Every broker that internalises any part of its flow carries market risk, whether or not it manages
it. MetaTrader records the positions. It does not tell you what they add up to, it does not offload
them, and it does not separate a client with an edge from one arbitraging your latency.
These components run as plugins inside the trading server: a threshold evaluated in-process as
positions change is a different product from a script polling the manager API every thirty seconds.
The suite is book-agnostic — the same exposure view proves an A-book desk's coverage and tells a
hybrid desk what to cover.
ComponentRisk Dashboard
Real-time dealing-desk visibility: net exposure by symbol and asset class, an A-book and B-book
breakdown, top clients by exposure and P&L, margin utilisation, unusual-activity alerts and
historical exposure analytics.
Aggregation is the point. Rolled up per asset class, six positions across the metals group read
as one trade in different clothes, with internalised flow separated from covered.
What it solves
- Answers "what is our exposure right now" on one screen
- Surfaces concentration while it is still a position, not a loss
- One factual picture for dealing and management, not two opinions
- Historical series let you review a bad session with data
ComponentAuto-Hedger
Rule-based automatic risk offloading. When net exposure on an instrument crosses a configured
threshold, the excess is hedged to a coverage account with no human in the loop. Thresholds and
hedge ratios are per symbol, throttles govern how often a rule fires, and executions are logged.
The hedge ratio is what makes it usable rather than blunt: full coverage removes the risk and the
revenue together, while a partial ratio caps the tail and leaves the book earning. Throttles
matter as much — exposure oscillating around a threshold otherwise pays the spread on a stream of
offsetting hedges.
What it solves
- Exposure is covered in seconds, including every unstaffed hour
- The written risk policy and the executed risk policy become one
- A small team can run a book that would need shift cover
- Each hedge records the exposure that caused it
ComponentDynamic Leverage
Automatic leverage and margin management. Leverage is tiered by position volume — full leverage on
retail-size positions, stepped down as volume grows — with settings per instrument class and
scheduled reductions ahead of news, weekends and market close. Client-facing tier tables publish
the policy.
Tiering by volume rather than by account is what makes a high headline figure honest: it applies
at the size most clients trade, and the ceiling engages only where an oversized position is
dangerous to both sides. A position opened at maximum leverage on a Friday afternoon is the
textbook weekend-gap loss.
What it solves
- Caps what one oversized position can do to the book and the client
- Gap risk over news, weekends and market close is handled by schedule
- High leverage can be advertised honestly, because it applies where it is harmless
- Published tier tables make a margin change a documented rule
ComponentPrice Feed Jammer
Control over the quote stream before it reaches clients: spike and outlier filtering, gap control,
spread policy by group, session and news window, minimum-spread enforcement, feed smoothing and
automatic failover between quote sources.
Start with the bad tick, because it does more damage than a bad day. One outlier from an upstream
source fires stop-outs that should never have fired and fills stops at prices that never traded,
and the broker pays the claims. Filtering protects clients and the book at once.
Spread policy is the other half. Through rollover, holidays and the seconds around a release,
liquidity is thin and the honest price is wider than at midday — which minimum-spread enforcement
per group and session states in advance.
What it solves
- Bad ticks stop causing wrong stop-outs, phantom fills and claims
- Thin-liquidity conditions are priced deliberately, not inherited
- A quote source can fail without clients noticing, with per-symbol overrides
- Account tiers become a pricing tool: one feed, several published policies
ComponentAnti-Abuse & Toxic Flow Protection
Detection of predatory trading patterns — latency arbitrage, tick scalping, coordinated
multi-account activity, news-spike gaming — with graduated responses from monitoring and flagging
through to per-group execution restrictions.
Detection reads server-side trade and quote data: holding times, fill timing relative to tick
arrival, and correlation across accounts that appear unrelated. That last pattern is the one
manual review rarely catches, because each account alone looks ordinary. Responses are graduated
for a reason: widening spreads for everybody prices out honest clients and leaves the arbitrageur
in place.
What it solves
- Toxic flow is identified while it runs, not in the month-end P&L
- Measures are per-account, so honest clients are not repriced for someone else
- Coordinated multi-account patterns become visible, which manual review misses
- Every restriction carries an evidence trail for disputes and audits
How it ships. Server-side plugins plus a dealing console. Thresholds, leverage tiers, spread
policy and detection rules are configured in one place, per symbol and per group. The dashboard
reads the same state the enforcement acts on — no second data path, no lag.
A-book, B-book and the line between them
These are not five products bolted together. The dashboard measures, the hedger acts, dynamic
leverage bounds how large one mistake can become, and the feed controls decide what price all of it
is measured against. Thresholds are per symbol throughout: gold and a small-cap share CFD need
different hedge ratios.
Routing is a risk decision. Almost every desk is hybrid, and the choice to internalise one
client and cover another is only as good as the exposure figure behind it — which is why the
dashboard and the hedger read one state.
Toxic flow is a routing input first. Flow that consistently beats your quotes belongs on the
A-book, not a restriction list. Naming it correctly turns a recurring loss into covered volume.