Most people who set out to launch a retail brokerage start from the same place. They have watched the industry from the inside, usually from a sales or affiliate seat, they know roughly what a broker charges and what a client is worth, and the arithmetic looks generous. Spread times volume, multiplied by a few thousand accounts. On a spreadsheet that business prints money.
Then the first year happens. I have spent a long time building the software that sits behind these companies, which means I get a close look at the ones that work and an even closer look at the ones that quietly stop answering emails. The pattern is boring and it repeats. Very few of them fail because nobody wanted to trade with them. They fail because the money ran out somewhere between the licence application and the point where marketing started paying for itself, and nobody had modelled that gap honestly before they started.
Here is where it usually goes.
The licence you pick is really a budget decision
Founders tend to treat licensing as a legal step: pick a jurisdiction, hire a consultant, wait. It is actually the single decision that sets the shape of the whole business, because the regulator prices your business model for you.
In Cyprus, for example, the initial capital a firm has to put up depends on what it is allowed to do. A firm that routes every order out and never holds client money sits in the cheapest tier. A firm that safeguards client funds sits above it. A firm that deals on its own account, which is what most people mean when they say market maker, sits at the top, in the region of three quarters of a million euros. That is capital that has to be there in liquid form and stay there. It is not working capital and you do not get to spend it on Google Ads.
And that is only the regulatory floor. CySEC also wants to see the first year’s operating budget alongside the capital, which is the regulator politely asking the question most business plans dodge: how do you stay alive for twelve months if revenue is late?
The version of this mistake I see most often is a founder who plans a B-book operation because the margins look better, then discovers halfway through the application that the capital requirement for dealing on own account is several times what they raised. The application stalls. The runway keeps burning while nothing earns.
Regulation quietly changed what a client is worth
The revenue side of those early spreadsheets is usually built on numbers from a different era.
When ESMA capped retail leverage in the EU, it capped it hard: 30:1 on major currency pairs, 20:1 on non-majors, gold and major indices, 10:1 on other commodities, 5:1 on single equities, 2:1 on crypto. It also required margin close-out at 50 percent of required margin and negative balance protection on every retail account. Similar regimes followed elsewhere.
Lower leverage means smaller position sizes, which means less volume per deposit, which means the revenue you extract from an average client is a fraction of what it was before 2018. Meanwhile the standardised risk warning tells every visitor to your site that between 74 and 89 percent of retail accounts lose money, which does not make conversion easier.
None of this makes the business unviable. Plenty of firms are doing fine. But if your model assumes a client generates what a client generated ten years ago, your break-even sits somewhere much further out than you think, and you will only find that out in month eight.
The second budget nobody writes down
Almost every plan I see budgets for the launch. Licence, incorporation, platform licence, liquidity connection, website, a CRM, some staff. That number gets raised and the company gets built.
What rarely gets written down is the cost of acquiring clients until acquisition becomes efficient, which is a separate and usually larger number. Your first campaigns will be expensive because you have no data, no creative that works yet, and no reputation. Your affiliates will want to be paid before your clients have traded enough to cover them. Your payment providers will hold a rolling reserve. Your first support hires will be paid for months of answering questions from people who deposit two hundred dollars and leave.
The honest way to plan this is to write two budgets. One gets you licensed and technically live. The other keeps the lights on from the day you go live until the month where inbound revenue covers monthly cost, and you should assume that month is further away than your optimistic case. If you cannot fund both, you are not underfunded for a launch, you are funded for a launch and a slow closure.
Manual operations look cheap right up until they aren’t
The other quiet killer is operational, and it is the one I have the strongest opinions about because I watch it happen in the software.
A new broker with fifty clients can do everything by hand. Someone reads the KYC documents, someone else checks the deposit landed, a third person approves the withdrawal, and the whole thing runs on a shared inbox and a spreadsheet. It works. It even feels efficient, because hiring one more person is cheaper than buying a system.
At five hundred clients it stops working, and it stops working in the most expensive way possible. Withdrawals get slow. Slow withdrawals get posted to forums. Onboarding takes three days and the client funds an account somewhere else on day two. Nobody can produce a clean audit trail when the regulator asks, so someone spends a week rebuilding one from emails. The staff cost you were avoiding shows up anyway, just with churn and a damaged reputation attached.
Decide early which parts of the client lifecycle you will never do by hand, and build or buy for that from the start. Onboarding and verification, deposits and withdrawals, the partner and IB structure, and the reporting your compliance officer needs. Everything else can wait.
A pitch deck is not a plan
The last one is less about money and more about thinking.
Most founders arrive with a deck. A deck is a persuasion tool. It has a market size slide, a team slide, and a hockey stick, and it is designed to make somebody feel comfortable enough to write a cheque. A plan is a different document. It is the one where you write down which client segment you are actually going after, which execution model you are running and why, how much cash leaves the account every month before revenue arrives, what happens to your risk book on a gap, and what the first ninety days after launch look like day by day.
That document is also what regulators and banking partners want to read, and it is the only thing that will tell you, before you spend the money, whether the plan you have in your head survives contact with numbers.
If you want a sense of what belongs in one, there is a downloadable forex brokerage business plan that walks through the market description, the choice between STP, dealing desk and hybrid models, platform technology, cash burn and sales cycle, marketing, competition and a launch checklist. Use it as a template or use it as a list of the sections you have not thought about yet. Either way it is a cheaper way to find the holes than finding them in month nine.
The version I would tell a friend
Raise for eighteen months, not twelve. Pick the licence that matches the business you can actually fund, not the one with the better margins. Model client value using today’s leverage rules, not the ones you remember. Automate onboarding and withdrawals before you need to. And write the boring plan before you write the exciting deck.
None of that guarantees anything. Brokers with good plans still close. But almost none of the ones that closed on me had a plan that was wrong. They had no plan at all, and a spreadsheet that only worked if everything went right.
