How to start a prop firm.

A rule is finished when two people configure it identically and a trader who fails by it can be shown exactly why. "Maximum drawdown of 5%" leaves at least four questions open: five percent of what, measured when, against which balance, and does an open position count. Every one of those becomes a support ticket, and eventually a public complaint, if the answer is decided after the fact.

The related decision is who is allowed to say what happened. If the trading platform computes a fill and the back office computes a different one, there is no defensible position when a trader disputes a breach — the firm is arguing with itself in front of a customer. Risk authority covers how to draw that boundary and what to test before signing with any vendor.

Traders need real-time prices, and real-time prices are licensed. This is usually the largest recurring line in a new firm's budget and the one first-time operators underestimate most, because it is billed per user per month rather than as a flat platform cost — so it scales with exactly the thing you are trying to grow.

Which provider you need is decided by the asset class you are evaluating on, not by preference:

For futures specifically, most firms end up on Rithmic, Tradovate or NinjaTrader, or dxFeed for real-time data and simulated execution. These are proven and widely used, and there is nothing wrong with any of them — but they are all priced per active user per month on top of whatever you pay for the platform itself, and that combined per-seat figure is what determines whether your unit economics work at 50 traders and at 5,000.

This is the specific problem Nexus Connect exists to solve. It is licensed real-time market data plus the simulated execution environment, at $20 per active trader per month with no minimum and no platform fee — where an active trader means one who placed at least one order that month, not an account you provisioned. Run your own numbers against your current or quoted stack on the pricing calculator; the comparison that matters is total per-seat cost at the trader count you actually expect.

Almost every first-time operator plans to take payments with a mainstream processor and discovers the problem late. Prop firm evaluations sit in a category that mainstream processors treat as high risk: the product is an intangible with a payout obligation attached, chargeback exposure is real, and the model is close enough to categories they decline outright that underwriting rarely gets comfortable. Firms are routinely declined at application, and — worse — sometimes onboarded and then offboarded months later with funds held during a review.

That second failure is the dangerous one. Losing your processor after launch means you cannot sell evaluations, cannot process resets, and are explaining a freeze to traders who are mid-evaluation. It is a business-continuity risk, not a procurement detail.

What actually works is a specialist high-risk acquirer, or several. The catch is that the good ones have volume thresholds, and a pre-launch firm with no processing history is exactly who they will not call back. So the sequence matters: start these applications before you select a platform, because approval timelines are weeks and they are outside your control.

Written rules are worth nothing if they are enforced after the fact. Breach determination has to happen against live account state — the same state the trader can see — or you are informing someone hours later that a trade they placed and watched settle has failed them, which is the single most common source of public complaints about prop firms.

Practically this means real-time position and P&L that reconcile to the authoritative record rather than being computed independently in the browser, protective orders applied at platform level so a rule cannot be traded around, and a scheduled liquidation path for daily boundaries. It also means an operator can answer "why did this account fail" precisely, without an engineer.

Funded-trader programs attract organised abuse because the payoff is asymmetric: the cost of an evaluation is small and the payout is not. The common patterns are coordinated trading across accounts, hedging the same instrument in opposite directions across two accounts so one always passes, identity reuse across sign-ups, and payment fraud at purchase.

None of these are visible from a single account viewed on its own, which is why firms usually discover them at payout — the most expensive possible moment, when the money is already owed and refusing it is a public argument. Detection has to run across accounts and across the customer graph, and it has to inform a human decision rather than execute an automated denial: denying a payout on a probabilistic signal is a reputational cost that compounds.

Nexus firms also see signals from the shared fraud network — patterns and identities flagged across participating firms, so abuse that has already been seen elsewhere does not have to be learned again at your expense. Fraud prevention covers the signal types and how cases are assembled.

Three things, consistently. First, that program rules need to be written to a standard that survives an argument — 'maximum drawdown of 5%' leaves at least four questions unanswered. Second, that payout operations are a real function requiring people, evidence and consistency, not an afterthought. Third, that support volume is a direct function of what the trader cannot see for themselves.

The firms that struggle are rarely the ones with the worst software. They are the ones that launched before their rules were unambiguous.