Education
Exposure management for prop firms: the problem that shows up as you scale
Why hundreds of accounts running similar strategies concentrate risk invisibly, how to separate evaluation from funded, and what to measure before the mismatch catches up with you.
A prop firm with twenty funded accounts and one with six hundred don't have the same business at a different size: they have different problems. The one that shows up as you scale is almost never about acquisition, and almost always about aggregate exposure.
The problem you don't see coming
Each individual account has its loss limit, its maximum drawdown and its rules. All controlled, account by account.
The problem is that the accounts aren't independent of one another.
Your traders read the same analysis, follow the same signal accounts, trade the same pairs in the same sessions and react to the same news. Many use strategies bought from the same places.
The result is correlation. And correlation turns six hundred small positions into one large position that nobody decided to take.
On a normal day it doesn't show. On an employment print with everyone positioned the same way, it does.
What to measure
The metric that matters isn't exposure per account. It's the aggregate net exposure per instrument, and how it evolves over time.
Questions your system should be able to answer in real time:
- What's my net exposure in EUR/USD right now, summing all accounts?
- What percentage of my funded accounts is positioned in the same direction in the same instrument?
- How does that change in the thirty minutes before a high-impact release?
- If the market moves 100 pips against me, what's my aggregate loss?
If those four don't have an immediate answer, you're managing risk with a lag.
Evaluation and funded are different risks
This is the most important separation and the one most firms make late.
Evaluation accounts. The trader pays to try. Most don't pass the phase. Their flow, in aggregate, tends to be predictable in its statistical behavior and its risk impact is bounded.
Funded accounts. Here the trader has already proven something, trades real capital and their result is your result. The risk profile is completely different.
Treating both groups with the same routing policy leaves economics on the table on one side and takes on unnecessary risk on the other.
The reasonable configuration is to route them separately: the evaluation phase with one policy, the funded phase with another, and the ability to change the split without redoing the integration. It's the reason routing configurable by account group matters so much in this specific segment.
The step where it breaks
There's an identifiable moment in a prop firm's growth where the management model stops working:
Up to ~50 funded accounts: it can be managed with manual review and spreadsheets. Uncomfortable but viable.
Between 50 and 200: correlation starts to matter and manual review arrives too late. You need automatic aggregation by instrument.
Above 200: you need aggregate limits applied before execution, not reports after. A dashboard that tells you what happened is no use; you need controls that stop it from happening.
The classic mistake is passing the second step without changing tools, because "it worked until now." The cost of discovering it is a volatility event.
Pre-trade controls
The difference between monitoring and controlling is whether the system can reject an order that breaks an aggregate limit.
The controls worth having at scale:
Net exposure limit per instrument. An aggregate ceiling that can't be exceeded even if each individual account is within its rules.
Limit per account group. Applied to the set, not account by account.
Window restrictions. Different behavior in the minutes around high-impact releases, defined in advance.
Prior credit check. That the order is validated against the aggregate state before going out, not after.
All of these are pre-trade controls. A limit evaluated after executing is a report, not a control.
The conversation about strategies, before integrating
Prop firms are the segment where the most conflicts arise over trading profile, and almost all of them are avoidable by talking beforehand.
Your traders are going to scalp, are going to trade news and some are going to use automated systems. That's neither a secret nor a problem in itself. What is a problem is discovering in production that your counterparty considers it unacceptable.
Put the topic on the table during onboarding:
- What strategy profile your traders run, honestly
- What the counterparty considers problematic, with specificity
- What happens if they detect something: warning, routing change, or termination
- Whether you can segment to isolate a subset instead of contaminating the whole relationship
A counterparty that answers these four with specifics has a process. One that dodges them is reserving the discretion for when you already depend on it. The article on toxic flow develops how to defend yourself if the label is applied to you unfairly.
Latency: why it matters differently here
For a prop firm, latency isn't a vanity metric: it determines which strategies are viable on your platform.
If your traders run time-sensitive strategies and your real round-trip is 120 milliseconds over public internet, those strategies won't work — and your traders will notice before you do.
What's relevant isn't the number your counterparty publishes, but the round-trip from where your servers are. Ask for that specific mapping. Sometimes the honest answer is that you have to move your infrastructure, and that answer is worth more than a generic "sub-millisecond."
What to review each month
A minimal routine that avoids most of the scares:
- Maximum net exposure reached per instrument during the month
- Correlation between funded accounts: what percentage coincided in direction at the peaks
- Behavior in news windows: aggregate exposure in those minutes
- Accounts that approached limits, individual and aggregate
- Execution quality by group: if the funded accounts get worse fills than the evaluation ones, there's something to look at
Point 5 surprises more than one and is easy to check with your own logs.
Frequently asked questions
Should I hedge the evaluation accounts? Usually it makes no economic sense: evaluation flow rarely justifies the hedging cost. But the decision should be yours and configurable, not imposed by your counterparty's infrastructure.
How do I detect correlation between traders without analyzing each account? Start with the simple thing: the percentage of accounts positioned in the same direction per instrument, sampled every few minutes. You don't need a sophisticated model to see the pattern — an 80% spike in the same direction before a news release is visible with basic arithmetic.
How many accounts can I manage without an aggregation system? It depends on the homogeneity of the strategies more than the number. A firm with fifty highly correlated traders has more aggregate risk than one with two hundred diversified ones. If your traders come from the same channel or community, assume high correlation and bring the step forward.
Do aggregate limits bother traders? Well designed, they almost never trigger: they're there for the extreme case, not for daily operation. If they trigger often, either the limit is badly calibrated or your book is more concentrated than you thought — and in both cases it's useful information.
The detail of how this works in our infrastructure — routing by account level, aggregate exposure by group and pre-trade controls — is in execution for prop firms. Prop firms onboard as institutional clients and execute through us; the split between evaluation and funded is a configuration change, not a migration.
