Advanced fraud detection
Catch it before you pay it
Most of what damages a prop firm is not market risk. It is coordinated abuse: copy trading across accounts, hedging against your book, farming payouts across firms. The engine holds the payout.
Evidence attached to every flag
How abuse actually shows up
Six patterns the engine is built around. None of them are exotic. All of them are running on somebody’s book right now.
Copy trading
Timing, direction, size and instrument compared across every account. Clusters get a score.
Internal hedging
Opposite positions inside your own firm, so one account passes whichever way price goes.
Reverse hedging across firms
Mirrored books at two firms, one payout guaranteed. Caught on pattern and timing, not one trade.
IP and location anomalies
Impossible travel, shared addresses and VPN patterns across accounts that claim to be unrelated.
Device sharing
Hardware and browser fingerprints matched across sign ins. Ten accounts, one machine, one person.
Banned trading behavior
Martingale, one sided betting, high frequency bursts and other prohibited strategies, flagged as they form.
Operator view
Two ways this goes wrong
Every operator has felt both of these. The engine exists because judgment alone does not scale past a few hundred accounts.
You pay too early
A clean looking payout clears, then the same trades show up on four accounts you already paid.
You hold on a hunch
You are wrong, and an honest trader tells a Discord of thousands what your firm did.
The difference is evidence: the trades, the timings and the matches next to the decision.
Account #48213
Copy trading, 96% correlation
Matched against #48197 and #48166 across 214 trades
- Device fingerprint
- 2 shared
- Reverse hedging
- 1 match
- IP anomalies
- 4 countries
Illustrative flagged account shown in the Admin Portal.
What the engine watches
Signals scored continuously, including how each trader normally behaves. A flag is a pattern, not a hunch on payout day.
Account scoring
A live risk score that moves with behavior, not a check at withdrawal.
Trade matching
Entries, exits, size and timing compared across the whole book.
Behavior patterns
Session times, sizing and reaction to drawdown set each trader baseline.
Device and network
Shared devices, addresses and impossible travel link unrelated accounts.
Payout patterns
How often, how much and how close to a rule edge an account withdraws.
Payout gate
Flagged accounts stop with evidence attached, clean ones pass through.
From flag to decision
The engine builds the case. Your team makes the call. Four steps, and a human decides at the end of all of them.
Step 1
Detect
A pattern crosses your threshold and the account is scored. Nothing is decided yet, the signal is recorded.
Step 2
Correlate
Matched accounts, shared devices, timing windows and prior flags are pulled into one case, not six alerts.
Step 3
Hold
A payout in flight stops at the gate with its case. The account keeps trading, only the money waits.
Step 4
Decide
Your team approves, rejects or asks for more. Decision, evidence and reason go on the record.
Why evidence matters
Every hold carries its proof, so the decision survives contact with the trader, the payment provider or a lawyer. Clean accounts clear faster too.
- Matched trade list
- Timings and windows
- Correlation score
- Decision on the record
The engine flags. Trade Guard works the case.
Trade Guard, our managed operations team, is trained on this engine. Its agents review every flag, from challenge to funded stage.
The things people ask about fraud detection
What about false positives?
The engine produces a score and a case, never a verdict. Nothing is closed, refused or messaged automatically. Thresholds follow your rulebook and are tuned once you see how your own book behaves.
What happens to the trader when an account is flagged?
The payout stops at the gate. The account is not closed and trading is not suspended unless your team decides it should be. The trader is told the withdrawal is under review, then cleared or given the decision with the evidence.
Can the rules be tuned to how my firm actually operates?
Yes, and they should be. Firms differ on grid and martingale, news trading and how much correlation between friends is acceptable. Thresholds, watched patterns and what escalates are all configured to your rulebook.
Does this slow payouts down?
Not for clean accounts. Scoring runs continuously rather than starting at the withdrawal request, so most payouts pass the gate without stopping. The ones that stop arrive with the case already assembled.
What data does it use?
Your own firm’s records: trading activity, sign in metadata, device and browser fingerprints, and the account and payment details traders give you. Cross firm hedging is inferred from pattern, timing and behavior inside your own book.
Do I need Trade Guard to use the engine?
No. The engine works on its own, with your team reviewing the cases. Trade Guard adds a managed team trained on it, so every flag is reviewed by agents who already know the evidence and your rules.
Next step
Hold the payout, not the relationship
Thirty minutes on your rulebook, the abuse you are already seeing and the thresholds that fit your programs. You leave with a plan and a fixed price.
Figures shown in product screens are illustrative.