David Wright, managing director and global head of FX electronic distribution, and Jian Chen, head of Quantitative Solutions and Innovations (QSI), at Morgan Stanley, talk about how Transaction Cost Analysis (TCA) is moving to real-time.
Profit & Loss: Given that all the major single-dealer platforms have a TCA component now, how do you look to differentiate yourself in this area?
David Wright: So where the post-trade TCA adds value from the client perspective is as a basic benchmarking tool for the algos that they’re running. Then on the pre- trade side where we initially differentiated ourselves was on calculating transaction costs under a number of regimes, giving clients the ability to better understand the expected execution costs of one algo versus another.
Now, we see the next generation of TCA developing as a real-time tool. What’s important here is not just that clients are able to see the transaction cost in real-time, but that this TCA helps inform them about how they should be managing their algos and the execution itself.
Jian Chen: The differentiation here is that we’re trying to make the TCA as fast and real-time as possible, as accurate as possible and then we’re trying to link it as closely as possible with trading.
P&L: What are the technical challenges associated with real-time TCA?
DW: From a technical perspective, doing anything in real-time is always more of a challenge. It’s one thing for someone to visually look at the data on Matrix minute- by-minute, it’s different if you’re looking to start driving an algo based on some of the same analytics.
JC: Predicting the future is always a challenging task in the FX world. There are a number of different hurdles, but from the beginning just trying to get all the data together and then turn that into a real-time prediction of market conditions is, by itself, challenging. This is especially true for less liquid currency pairs where there is less data or on very large trade sizes. Then representing that prediction accurately in real-time is another challenge.
P&L: Has your clients’ focus shifted with regards to TCA? There seems to be a lot of attention on market impact at the moment….
DW: Market impact is certainly the topic de jour, but honestly, it’s something that we’ve been looking at for some time. We think it’s an extremely important metric, not just for the broader impact of the order, but also for the individual trade slices and the actual placement of orders. We actually think that market impact is a metric that needs to be used much more broadly than it is even now.
There are lots of things that can cause market impact like information leakage or whether your algos are participating in a broader market trend – which can look like market impact when really there’s other trading happening at the same time. So we aim to give some leading indicators about what could be causing market impact, and then diving into it and fixing that is a different can of worms.
P&L: What implications does real-time TCA have for the human trader?
JC: Transaction cost can normally be broken down into different components. There are the visible costs, such as the margin spread or up-front fees, and then there are implicit costs, such as the risk premium if you’re doing a risk transfer trade or the algo performance cost. And even this performance cost can be broken down into two main aspects, one is the cost coming from the execution of a particular algo and the other is the decision making. So the trader chooses a specific algo and what time to run it and what limits they want to input, etc and this is all part of the decision- making process.
What we’ve seen over the past year is that clients have become much more conscious of their workflows, they’re trying to compare the execution from different algos in order to improve their decision-making process. So in the next five to 10 years we will see human traders become more systematic in terms of this decision-making process, using tools provided to them by bank liquidity providers or other firms to inform decisions about how they execute a trade. This decision making process will become more dynamic as TCA becomes real time and, in the longer run, we will gradually have more self-adaptive algos.
P&L: What does the roadmap look like for you on the TCA technology side?
JC: I think that real-time TCA will take a long time to mature, so right now we’re trying to build something that can inform human traders and then gradually try and give them more intelligent suggestions to help them make decisions, and then seamlessly connect the trading ability of firms with those suggestions.
DW: I think we’re just at a starting point right now in this space. We’re approaching FX TCA very differently from equities, where the algos have a lot more acceptance and market share.
What we’re trying to do here is expose pre-trade information that clients should be thinking about and then offering real- time information to help them drive their algos during execution. It’s an area that we’ll be investing heavily in and we expect to have some product releases by the end of the year.