Baillie Gifford’s head of trading gives BEST EXECUTION an exclusive insight into his desk’s data literacy framework, along with a detailed dive into the real value of data (as opposed to its cost) and a discussion around the controversial question of ownership – is the monetisation of trade data disadvantaging the buy-side, and how can this issue be addressed?Â
To some, data is a four-letter word. Sometimes using data leads to great decisions, but at other times to frustration.
By definition, data analysis is the collection, transformation and organization of data in order to draw conclusions, make predictions and drive informed decision-making. The way people interact with data is unique.
Some of us will not act without the data to back up their decisions. Others use it to justify or measure the outcomes of their decision-making. There are those who gather as much data as possible to figure out what other people are up to (more of that later). And there are people who don’t want their decisions measured at all.
Whichever group or you sit in, data is only going to become an increasing part of our world.
On the Baillie Gifford trading desk, we recognise data literacy as a critical skill that is growing more important. In our Critical Skills Framework, we define data literacy as the ability to read, understand and structure data and processes, to inform decision-making and communicate the meaning, use and importance of data to others – internally and externally. I realise that’s a long description, but for me it demonstrates the need to embrace and develop these skills.
With the advent of advanced technology and proliferation of electronic trading platforms, swathes of trading data are generated and collected in real-time. There is no escaping that new technologies are evolving fast. The opportunities to interrogate our data, learn from it, and one day perhaps even monetise it, will only multiply.
When it helps articulate a point, data can be our best friend. With this in mind, the phrase ‘garbage in and garbage out’ becomes critical. Trade data is incredibly powerful when it is clean, properly identified and correctly labelled. At Baillie Gifford, our Trading Data Analytics team spends considerable time getting this right, even if it means a slower output.
This is not a small project. Data can be seen as a living organism that evolves and matures over time; a strong foundation is important. Above all, any data set must represent real, actual, client trades. It cannot be polluted by phantom or internal trades that do not help drive future decision-making.
The time it takes to fully complete this task fully may frustrate even the most sympathetic C-suiters but it is essential to get it right and there are few short-cuts. Every field must be exactly defined, verified, and not duplicated. When you get it right, delivery accelerates, and discoveries will be made.
So, what comes next? Clean data allows us better examination of how and why we do things. It also helps us to fill any gaps. Led by our terrific trading data analytical team, we are excited about the possibilities of FIX Tag 527.
This field, which our traders and analysts helped develop, means we can identify the algorithms being used by us, program traders or high-touch sales traders. Over time, it will allow greater comparison of similar algorithmic tools and how they perform. And it will provide a more statistically significant sample of data with which to do forensic analysis.
The irony of using data to analyse trading algorithms – which themselves analyse vast amounts of data to decide how and when to trade – is not lost on us.
The catchphrase du jour is ‘artificial intelligence’ or just AI. From simple machine-learning algorithms to robotics, we already use a multitude of applications daily, based on machines learning from previous experience. It will be interesting to see how this can be deployed in financial markets.
It is our responsibility not to get caught up in the hype. We should consider how our counterparts and suppliers are using it, rather than relying on blind faith for it to make trading decisions.
This seems a good point to consider the humble human being. Data helps us incorporate different behavioural metrics into our decision-making process. But our team is equally keen to maintain and value the relationships we have through high-touch sales trading with key partners.
While commentators love catchphrases such as ‘march of the machines’, the ability to create (and curate) liquidity is not yet the sole domain of the algorithm. We rely on skilled analysts to validate data accuracy, interrogate outcomes, and look at ways to improve what we do. Our trading data analysts are an integral part of the team and already possess skills that buy-side traders of the future will need to learn.
So, what is the value of data (as opposed to its phenomenal cost), and what is the answer to the question of ownership?
How we use our data and whether trade data (to which Baillie Gifford has been a counterparty) is being made available by others to impede the buy-side trader, is a matter of great importance. While data can be source of good, some arguably feeds the devil’s work. Let’s face it, no one would buy data if it wasn’t valuable.
Technology makes it easier to scrape data and find correlations hidden to the human eye. It can and create signals for some proprietary trading strategies to try and trade ahead of us, creating poorer outcomes for our asset owners. We think that every buy-side trading desk should own the problem and stop information leaking to trading strategies based only on personal gain. There is nothing inherently wrong with proprietary trading strategies so long as the playing field is level.
When trading venues and clearing houses consider selling flow data that signals our intent, then it deserves our attention. When that data has the power to penalise our own outcomes, whilst being monetised by unrelated third parties, the industry needs to consider how to approach the situation.
That ‘data is the new oil’ is not a new revelation, but when third parties are monetising data away from the buy-side trader (without whom there would be no trade) there should be a consistent approach.
While regulators also use data to monitor market behaviours, it is easy to abdicate responsibility and pretend it’s someone else’s problem. We believe it is the duty of both the buy-side and brokers to highlight the problem, articulate our concerns, and seek to stop the proliferation of data aggregation tools that do not serve our purpose.
Our conversation so far has focussed on equities, but data is asset class agnostic. We have equally interesting learning points in OTC markets, with significant progress in recent years for currencies and fixed income trading.
The collaboration between ourselves and our TCA vendors, BestX and Virtu, is a key part of our strategy. How to use trading data as learning points, shines a spotlight on both our trading performance and those of our counterparts. In FX for example, regulatory changes such as accelerated settlement cycles or SA-CCR potentially changes the ability to curate liquidity. The data will give us the evidence, illustrating which banks are reacting to change and providing the most competitive liquidity and spreads. Counterparty meetings going forward, informed by better analysis, will grow more important.
Earlier in this article, I mentioned our Critical Skills Framework. An increasing proportion of active trading is now dominated by data-driven strategies. Increasingly prevalent, these strategies are making a significant contribution to market liquidity. Clearly, trading data is the lifeblood of algorithmic trading and quantitative strategies, as they leverage historic and real-time data. It looks as if pre-trade consolidated tapes in Europe will not offer what we hoped for. t’s imperative that traders and analysts learn to understand the basis on which these sophisticated models work, and to reverse recent trends by deriving opportunities to benefit from them.
Data, and tools associated with managing, processing and using data are developing rapidly, as new techniques for making predictions and finding patterns come to light. For us, data literacy is a way to navigate through this increasingly complex world, by achieving what is best for our clients.
We’ve talked trading but should finish by talking about risk management. Our hope is that the eventual consolidated tape will (finally!) accurately define what is addressable or executable volume. I fear some decision-makers may have lost sight of how important accurate post-trade data is for liquidity and capacity analysis. Getting this right will play a major role in mitigating future systemic market risk.
I conclude by thanking our traders, analysts, counterparts and vendors for working together to develop these data products. They give us far greater insight into trading performance and risk management metrics. All this hard work is ultimately for the benefit of Baillie Gifford’s clients, and more widely for helping to maintain the integrity of financial markets.
©Markets Media Europe 2023
