Licensed FX, CFD and securities brokers

The overnight surveillance desk your MT5 book never had.

Marqib reads the end-of-day trade, order, funding and login files a broker already produces, runs market-abuse and money-laundering rules across the whole book, and hands the MLRO complete case files by morning: matched trades, shared beneficial owners, funding sources, rebate exposure and a draft report. Deployed in the broker's own cloud account; no client data leaves.

MQB-260926-006 · SPOOFINGSCORE 0.89
00:00Nightly scan — 516,204 orders, 2,910 clients
00:02Rule hit — 212 EURUSD orders in 40 min, 209 cancelled < 3 s
00:03Fill check — 9 fills, all on the opposite side
00:04Client — Vantage Prop Ltd · proprietary trading · AE
00:05⚑ Pattern — layering consistent with SPK/DFSA market-abuse guidance
00:06Draft — STR narrative written from evidence rows E1–E18
→ Case file ready for MLRO review. Evidence pack hashed; goAML XML prepared.
Why brokers

Three things a broker's compliance desk cannot do by hand.

Watch every pair of accounts.

Wash trades hide between accounts that share a passport, a device or a funding IBAN. Finding them means joining the trade file to KYC and to deposits for every client, every night. Marqib does that join and shows its work.

Prove the IB was paid for it.

Volume-based rebates make matched trading profitable. Marqib links the flagged volume to the introducing partner's rebate ledger, so the case file states what the IB earned and the MLRO can decide with that number in front of them.

File on time, with evidence.

SPK, MASAK, DFSA and CMA expect a filing that cites records, not impressions. Every sentence in a Marqib draft cites an evidence row; the pack is hashed; the goAML XML is prepared. The reviewer approves, edits or declines; nothing is filed without a named person.

Coverage

Coverage for a brokerage book.

Rules are deterministic and tunable per firm. A language model never decides what is suspicious; where enabled, it only rewrites the reasoning from the same evidence table.

Market abuseWash tradingOffsetting trades across accounts sharing a UBO, device or funding source; rebate abuse.
Market abuseSpoofing and layeringOrder bursts cancelled within seconds with fills on the opposite side.
AMLThird-party fundingDeposits from a counterparty other than the account holder; relative-name heuristics reduce noise.
AMLDormant reactivationLong-idle account, new device or country, bank-detail change and a withdrawal in the same hour.
AMLStructuringRepeated deposits just under the enhanced-due-diligence threshold, followed by a crypto or wire withdrawal.
RoadmapCrypto wallet exposureScreening of funding wallets against sanctions and mixer lists.
RoadmapFront-runningClient orders placed ahead of large house or IB flow.
RoadmapSanctions and PEPScreening API in the firm's region.
Data

Five end-of-day files. No integration project.

Marqib reads CSV exports every broker already produces from MT4/MT5, the CRM and the payment gateway. A pilot starts on last quarter's history the day the files arrive.

  • trades.csv deal ticket, login, timestamp, symbol, side, volume, price, IB partner.
  • orders.csv pending orders with placement, fill and cancellation times (for spoofing).
  • transactions.csv deposits and withdrawals with channel, counterparty name and reference, status.
  • clients.csv KYC extract: type, country, beneficial-owner reference, occupation, expected turnover.
  • logins.csv · bank_changes.csv optional; enable dormant-reactivation and bank-change rules.
01

Morning queue

Cases ranked by severity, with holds already placed on pending withdrawals where the rule warrants it.

02

Case file

Why it was flagged, every evidence row, the subject's recent transactions, and the draft narrative; rule text and model text side by side.

03

Filing

Evidence pack with SHA-256, goAML XML for the UAE, MASAK format for Türkiye on the roadmap, and a hash-chained audit log of every decision.

Pilot programme

A ninety-day pilot on the broker's own history.

Installation in the broker's cloud account in week one; calibration on last quarter's files; live overnight queue from week seven. Three measures agreed in advance: reviewer hours per alert, STR turnaround, and matters the existing process did not find.

Fixed pilot fee
Credited against the first-year licence on continuation.

Request a pilot View the product demo →

Or write directly: anilabbak@gmail.com