The complete roadmap to starting a forex brokerage
Every decision a new MetaTrader 5 (MT5) brokerage faces, in the order you will face it — model, licence, technology, launch and risk. Vendor-neutral, written by the team that builds and operates this stack for brokers every day.
Choosing your model
This is the decision everything else inherits. It sets your revenue structure, your risk exposure, the capital a regulator will ask you to hold, and how complicated your technology has to be. Change it later and you are effectively relaunching.
A-Book (STP / ECN)
Client orders pass straight through to liquidity providers. You never hold the other side, so you earn from commission and spread markup rather than from client losses.
It buys you: no structural conflict of interest, pricing you can explain to a regulator, and the easiest story to tell a professional client. It costs you: thinner margins, real dependence on the quality of your liquidity, and less control over the price your client sees.
Suits: brokers going after professional and institutional flow, and anyone whose plan depends on reputation before volume.
B-Book (market maker)
You take the opposite side of client trades. Most retail traders lose over time, so the position is profitable in aggregate — which is precisely why it attracts regulatory attention.
It buys you: higher margin per client, a simpler technology footprint, and full control of pricing. It costs you: an inherent conflict of interest you must disclose and manage, exposure to a client who is consistently right, and a harder conversation with both regulators and sophisticated clients.
Suits: smaller launches with limited capital serving a retail audience — provided the risk side is genuinely built, not improvised.
Hybrid — what most brokers actually run
In practice the serious answer is neither. You internalise the flow you can safely hold and hedge the flow you cannot, deciding client by client rather than book by book. That is the model most established brokers converge on, and it is the one that makes the heaviest demand on your risk system: routing is only as good as the profiling behind it.
Where you licence
Jurisdictions are usually discussed in terms of cost and speed. The question that actually decides it is banking: a licence that no reputable payment provider will work alongside is not a cheap licence, it is a stalled business. Check banking access before you shortlist.
The second question is who you intend to sell to. Institutional clients and larger introducing brokers will filter you by regulator before they read anything else on your website.
| Jurisdiction | Indicative timeline | Tier | What the tier buys you |
|---|---|---|---|
| UK (FCA) | 12–18 months | Tier 1 | Widest institutional acceptance and the least friction with banking; the longest and most demanding application. |
| Australia (ASIC) | 6–12 months | Tier 1 | |
| Cyprus (CySEC) | 6–12 months | Tier 2 | Credible with most partners and clients, and a common base for EU and MENA-facing brokers. |
| UAE (DFSA) | 3–6 months | Tier 2 | |
| Seychelles (FSA) | 2–3 months | Tier 3 | Fastest route to trading. Expect to work harder for banking, PSPs and institutional counterparties. |
| Vanuatu (VFSC) | 2–3 months | Tier 3 |
Timelines are indicative and assume a complete application. Capital requirements and fees change, and differ by licence class within the same regulator — take them from the regulator’s own current fee schedule rather than from any vendor’s summary, including ours.
How brokers run more than one licence → · The licensing process in detail →
The technology stack
A brokerage is five systems that have to agree with each other. Most launch problems are not one system failing — they are two systems disagreeing about the same client.
Trading platform
MT5 is the default your clients will expect and your IBs will ask for. The decision is less which platform than whether you run it yourself or have it operated for you.
Liquidity and bridge
The bridge routes orders and carries your pricing. Connect more than one provider: a single LP means no failover, and no leverage in a pricing conversation.
Risk management
Net open position, exposure by symbol and client, and the ability to hedge without leaving the screen. This is the system brokers most often postpone and most often regret postponing.
CRM and back office
Onboarding, KYC, payments, withdrawals and IB commission. If this is a spreadsheet at launch it will still be a spreadsheet when it breaks.
Payments
More than one PSP, chosen for your actual client geography, plus a withdrawal process a compliance officer can defend.
Monitoring and support
Someone has to notice at 03:00 that the bridge has dropped. Decide whether that is your team or a managed service before you launch, not after the first incident.
MT5 platform support → · Payment gateway integration → · Bridges and liquidity partners →
A realistic launch sequence
The order matters more than the speed. Licensing is the long pole, so it starts early and runs alongside everything else; technology cannot be finished before you know which entity it belongs to.
Phase 1 — Planning
Define the client you are actually going after and the region they are in. Decide the model. Shortlist jurisdictions on banking access first, marketing second. Write the budget down.
Phase 2 — Legal and licensing
Company formation, corporate banking, directors and shareholders, then the application itself: documentation, fit-and-proper assessments, regulator queries, in-principle approval, and the pre-operational conditions that follow it. Expect queries; they are normal, not a setback.
Phase 3 — Technology
Platform and bridge selection, branding, server provisioning, CRM, payment integration. Run this in parallel with licensing so you are not starting a three-month build the week your licence lands.
Phase 4 — Operational readiness
Compliance manuals, KYC and AML procedures, support staffing and hours, the IB programme, and a tested withdrawal process. This is the phase most commonly skipped and it is the one regulators and clients both notice.
Phase 5 — Launch
Start with live testing on real money and small size, then a controlled group of clients before any marketing spend. Your first week should be boring by design.
Our advice on timing: add a buffer of roughly half again to whatever schedule you arrive at, and budget eighteen to twenty-four months of runway. Launches rarely fail because the plan was wrong; they fail because the plan had no slack in it.
Risk management from day one
If you hold any client flow at all, you are carrying market risk from your first trade. The question is only whether you can see it. A broker who cannot state their net open position right now does not have a risk problem in the future — they have one already.
What a risk desk needs to answer, continuously:
- Where am I exposed? Net open position across the whole book, by symbol and by client, updating in real time rather than at end of day.
- Who is costing me money, and why? Profitability alone is not a red flag — a client can simply be good. What matters is behaviour that reliably profits at the broker’s expense rather than from market direction, from latency arbitrage to swap farming, each recognised as a named pattern with the evidence behind the flag rather than collapsed into one number.
- What do I do about it? The defensible answer is routing: hedge the flow you should not hold, keep what you can. Decisions taken client by client, recorded, and explainable afterwards to a regulator or to the client themselves.
Two things worth saying plainly, because the industry often does not. Profiling exists to price and route risk, not to penalise clients who win — and every action it triggers should be one you would be comfortable explaining to the client it affected.
How Finnovic Shield handles this → · Client profiling and toxic-flow detection → · Real-time exposure monitoring →
Ten ways launches fail
| Mistake | What it costs you | What to do instead |
|---|---|---|
| Undercapitalisation | Running out before profitability | Budget 18–24 months of runway |
| Jurisdiction chosen on price | Banking refused, reputation damage | Check banking access before licence cost |
| Weak technology partner | Outages and client complaints | Take references from their existing brokers |
| No risk system | Losses you did not see coming | Deploy alongside the platform, not after |
| Compliance shortcuts | Fines, restrictions, closure | Budget for compliance as a running cost |
| No CRM | Operational chaos, poor onboarding | Choose it before launch, not after 500 clients |
| A single liquidity provider | No failover, no pricing leverage | Connect more than one from the start |
| Neglected support | Churn and public reviews | Staff the hours your clients actually trade |
| No IB programme | Expensive client acquisition | Launch the IB portal with the platform |
| An unrealistic timeline | A rushed launch that breaks | Add half again to every estimate |
Judging a technology provider
We sell technology to brokers, so treat this list as one you should apply to us as readily as to anyone else. Walk away, or ask harder questions, when you see:
- No live demo. A vendor who will only show slides is showing you a roadmap, not a product.
- Pricing that will not resolve. If the number changes depending on who asks, it will change again after you sign.
- No reachable references. Named clients you can actually speak to, not logos.
- Slow support during the sale. This is the fastest they will ever be.
- No compliance documentation. You will be asked for it by your regulator even though it is their software.
- Long lock-ins. A contract that is hard to leave is usually protecting the vendor from the product.
- Integration fees that appear late. Ask what the second year costs, in writing.
Planning a launch? Bring us the hard part
Thirty minutes with the people who build and run this stack — not a sales call. Bring the decision you are stuck on: model, jurisdiction, bridge, or what your risk desk should look like on day one.