Welcome to the latest edition of At the Forefront: Fintech Conversations!
To learn more about this podcast and explore our episode archive, click here.
In this episode, Forefront VP and Head of Content Sam Belden is joined by Martina Rejsjö, Head of Product Strategy at Eventus, and Vinod Jain, Strategic Advisor at Datos Insights, for a deep dive into the current state — and future — of trade surveillance. The conversation centers on insights from their newly released research report, The Trade Surveillance Revolution, based on in-depth interviews with 20 global sell-side institutions.
The group discusses the factors driving firms to modernize their surveillance programs and how AI is changing the game, with explainability and governance as key pillars.
To start, Vinod shares how the study was conducted, emphasizing its global reach and depth. “We had approximately 45% of the respondents from the U.S. region, followed by 35% from EMEA and the remaining 20% from APAC,” he explains. The result was nearly 30 hours of qualitative insight from heads of compliance and surveillance across the industry. The goal? To better understand evolving strategies around regulation, technology and operations.
Martina then provides perspective on why firms are rethinking surveillance now. “Budget concerns is the biggest driver,” she says. “I do think they get asked a lot to do more with less.” But budget isn’t the only factor — regulatory pressure is another constant. “Fines come out from the regulatory agencies… the statements are often very detailed on the deficiencies that have been found, but also the expectations on what you need to have as adequate surveillance,” she adds. Market shifts, such as the move to 24/5 trading and expansion into crypto, are also creating natural inflection points for surveillance program reviews.
The discussion then turns to one of the most complex and important areas covered in the report: surveillance data. According to Vinod, firms are wrestling with data quality, reference data management and system integration challenges. “A small difference of a couple of seconds between various systems can really offshoot the market history you are going to present,” he notes. But even with these hurdles, both guests emphasized that structured, well-governed data offers a critical edge. “Once you have mapped it, you still have to continue to review it — because changes happen,” Martina explains. “And do you know how that change impacts your flow through the system down to the surveillance system?”
The conversation also challenges some common language in the trade surveillance space, particularly around false positives. Martina makes a strong case for replacing the term with “low-value alerts.” “They are triggered by your surveillance system, by your set parameters and thresholds, so they are in fact alerts,” she says. “But once triggered, you can deem them not to be of high value.” Her advice: don’t discard these alerts — use them to garner insights. “You can do a near-miss review… if an account is always being closed out by the system but triggers a lot, that should warrant a review of that pattern specifically.”
Next, Vinod highlights how firms are already using AI to reduce noise and improve alert quality. “The majority of the firms can reduce the automation and the alerts by 25% to 50% by using large AI models,” he says. In one case, a large investment bank implemented machine learning to detect rogue trading on a specific desk and reduced false positives by 80%.
Martina adds that while AI holds promise, regulators will demand transparency. “You have to be able to verify and demonstrate the output,” she notes. “That’s the challenge many firms are having right now.” She contrasts probabilistic models with deterministic ones, where the latter provides “clear structure and repeatable output.” That’s the direction Eventus is taking with its forthcoming deterministic AI solution, Frank. Designed to bring relevant data directly into analysts’ hands, Frank will provide enriched context with traceable, auditable logic — allowing firms to embrace automation without sacrificing oversight.
To close the episode, both guests emphasize that trade surveillance must evolve from a compliance checkbox to a strategic, insight-driven function. “Firms need to ensure they have the right balance of risk management to navigate the kinds of volatility that can happen in the system,” Vinod says. Martina highlights AI, data analytics and dynamic thresholds as key to better calibrating alerts and managing noise across markets and asset classes.
For more on how Eventus is helping global firms modernize trade surveillance through its data-driven, AI-ready platform, visit their website.
Download the full “The Trade Surveillance Revolution” report here to benchmark your strategy against global peers.
See below for a breakdown of what was discussed. Happy listening!
Timestamps:
1:22 – Top-line findings from the Datos Insights report commissioned by Eventus
3:24 – Factors driving firms to rethink their trade surveillance programs
6:05 – Core data challenges: quality, sequencing and integration across systems
10:30 – Why data ownership and governance are critical for surveillance effectiveness
11:57 – Reframing false positives and how automation can help
17:00 – How firms are using AI to reduce alert fatigue and boost case management
20:42 – Why explainability is essential for AI adoption in compliance programs
24:56 – How AI, cloud and APIs combine to cut costs and scale surveillance
26:24 – The challenge of unifying surveillance across regions and asset classes
30:01 – How standardized data unlocks operational efficiency and 24/5 oversight
34:01 – Why surveillance should be viewed as a risk management function
36:13 – What’s next: dynamic thresholds, better data analytics and AI-powered insight
