How traders can use their Python skills on the desk

Published on 20 July 2021

Developments in execution management systems (EMSs) have enabled equity traders to adapt their trading tools using Python coding. This can allow for more efficient flowing of data through the trading and investment workflow, and ultimately lead to greater levels of automation based on a trading desk’s own execution model.

Dan Barnes: Welcome to Trader TV – your insight into trading for professional investors. I’m Dan Barnes. If buy-side traders can use their coding skills to incorporate data into their execution management system, they can affect better execution. Joining me today to discuss how to do this is Quentin Limouzi, head of order and execution management systems & equity trading solutions at London Stock Exchange Group.

Quentin, welcome to the show.

Quentin Limouzi: Dan, it’s a delight to be here. Thanks for having me.

Dan Barnes: So tell us, can TCA data be incorporated into the equity trading work flow to better support execution at trade?

Quentin Limouzi: Can it be? Yes. Is it? Absolutely.

And that’s definitely where we’re spending a lot of our time and effort. Traditionally, we’ve always had data talking to trading data, to market data, to real time data and analytics, and then you’ve got your trade data in your EMS coming out of your OMS workflow. What we’re trying to do in a better way now is really to overlay that trade data that’s live, that’s sitting in your blotter as an execution trader with the analytics that are supporting your trading decisions, so we have better signals and we help you in automating that view of your blotter and decision making.

In my group fx, right now we’re working on a product called Trade Performance Analytics, or TPA, and that’s basically overlaying market impact models, new liquidity measures for a lot of our clients around venue entropy, around fractions, to our blotter coming from our EMS. So our clients will not only have a good view of what potential market impact they would have on their trades, but they will also have the tool to see if they made the right decisions, if the market has changed. What impact does it have vs different benchmarks vs their own execution? So therefore, we see pre-trade and at-trade are very complementary as the loop keeps turning, because really at-trade is also pre-trade with the leaves of the orders, throughout the time of the order.

Our analytics platform on top of the execution platform, rather than being a black box and giving you models and numbers, we open that model and we let our clients, through Python, add on their secret sauce and let them see what’s behind the model. And we’ve heard from the clients that it’s definitely something that they were looking for as it allows them to better track, but also change the analytics that are better suited to themselves in a way that is not impossible, because it’s right there in Python, in front of them with the data flowing through. So the work to get there hopefully is lesser and the clients can spend more time working the alpha out of the model rather than how to get it in there.

Dan Barnes: As we see skills develop on the trading desk, there’s always been the risk that technology doesn’t involve with the skill sets the traders have. As traders now learn Python, that means they can evolve their tools with their own skill sets, is that right?

Quentin Limouzi: Absolutely. I mean, you’ve read in the news over the last four or five years, that being able to code on the trading desk is a prerequisite for a lot of people. I would also say that having a Python-capable analytics platform within our distribution model is also an easier way for a buy-side or a sell-side desk to have the traders who code, code in the right numbers and then distribute them to the rest of the desk, rather than having everyone running their spreadsheet, which has always been costly and difficult on the trading desk.

Dan Barnes: That’s really interesting. I was going to ask next about the different types of asset managers and what sort of interests they have, because obviously if you’re trading for one portfolio vs another type of portfolio, you will be looking to reduce market impact, reduce implementation shortfalls, etc., but some of your priorities might change and that presumably affects what you’re trying to achieve via your EMS?

Quentin Limouzi: Absolutely. And I would say in general, there is an understanding that automation is usually for larger investment managers or perhaps for the systematic trading desks. But really what automation is, is simplifying your menial tasks and helping you make the right decision. So automation is everywhere in the trading workflow; like putting an order from APM to a trader via the portfolio management system, through the order management system, going through the what-if-scenario that a PM would do, going through the compliance and risk, going through locking and slicing and then ending up to the trader. You know, the trader should have a choice and should also be able to have an automated tool to help them make their decisions on how to send that order. But to keep that order as easy, medium hard or as better destination, better algo. It doesn’t necessarily mean that you’re automating the entire workflow and the trade goes all the way to the exchange via a DMA order through broker X, Y, Z or without the wheel.

And I think that’s something that’s very important for the clients to understand. You can automate all the way or you can automate part of it. In any case, you’re making your process better.

Dan Barnes: So then what are the challenges with incorporating internal and external data within the EMS today?

Quentin Limouzi: I’ve gone through the stats in the past and I will not today because they change every second. And basically what it means is, there is so much data that is being created all day long that it can be very costly and it can be very time consuming, and can be very difficult to prioritize what is good data vs what is bad data is to a trading strategy. In our position at LSEG, we produce data and we ingest data and we distribute data, so it is not a trivial task. But I’m very fortunate to work for a group where I have all of those very difficult tasks taken care of by a group of very professional people, so that’s a huge advantage.

Dan Barnes: And then how is the equity EMS today? Incorporating trading algorithms and the automation we discussed a little earlier to deliver better execution.

Quentin Limouzi: It always goes back to the quality of the data from which you can derive quality analytics and then triggers to potentially trade or bucket, as we talked about. So if you have the workflows and the mechanics around automating your trading, but you’re lacking that data, you really don’t go anywhere. Incorporating these tools together is really key for the trader of today and tomorrow, in order for them to be able to do more with less and have a better understanding of what their blotter looks like and what their day looks like, without having to go through and repeat actions every day that frankly do not add any value.

You know, in the old days of going through your blotter in the morning and as the day evolves and looking line by line of what the impact may be because of today’s liquidity measures on the back of historical volume, what you want to achieve does not make sense when the data can give you that. I still think that having a trader pulling the trigger and making sure that everything makes sense is essential, but I think the best of both worlds is having great automation tools with great traders.

Dan Barnes: That’s great. Quentin, thank you so much.

Quentin Limouzi: Thank you. It was a pleasure.

Dan Barnes: I’d like to thank Quentin for his time and insights and of course you for watching. To catch up on our other shows or to subscribe to our newsletter, go to TRADERTV.NET.