Stop Flying Blind: A Real-Time Exposure Dashboard for A-Book, B-Book & Hybrid Brokers

Running A-book, B-book, or hybrid isn’t the hard part—running it with visibility is. Most brokerages can tell you yesterday’s P&L, but struggle to answer simple real-time questions like: “What’s our net exposure on EUR/USD across all books right now?” or “Which client segment is driving toxic flow in the last 15 minutes?”

A real-time exposure dashboard solves that by combining two data sources you already have: CRM context (who is trading and why it matters) and bridge/execution data (what actually happened in the market). Below is a concrete blueprint for building a dashboard that supports A-book/B-book decisions, hybrid routing, and hedging—without relying on gut feel.

1) What “real-time exposure” means in A-Book, B-Book, and hybrid risk

A-book exposure is often misunderstood as “no risk.” You still carry operational risk (LP rejects, slippage, last look), basis risk (markup vs hedge fills), and concentration risk (one symbol/session dominating volume). Your dashboard should reflect these realities, not just label trades as STP.

B-book exposure is more direct: you’re warehousing client risk. That’s fine when controlled, but dangerous when you’re blind to net open position (NOP), client concentration, and behavioral shifts (news spikes, latency arb, sudden scalping).

Hybrid risk sits in between, and it’s where dashboards pay off most. The key isn’t picking “A or B,” it’s measuring whether your routing logic is producing the intended outcomes:

• Are you internalizing the right flow?
• Are hedges firing fast enough when limits are hit?
• Are you seeing toxicity early enough to re-route?

2) The minimum dataset: what to pull from CRM vs what to pull from the bridge

You can build a useful dashboard without boiling the ocean. Start by defining the “contract” of fields you need from each system.

From the CRM (identity + segmentation + money movement):

• Account identifiers (login, account group, entity)
• KYC/region flags (for jurisdictional controls—check local regulations)
• Client segment tags (retail, VIP, IB-referred, prop evaluation, funded)
• Risk profile fields (manual risk tier, automated score, notes)
• Balance/equity/free margin snapshots (for stress testing)
• Deposit/withdrawal status (recent deposits can correlate with behavior changes)

From the bridge / execution layer (truth of execution):

• Symbol, side, volume/notional
• Execution venue/LP, fill status, reject reason
• Price, slippage, spread at execution, latency metrics if available
• Markup/commission components (so you can separate revenue from market risk)
• Routing label (A-book, B-book, internal match/C-book, hybrid rule ID)

If you have RiskBO or similar tooling, treat it as the “risk event stream” that ties these together: exposure updates, hedge orders, limit breaches, and routing decisions.

3) Exposure metrics that actually drive decisions (not vanity charts)

A dashboard becomes operational when every widget answers: “What action do we take if this moves?” Focus on a small set of decision-grade metrics.

Core exposure metrics (real-time):

• Net exposure (NOP) per symbol: base and quote currency exposure where relevant
• Exposure by book: A-book vs B-book vs internalized
• Exposure by segment: VIP vs standard, IB channel, prop evaluation vs funded
• Concentration: top 10 accounts contributing to exposure on a symbol

Execution quality overlays (especially for A-book):

• Reject rate by LP/venue and by symbol
• Slippage distribution (median + tail, not just average)
• Fill ratio and time-to-fill (if your bridge provides it)

Toxicity / behavior indicators (hybrid triggers):

• Rolling profitability by account and by segment (short windows like 15m/1h/1d)
• Latency outliers (accounts consistently beating your median execution time)
• News-spike sensitivity (volume and P&L clustered around high-volatility windows)

The goal is to connect exposure to who is creating it and how it’s being executed.

4) Building the dashboard: a practical architecture for CRM + bridge data

You don’t need a massive rebuild—just a clean data path and a consistent “position state.” A common pattern is:

1. Ingest events from the bridge (fills, rejects, amendments) into a stream or queue.
2. Enrich events with CRM context (segment, risk tier, region, IB) using account IDs.
3. Update a real-time position store (per account, per symbol, per book).
4. Compute aggregates (symbol NOP, segment exposure, LP exposure) on short intervals.
5. Render dashboards + alerts for dealing/risk/ops.

Where teams get stuck is step 3: you need a single, authoritative method for calculating open positions from execution events. Keep it deterministic:

• Only count filled quantities into exposure.
• Track partial fills explicitly.
• Treat cancels/rejects as execution quality metrics, not exposure.

From there, you can build views for different roles:

• Dealing/Risk view: symbol exposure, hedge status, limit breaches
• Ops view: rejects, bridge health, LP connectivity
• Compliance view: region flags, unusual activity, audit logs (check local regulations)

5) Hybrid routing: turning dashboard signals into rules and guardrails

Hybrid models fail when routing logic is static while behavior is dynamic. Your dashboard should feed a controlled loop: observe → decide → route → verify.

A practical way to operationalize hybrid is to define routing “lanes” with measurable criteria:

• Lane A (A-book): higher-risk or consistently profitable flow; accounts flagged for latency arb; large ticket sizes
• Lane B (B-book): small tickets, low leverage, historically unprofitable flow, low toxicity score
• Lane C (internal match): offsetting retail flow where your internalization is stable

Then add guardrails your dashboard can enforce with alerts:

• Max NOP per symbol per book
• Max exposure per segment (e.g., one IB network)
• Max LP concentration (avoid single-venue dependency)
• Auto-hedge thresholds (partial or net hedge) when limits are breached

The dashboard’s job is not to “pick winners,” but to prove your routing is behaving as designed—and to show when it isn’t.

6) Alerts and workflows: what to automate vs what to escalate

Real-time dashboards are only as good as the actions they trigger. Start with a small set of alerts that map to clear playbooks.

High-signal alerts (recommended):

• Symbol NOP breach (soft + hard limits)
• Sudden exposure jump (rate-of-change) on majors during session transitions
• LP reject spike (by LP and by symbol)
• Slippage tail event (e.g., 95th percentile worsens beyond threshold)
• Account/segment profitability spike (short window) paired with latency outliers

Workflow examples:

• Soft limit breach → notify dealing + suggest hedge size based on net exposure
• Hard limit breach → auto-hedge (if permitted) + lock routing to A-book temporarily
• LP reject spike → failover routing + ops ticket with bridge logs attached

Automation should be conservative and auditable. For anything that impacts execution outcomes, maintain change logs and approval controls, and consult compliance for jurisdiction-specific expectations.

7) Implementation checklist (what to do in the next 2–4 weeks)

If you want this live quickly, treat it like an ops project, not a data science project.

• Define exposure definitions: NOP, per-book exposure, and how you treat internalization
• List required identifiers: one consistent account ID across CRM, platform, and bridge
• Build the event pipeline: fills/rejects → enrichment → position state
• Create 3 dashboards: Risk/Dealing, Ops, Compliance
• Set initial limits: start wide, then tighten based on observed volatility and liquidity
• Add 5–8 alerts with playbooks and owners
• Run a two-week “shadow mode”: compare dashboard exposure vs backoffice reports daily
• Document routing rules and keep versioning (so you can explain changes later)

This approach reduces surprises and avoids the classic trap of building a beautiful dashboard that no one trusts.

The Bottom Line

A-book, B-book, and hybrid models are only as safe as your ability to see exposure as it forms, not after the fact. By joining CRM segmentation with bridge execution data, you can monitor NOP, concentration, execution quality, and toxicity in one operational view.

If you want a practical implementation path using Brokeret’s CRM + risk tooling, start here: /get-started.