ICE offers advancements with customisation, AI and a depth of historical data
The DESK spoke to Varun Pawar, ICE Fixed Income & Data Services’ Chief Product Officer, about the latest trends in funds management, the use of AI, and how index providers are keeping pace with growing demand for customisation.
The DESK: What are the key index trends you’re seeing currently?
Varun Pawar: Right out of the gate, younger investors have automatically started gravitating towards ETFs. If you take a step back and compare an ETF complex versus a mutual fund complex, the ETF complex is a lot more liquid, and investors have full control over what they want to do, depending on their time of execution.
If you compare passive fees versus actively managed fees, especially for mutual funds, it’s day and night. The community of investors still using mutual funds who want potentially higher returns than traditional passive funds are engaging with asset managers to look into actively managed ETFs. If you look at 2023/2024 the numbers speak for themselves – more actively managed ETFs were launched in those two years compared to passively managed ETFs, and even mutual funds.
Index providers are assessing what’s already in the marketplace and how to offer something different and unique. This is an opportunity for us to offer clients the ability to look at indices on a very broad spectrum, moving from an idea to back testing it rapidly. The whole conversation has shifted to the speed at which we can stand an index up, and give it back, to let the clients adjust it with our Custom Index tool and pitch it to end investors.
TD: What demands are they making of you in the active space?
VP: One key demand is customization. Clients keep evolving traditional indexes, and traditional index rules, to come up with new concepts. They want us to tweak the rules and back test ideas and come back to them with data so that they can make decisions. If we had to deploy a team of people for every idea it would be very expensive, but about five years ago we built an application internally that allows our product managers to very quickly stand a new idea up. The application sits on top of approximately two decades of clean, historical data (at the bond level) that makes back testing and rules optimization a very straightforward process.
We take a broad base index and then start writing filters and caps, or objectives and constraints, and very quickly have the system ready to generate a historical timeseries.
Increased efficiencies in our R&D environment means we’ve gone from taking about two weeks to launch new customized indices to now doing it in a day or two.
TD: Looking at that application in more detail, how are you using things like artificial intelligence and machine learning to improve efficiency and precision?
VP: We are using AI right now to have full control of the index supply chain to make sure that the ingredients you use to build an index are pristine.
AI is helping us with capturing those raw ingredients, where we are transforming the workflow to be “validating data” vs “entering data”. For example, when there’s a new bond issued, we start seeing information trickle in from a variety of sources before any official documents. Then you see more information through documents such as term sheets, and finally, you end up with a broader prospectus of the final Terms and Conditions.
We have an in-house AI embedded system that looks at all this information and has been trained on those specific document sets.
From the initial information we want to identify who the issuer is, and the maturity that the issuer is thinking about launching. When the next term sheet comes out you get more clarity on the size of the issue and who the lead underwriters are. Finally, the end prospectus comes out, which has all the terms and conditions of the bond. The workflow has changed from an individual reading a document and typing the data into a database to AI reading the document and the user checking the confidence threshold, to decide if they hit ‘approve’ or ‘not approve’. We have dramatically reduced the probability of data errors. We’ve gotten very comfortable with data going straight into our databases with a 90% confidence threshold. As we get more data and document sets, we can scale this quickly across the stack.
TD: What does the future look like for the business?
VP: Our indices are rules-based, meaning there’s no personal judgment used for rebalancing and re-constituting an index. When the index rebalances each month/quarter, investors want a report on which names are entering and exiting, the latest returns, duration, yield, spread, and so on. The future of AI for us, is quickly taking that information and having an AI engine write the report. It will be sanity checked by a person for accuracy, but the world we’re moving towards, is one where we very quickly produce some of these reports in a nice, clean way that our clients deliver to their end investors or to downstream clients. We produce more than 7000 indices, so scale is key.
TD: How would you summarise your competitive differentiation?
VP: History is definitely one of the key factors – we’re one of the few providers that have deep historical data on constituent bonds going back up to two decades. The second key factor is our customization. The ability for us to take a concept or an idea, and in some instances, turn that around within the day. We can handle the entire life cycle and workflow involved with developing an ETF, from index idea, to research and rigorous testing, and then creation of that index. The third thing to manage with ETFs, is the whole create-redeem process. ICE Bonds is an electronic trading platform that helps clients manage that. ICE Bonds offers clients access to the ETF Hub, where we can help clients make that create redeem process more efficient. Finally, NYSE Arca is the home of ETF listings. Connectivity to this wider ecosystem to support ETF issuers means ICE Indices is in a unique position versus any other competitor. Our thesis is that ETFs are going to continue to grow, and we can play a role in various aspects of that workflow.
