The Future of CFD Brokerage: AI and Machine Learning in Panda’s Brokerage Systems

Artificial intelligence is gradually creeping into all walks of human life. And yet, if you ask the management teams of most brokerages about the day-to-day running of their businesses, you’ll learn that it hasn’t really found its way into our industry yet, until now.  

At Panda, we always strive to be at the cutting edge of what’s possible. As a technology provider for the online trading segment, our reputation depends on us being ahead of the curve, so that we can have solutions ready to go when the rest of the industry starts inquiring about them. We also have the luxury of a large and very capable development team that’s keen on experimentation, and is chomping at the bit to try new things.  

So, a few years ago we decided to put a few of our best and brightest on the case of AI, to explore how the space was evolving, and to see if there were any interesting synergies to be achieved between recent developments in the AI and machine learning communities, and all the work we’ve been doing on the brokerage tech side of things. It turned out to be an extremely fruitful experiment that resulted in a couple of groundbreaking additions to our existing suite of brokerage products.  

Our goal for this project was to try and find novel solutions to the persistent pain points we hear from our brokerage customers. Working so closely with such a diverse array of industry players, from the scrappy start-ups to the household names, we have a thorough understanding of brokerage operations, the industry’s unique business cycle, and the various bottlenecks experienced by individual businesses. This has massively helped us in knowing where to put our attention and what areas are the ripest for disruption. 

AI Document Verification 

A consistent brokerage bottleneck that many people in the industry will be familiar with, is the struggle to on-board new clients when appetites for online trading peak and brokers are inundated with new registrations. The last time we saw this was during the lockdowns, but it’s a cyclical occurrence, usually hastened by a bull market in a certain asset class that has retail traders clamoring to get a piece of the action.  

During these times, brokers struggle with processing the deluge of incoming registrations as each has to undergo KYC/AML checks before they’re permitted to trade. Up until recently, most brokers were still processing these registrations manually, which can be incredibly dull, repetitive, and time-consuming work that takes valuable human resources from more important tasks.  

One of the most mind-numbing of these tasks is the rejection of document scans that don’t meet the quality grade. Our engineers set to work on using Google’s Vision AI algorithm in Panda CRM to scan incoming documents and automatically reject the ones that aren’t of sufficient quality. The CRM then requests a re-upload from the customer without a human member of the team having to get involved. Our brokers have reported that this feature alone has massively aided back office and compliance departments when it comes to handling surging registrations. 

But there’s more to the module than just that. Google Vision comes with some powerful OCR (optical character recognition) capabilities, which our engineers have harnessed for reading every uploaded document, then comparing the details between them as well as to what the brokerage has on file for that customer. The module then ranks each application with a score, which alerts staff as to the likelihood that the application will be approved without further back and forth with the client. This allows back office and compliance teams to separate applications, rapidly moving through the highest-ranking ones, while delegating the more problematic ones to the relevant team members. 

Next Call AI 

Our next AI-powered module, Next Call AI, has proven to be a real game-changer for our clients. Next Call AI is a project that’s focused on helping a different set of departments than our document verification module. Whereas the former was designed for back office and compliance teams, Next Call AI was built specifically with sales and retention teams in mind, as well as those members of staff tasked with managing them.  

The idea behind Next Call was that so much usage data available to modern brokerages is wasted when there are team members who would benefit greatly from having access to it. It makes no sense to be able to track a lead with surgical precision, or to know exactly what part of the client area a customer is on, while your sales and retention teams are still cold-calling lists of people from a spreadsheet.  

What Next Call AI does is combine customer usage data with an AI algorithm designed to sort through this data and identify the most important next call for these teams to make, based on the organization’s priorities. So, for instance, a customer is struggling to make a first-time deposit because the card is being declined. Next Call AI is designed to listen for these event triggers and inform available staff as to the highest priority next call to make. It could be as simple as reaching out to the customer and advising them to try a smaller amount. 

These event triggers are completely customizable and include things like client registrations, deposit page visits, card rejections, withdrawal requests, and more. The module is changing the way the brokers we work with organize crucial departments such as sales, and retention. They’re also finding that Next Call AI can be incredibly useful in the area of pre-emptive customer support.  

So, for example, customers who would ordinarily find themselves stuck on a call or live chat queue, can have their queries addressed proactively, especially when they are as predictable as the results of a failed deposit, or a beginner having trouble with the trading platform. In the past, the only way staff could know there was an issue was after the client had logged off and a report had been produced, or the customer contacted support themselves. By combining the existing knowledge regarding customer behavior of human teams, with the customizable triggers of Next Call AI, you can now be much more attentive to your customers’ needs, ensuring that every call made provides the maximum value to both them and to your business. 

More to Come!

As you can see, this is an incredibly interesting and useful area for modern brokerages trying to be that little bit more competitive in a crowded marketplace. At Panda we’re busy working on the next batch of AI inspired modules that will make our products even more valuable to brokerage businesses. To see the above modules in action, recommend a product development, or discuss your business’s unique requirements, please get in touch with one of our success liaisons, who’ll be happy to assist you.