THETRADETECHFX DAILY from the floor
How do you decide which data is most useful for an FX investment decision? The expanding data universe creates real opportunities to improve how we measure economic and market states. However, just because the data universe keeps expanding doesn’ t mean our models should keep expanding proportionally with it. We start from the investment question or economic theme we want to understand, not from the datasets that happen to be available. The first question is therefore not‘ does this dataset back test well?’ but‘ what economic variable is it measuring, and why should that matter for the assets traded?’
A new dataset must clear three hurdles. Firstly, there needs to be a credible economic rationale. Secondly, it needs to provide genuinely incremental information relative to what we already capture. Thirdly, that additional information needs to be valuable enough to justify the cost, complexity and the effort required to implement and maintain it.
We particularly like data that gives us a faster, higher-frequency or genuinely different view of an economic state we already understand. In FX there is an additional challenge because relative information matters: a global growth signal may be useful, but currencies often depend on relative growth, inflation, policy or capital flows. Often the right response to more data is better measurement of those themes, not simply more inputs.
What signals do you find most useful for spotting changes in the FX market early? Indicators of change and surprise tend to be more useful than levels or more structural measures. The interesting information is often not that growth is strong or inflation is high, but that our estimate of the underlying state has changed materially relative to previous expectations.
Faster-moving data can therefore be particularly useful when it starts to diverge from slower official or modelbased indicators. The focus on change also reduces the need for any single dataset or model to be all-encompassing – each item provides a separate perspective that contributes to a broader view.
In FX, we also distinguish between global regime information and more relative or regional information. Global risk appetite, volatility or broad USD conditions may matter across the whole market, which can be suitable for factor-style baskets. Individual currency views still tend to depend on relative growth, inflation, policy or flows. Positioning, flows, sentiment and higher-frequency macro data can all be additive within this framework.
When do you trust the data, and when do you rely more on your own judgement?
Using data smarter in FX
PER IVARSSON, quantitative researcher at Systematica Investments, sits down with The TRADE to discuss the increasingly important role of data across FX investment decisions, including which signals are most useful, and where the line lies between data usage and trusting your own judgement.
Our distinction is less‘ data versus judgment’ and more a question of where human judgment enters the process. We are systematic in implementation, but the system itself is the product of human decisions.
As such, our judgment is applied structurally rather than day by day. This brings greater rigour and clarity to the decision framework, while reducing the need for discretionary calls under shortterm pressure. We are therefore careful not to treat a drawdown, by itself, as evidence that the model needs to be overridden.
This judgment consists of guiding research towards economically plausible hypotheses, deciding which datasets map onto them and how the data should be transformed. Once this framework is established, the model is designed to apply those decisions consistently across markets and over time. Clear attribution is important here: we want to understand which building blocks are behaving unusually and whether the underlying data or economic relationship has changed. Those concerns then feed back into the longer-term research process, rather than prompting an ad hoc override.
Looking ahead, what new types of data or signals do you think will have the biggest impact on how you make FX decisions? One new exotic dataset is unlikely to transform FX trading. The underlying economic drivers will remain broadly the same, even if their importance varies. The bigger opportunity is better and faster measurement of states we already know matter, such as activity, inflation, policy expectations, flows, positioning and sentiment. Each new dataset is more likely to refine than revolutionise the process.
Higher-frequency data that helps identify turning points earlier is particularly interesting, provided it has a stable economic interpretation. The growing range of datasets lets us combine multiple noisy observations to improve our estimate of an underlying state, rather than simply create more standalone signals.
AI and newer tools will have a large impact on the research process by accelerating data cleaning, coding, testing, documentation and the use of unstructured data. There is, however, a paradox: cheaper research makes it easier to test thousands of weak ideas, while the FX universe is neither broad nor stationary enough for a purely statistical approach to be robust over our trading horizons. That means stronger priors, economic logic and outof-sample discipline become even more important. The principle remains: start from the themes you want to capture, not the datasets available.
The official newspaper of TradeTech FX 2026 TheTradeNews. com 23