Provider demos converge. Everyone shows a dashboard, a risk panel and a payout screen, and all three look fine in a controlled walkthrough. The differences turn up later, usually when something has gone wrong and you need the system to behave in a way nobody demonstrated. These are the questions that surface those differences early.
Data: who holds it and how do you get it back
The first question is not about features. It is whether you can leave. Ask for a full export of trader accounts, trade history, KYC status and payout records, and ask to see the format. A CSV or an API response is an answer. A PDF is not.
Firms discover this at the worst time, when they have decided to move and find that two years of trading history exists only inside somebody else's interface.
The risk engine, in numbers
Every provider says the risk engine is real time. The useful question is what happens between the breach and the close, and what the measured latency is under load rather than on a quiet demo account.
| Measure | Typical range | What to ask for |
|---|---|---|
| Breach detection to position close | 200ms to 2s | The 95th percentile, not the average |
| Accounts monitored per instance | 5,000 to 50,000 | Behaviour at the top of that range |
| Daily loss limit evaluation | Per tick or per minute | Which one, in writing |
| Platform outage fallback | Varies widely | What happens to open positions |
Payouts are an operations problem wearing a technology costume
The software side of a payout is simple. The hard parts are the ones a provider cannot fully control, and the honest ones will say so.
- Rails and coverage. Which corridors are supported, and what happens to a trader in a country that is not on the list.
- Approval flow. Whether a human must approve each payout, and whether that human can be you rather than the provider.
- Failed transfers. Who chases the bank, and how a failed payout appears to the trader in the meantime.
- Timing. The spread between request and settlement, in business days, measured rather than promised.
The answers that should worry you
- "We can export that for you" in place of "here is the export endpoint".
- Risk latency quoted as an average with no percentile behind it.
- No written answer on what happens to open positions during a platform outage.
- A migration path described as straightforward, with no mention of re-KYC.
- Pricing that changes shape when you ask about the exit.
None of these mean a provider is bad. They mean the question has not been asked often enough, which is itself useful information about who else is buying.
Frequently Asked Questions
Who should own the trader data?
You should, in a form you can export without asking. Ask for a sample export during the sales process, not after signing. If the only export is a PDF report, the data is not really yours.
How fast does a risk engine need to be?
Fast enough to close a breaching account before the next tick matters. In practice most firms are comfortable under 500ms from breach condition to position close; above about two seconds you are writing off the difference as slippage.
What does switching providers usually cost?
The licence is the small part. Budget for data migration, re-KYC of the active book, and a quiet period where you stop selling challenges. Firms that have done it tend to describe it in weeks, not days.
