The Broker AI Benchmark.
A quarterly, 12-axis scorecard of 40 retail CFD, FX and prop firms — the first public read on how the retail-derivatives industry is actually doing on AI readiness.
What the numbers say.
The gap between the top and bottom decile is 60 points.
Tier-1 FCA-authorised brokers score in the mid-70s composite. Emerging brokers in unregulated jurisdictions score in the mid-20s. There has never been a wider AI-readiness gap in retail brokerage.
Governance is the strongest axis in the FCA cohort.
Brokers regulated under the FCA average 16 points higher on AI governance than any other regulated cohort — reflecting Consumer Duty and SM&CR pressure driving investment in the control layer earlier.
Prop firms outperform brokers on dealing intelligence and underperform on governance.
Prop-firm cohort dealing scores sit level with growth-stage brokers, but governance and surveillance scores are 20+ points lower. The regulator has not arrived; when it does, this axis will close hard.
IP ownership is the single most-neglected axis.
Only 6 of 40 brokers own the majority of their AI stack. Everyone else is renting it — a compounding disadvantage as sector AI advantage widens.
ModelOps is the axis that correlates most tightly with composite.
Brokers with mature evals, drift detection and a model registry outperform on every other axis. The evals layer is the single highest-leverage investment for a broker in the middle of the pack.
The ranked table.
Composite score is the weighted average across the twelve axes. Rows are ranked descending. v1 uses anonymised broker codes; v2 will name opt-in top performers.
| Rank | Broker | Region | Regulator | Type | Composite |
|---|---|---|---|---|---|
| 1 | Broker B | Global | FCA | Tier-1 | 76 |
| 2 | Broker A | UK/EU | FCA | Tier-1 | 75 |
| 3 | Broker C | UK/EU | FCA | Tier-1 | 72 |
| 4 | Broker Z | APAC | ASIC | Tier-1 | 70 |
| 5 | Broker AG | UK/EU | FCA | Tier-1 | 70 |
| 6 | Broker D | APAC | ASIC | Tier-1 | 67 |
| 7 | Broker AH | Global | FCA | Tier-1 | 67 |
| 8 | Broker I | Global | FCA | Growth | 61 |
| 9 | Broker E | UK/EU | CySEC | Growth | 59 |
| 10 | Broker Y | UK/EU | FCA | Growth | 58 |
| 11 | Broker G | APAC | ASIC | Growth | 56 |
| 12 | Broker AL | UK/EU | FCA | Growth | 56 |
| 13 | Broker F | Middle East | DFSA | Growth | 55 |
| 14 | Broker H | UK/EU | CySEC | Growth | 53 |
| 15 | Broker AB | Global | FCA | Growth | 51 |
| 16 | Broker J | Middle East | VARA | Growth | 50 |
| 17 | Broker M | UK/EU | FCA | Growth | 49 |
| 18 | Broker K | UK/EU | CySEC | Growth | 47 |
| 19 | Broker U (Prop) | Global | Unregulated | Prop-Firm | 47 |
| 20 | Broker V (Prop) | UK/EU | Unregulated | Prop-Firm | 46 |
| 21 | Broker L | APAC | ASIC | Growth | 45 |
| 22 | Broker AA | UK/EU | CySEC | Growth | 45 |
| 23 | Broker AF | Middle East | DFSA | Growth | 44 |
| 24 | Broker O | Middle East | DFSA | Growth | 43 |
| 25 | Broker AI | APAC | ASIC | Growth | 43 |
| 26 | Broker AN | APAC | Regional | Growth | 42 |
| 27 | Broker N | LATAM | Regional | Growth | 41 |
| 28 | Broker T | UK/EU | FCA | Emerging | 39 |
| 29 | Broker AK (Prop) | Global | Unregulated | Prop-Firm | 38 |
| 30 | Broker P | UK/EU | CySEC | Emerging | 36 |
| 31 | Broker W (Prop) | APAC | Unregulated | Prop-Firm | 36 |
| 32 | Broker X (Prop) | Middle East | Unregulated | Prop-Firm | 34 |
| 33 | Broker AC | UK/EU | FCA | Emerging | 34 |
| 34 | Broker Q | APAC | Regional | Emerging | 32 |
| 35 | Broker AJ | UK/EU | CySEC | Emerging | 32 |
| 36 | Broker S | Middle East | Regional | Emerging | 30 |
| 37 | Broker R | LATAM | Regional | Emerging | 28 |
| 38 | Broker AM | Middle East | VARA | Emerging | 26 |
| 39 | Broker AD | APAC | Regional | Emerging | 24 |
| 40 | Broker AE | LATAM | Regional | Emerging | 22 |
What we scored on.
Onboarding & KYC AI
weight 0.10AI-orchestrated KYC over Sumsub/Ondato/similar, live document diagnosis, jurisdictional gating.
Retention & dormancy AI
weight 0.10Daily churn scoring, VIP early-warning, segment-of-one journeys.
Dealing desk AI
weight 0.12Toxic-flow scoring, A/B-book routing recommendations, LP scorecards.
Trade & comms surveillance AI
weight 0.10AI-first false-positive reduction across voice/chat/trade.
Compliance-safe marketing AI
weight 0.08Jurisdictional gating, disclosure engine, ad-copy guardrails.
IB & affiliate intelligence
weight 0.08Commission audit, sub-IB scoring, fraud-ring detection.
Deposit + funding intelligence
weight 0.07PSP failover, abandoned-deposit recovery, chargeback triage.
Internal AI copilots
weight 0.07Dealer, support, compliance and product copilots inside the firm.
Data platform maturity
weight 0.08Unified customer + trade + comms data readable by AI systems.
ModelOps & evaluations
weight 0.08Versioning, evals, drift detection, model registry.
AI governance & regulator posture
weight 0.08Explicit control map to FCA/CySEC/ASIC/VARA expectations.
IP + data ownership
weight 0.04How much of the firm's AI layer lives inside the firm's own boundary.
How the numbers were built.
40 named retail CFD/FX and prop firms observed by the KJ Capital team over the last twelve months, plus published disclosures and product analysis. Broker identities anonymised in v1 to publish freely; opt-in named benchmark planned for v2.
Twelve axes chosen because each has a direct P&L or regulatory consequence. Weights reflect the median importance we see inside broker leadership conversations, not any single firm's preference.
Scores are 0–100 per axis, with the composite weighted by the axis weights that sum to 1.00. Sub-axis rubric available on request.
Findings are the patterns visible across the cohort, not any individual broker. The benchmark is directional, not audit-grade.
Want your firm scored (and named in v2)?
Opt-in named brokers appear in the v2 benchmark (Q4 2026). Every opt-in receives the full per-axis scorecard plus a private gap-to-top-decile assessment. Free.