning engine that predicts trader risk category before enough data exists for algorithmic detection. Trained on historical patterns from thousands of classified traders, it recognizes early behavioral fingerprints that humans cannot see.

  • XGBoost gradient boosting with feature importance ranking
  • Registration metadata: country, account type, deposit size
  • First-session behavior: time-to-first-trade, initial lot sizing
  • Early trade patterns: volume-after-loss, speed acceleration
  • Confidence score increases with each new trade
ML Prediction
Trades 3–29
→
Seamless Handoff
Trade 30
→
Algorithmic Detection
Trade 30+

No gap in coverage. ML handles the first 29 trades with increasing confidence, then hands off seamlessly to algorithmic detection once enough behavioral data exists for deterministic classification. The latency arbitrageur gets flagged after 3 trades — not after 30.

ML Engine — Predicto4FX

Predict Risk Before the Data Exists

The Predicto4FX Engine uses XGBoost gradient boosting trained on historical data from classified traders. It analyzes registration metadata, early behavioral signals, and trade patterns — producing a risk prediction that improves with every new trade.

0
0
trades
Registration metadata
→
1
3
trades
72% A-book recall
→
2
5
trades
75% A-book recall
→
3
10
trades
78% A-book recall
→
∞
30+
trades
Algorithmic takeover

Registration & Metadata

Country, account type, initial deposit size, time of registration. Specific jurisdictions and deposit patterns correlate with aggressive trading styles — the model learns these patterns across brokers.

First-Session Behavior

Time from registration to first trade. Initial lot sizing relative to balance. First-session P&L predicts lifetime value — a technique borrowed from casino and gaming analytics.

Early Trade Patterns

Volume-after-loss ratio detects Martingale from trade #2. Speed acceleration reveals tilt. SL/TP usage on first trades. Position sizing patterns emerge by trade #3. XGBoost weights these signals by predictive power.

No more 2-week blind spot on new accounts. The ML engine routes intelligently from day 1 while the algorithmic engine waits for sufficient data. Once 30+ trades accumulate, deterministic classification takes over with full explainability.

Heartbeats — P&L

Know Exactly Where Your Profit Comes From

Slice profitability by symbol, group, classification, trader, order type. Identify which segments drive your P&L — and which silently erode it.

  • Average and total P&L across every dimension
  • Market orders vs pending — see execution quality impact
  • Top 20 traders by profit with one-click drill-down

Spot fading traders — yesterday's toxic A-booker whose edge has died. Move them to B-book before you waste another month of LP costs on losing flow.

P&L Analytics
Heartbeats — Volume

Volume Spikes Tell You Something. Listen.

A sudden volume increase on GBPUSD before NFP. A trader doubling lot size after three consecutive losses. Volume patterns are your early warning system.

  • Volume by symbol, group, classification, and trader
  • Market vs pending order volume split
  • Deal table with click-through to full transaction detail

Detect Martingale volume progression — the grid trader averaging down — before the position becomes unmanageable.

Volume Analytics
Execution Tracking

A-Book vs B-Book. Every Dollar Attributed.

Know exactly where your profit comes from. Dexa separates B-book market risk revenue from A-book LP markup income — by symbol, group, country, and trader classification. No more guessing which book drives your P&L.

  • B-book P&L, A-book revenue, and combined broker P&L at a glance
  • Volume split and P&L per lot efficiency metrics
  • Daily trends: track how your A/B balance evolves over time
  • By-symbol breakdown: spot which instruments drive each book

Your GBPUSD B-book lost $12K last week while EURUSD earned $45K. 60% of A-book volume comes from one trader group generating only 15% of markup revenue. These are decisions you can act on — today.

A/B Execution Summary
Analytics

Which Instruments Drive Your Business?

Top symbols by volume, P&L, and trader count. 9 sortable metrics per instrument. Know exactly where your book's concentration risk lives — and where the revenue comes from.

VolumeP&L Win RateAvg ProfitTrader Count
Symbol Analytics
Sessions

When Do Your Traders Trade?

24×7 activity heatmap across Asian, European, and American sessions. Colored by deals, volume, or P&L — spot session-overlap risk concentrations and plan your desk coverage accordingly.

Your desk is staffed London hours, but 40% of toxic flow comes during the Asian session. Now you know — and you can pre-configure automations for off-hours coverage.

Session Analytics
Groups

Your MT5 Groups, Decoded

Which groups are profitable for B-book? Which are leaking? See users, volume, P&L, deposits, and the dominant trader category per group — and act on it.

  • Top 10 by volume and P&L at a glance
  • Dominant category badge reveals each group's risk profile
  • Active traders, win rate, and deposit metrics
Group Analytics
Finance

Follow the Money

Deposit and withdrawal flows by country, group, and classification. Spot unusual patterns — a surge of withdrawals from a specific country, deposits concentrated in a single group.

Know if your best B-book group is about to churn. A spike in withdrawal requests from Consistent Losers may signal a marketing campaign pulling traders to a competitor.

Finance Analytics
Traders

Every Trader. 40+ Metrics. One Click Deep.

Configurable columns across 7 metric groups. Server-side sorting, multi-server filter, instant search. Click any trader for a full dossier: 15 KPIs, cumulative P&L chart, classification history with before/after comparison, and similar trader detection.

Custom ColumnsMulti-Server AI RecommendationClassification History Similar TradersP&L Chart

No more opening 5 tabs per trader. Login, category, risk score, P&L trend, similar accounts — all in one view. Investigate a suspicious account in 30 seconds, not 30 minutes.

Trader List Trader Detail
Classification & Actions

17 Algorithmic Categories. Automated Routing.

Each of 17 categories is defined by deterministic rules across 79 behavioral metrics — from Toxic Profitable to Consistent Loser, from Martingale to News Trader. Every classification comes with a transparent reason: which thresholds were met, which signals triggered. Fully auditable, fully explainable.

  • Color-code MT5 accounts by risk level automatically
  • Move traders to groups based on classification
  • Batch-apply actions to all traders in a category
  • Cooldowns prevent action storms on reclassification

Manual group changes at 7 AM based on yesterday's data are over. Classification changes trigger routing actions automatically, 24/7, with full audit trail.

Trading Patterns

Compatible with any Bridge

Dexa Analyze → Change Traders Groups → Bridge Execute

oneZero, T4B, Centroid, FXCubic, PrimeXM — works alongside your existing bridge.

Detection

Catch Coordinated Trading That VPNs Can't Hide

IP-based linking catches amateurs. Behavioral correlation catches professionals. Dexa analyzes trade timing, direction, instruments, and volume across all accounts — detecting coordination even when traders use different devices and VPNs.

  • Similarity scoring across 4 behavioral dimensions
  • Filter by group size: 2+, 3+, 5+, 10+ linked traders
  • Direct links to individual trader profiles

20 copy-trading accounts look like 20 independent positions. They're actually one correlated exposure. Detect the cluster before the LP notices the toxic flow pattern.

Similarity Detection
News Trading

Five Minutes That Swing Your P&L

NFP, CPI, FOMC — LP spreads widen to 10+ pips, liquidity thins, price moves 50–100 pips. News traders pile in. If they're on B-book and predict the direction, you absorb the loss.

  • Live economic calendar matched to your trading data
  • Time-delta analysis: who traded within ±1 min of the event
  • Impact filter: High, Medium, Low

Pre-route news traders to A-book before the announcement. See the straddle strategy building in real time. No other dealing desk tool integrates an economic calendar with trade data.

News Activity
Monitoring

Your Desk Doesn't Sleep.
Neither Do Your Alerts.

Configurable rules across 6 event categories. A classification change at 3 AM triggers an alert, an email digest, or an automated MT5 action — all before your desk opens in the morning.

  • Classification, Performance, Volume, Financial, Exposure, System
  • Immediate, hourly, daily, or weekly delivery
  • Test alerts before going live
Events & Alerts
No-Code Automation

Your Dealing Desk Rules. Built by Dealers, Not Developers.

Visual flow editor: Event → Filter → Calculation → Action. Drag blocks, connect them, save. No tickets to IT. No waiting. 10+ MT5 actions: SetColor, MoveToGroup, DisableTrading, EnableEA, and more.

"When a trader is classified as Latency Arbitrageur, set color to red and move to A-book group." One automation replaces a daily manual process that used to take 45 minutes and was always a day late.

Automation Editor
Reports & Data Export

All Your Data. Any Format.
One Click.

Generate detailed reports on traders, transactions, order flow, and execution quality. Export everything in PDF or CSV — ready to feed into your internal workflows, share with stakeholders, or use as a foundation for any reporting requirements you may have.

PDFCSV Custom PeriodsMulti-ServerAuto-Generated
  • Standard reports: traders, deals, volume, P&L summaries
  • Flexible date ranges from 1 day to custom periods
  • Built-in report manager with download history
  • Data foundation that can support your internal reporting needs

Stop manually exporting data from MT5, cleaning it in Excel, and formatting it for reports. Dexa generates structured reports in seconds — giving you a clean data foundation for any internal or external use.

Reports
Getting Started

Live in 5 Minutes. No Code. No Consultants.

1

Connect

Add your MT5 or MT4 server credentials. Test connection. Done — five minutes from signup to data flowing.

2

Analyze

Dexa syncs your data. ML predicts risk for new traders immediately. Algorithmic engine classifies experienced traders into 17 categories.

3

Act

Review AI recommendations. Configure automations. Generate reports. Your dealing desk is AI-powered.

Infrastructure

Cloud-Native. Infinitely Scalable.

From 500 traders to 100,000+. Tens of millions of orders processed in seconds. Your dashboards stay fast no matter how big your book grows.

⚡

Elastic Scaling

Auto-scaling containers adapt to your load in real time. No capacity planning, no over-provisioning. Scale with your business.

🛰

ClickHouse Engine

Column-oriented analytics processes billions of rows in milliseconds. The same tech that powers Cloudflare and Uber analytics.

☁️

Cloud or On-Premise

Managed cloud (AWS), your private cloud, or fully on-premise. Adapts to compliance and data residency requirements.

Multi-Server MT5 Incremental Sync Per-Broker Isolation Auto-Scaling 99.9% Uptime SLA
Comparison

Why Brokers Choose Dexa

Dexa
Traditional Solutions
✓ ML predictions from trade #1, algorithmic precision from trade #30
2–4 weeks of data before any insight
✓ 17 algorithmic categories + ML early prediction
Manual 3–5 buckets defined by humans
✓ Real-time exposure with live updates
End-of-day reports and spreadsheets
✓ Visual no-code automation builder
Developer-dependent rule changes
✓ 5-minute onboarding, no integration project
Weeks-long setup with consultants
✓ Native multi-server MT5 support
Single-server or complex workarounds
✓ Built-in reports with PDF/CSV export
Manual data extraction and spreadsheet formatting
✓ Millions of orders processed per second
Batch processing overnight
✓ Behavioral copy-trade detection
IP-based linking only (VPNs defeat it)
✓ Economic calendar matched to trades
No news-trading detection capability