The dominant data concern at FILS US this year is data is too siloed, sitting in separate systems that cannot talk to each other in real time. Holdings information, dealer inventory, research conviction, and live market pricing all exist within large asset managers, but combining them at the moment a trading decision needs to be made remains largely unsolved.
This constraint also caps the usefulness of AI, since any AI overlay is only as good as the data it can access, and if that data is fragmented, the analytical ceiling is hit quickly regardless of the sophistication of the model.
Even where data exists and is accessible, its quality and consistency varies significantly between instruments, geographies and market conditions, making it difficult to build models that transfer cleanly from one context to another. The substantial process work involved in extending a credit model in European markets vs US markets is clear, as a result of data availability, market conventions and liquidity dynamics in the latter.
Even where fixed income data is abundant, many firms find that extracting reliable, actionable signals from it, particularly in less liquid parts of the market, remains genuinely hard. AI has helped here at the margins, acting as democratiser by allowing firms with limited quant resource to ask better questions of large datasets, but the underlying data quality problem is not one that AI solves on its own.
While the lack of a unified post-trade data source in Europe is tangential to US trading, and being a market with a smaller notional value, less globally impactful, it is about to see a significant change to this structural competitive disadvantage relative to the US. fragmented snapshots of liquidity reported across multiple venues on one hand, and a single unified data source on the other, as “transformational” rather than incremental. The concern is both competitive — firms with larger balance sheets and deeper data resources extract disproportionate value from the current opacity — and operational, in that demonstrating best execution remains more expensive and complex than it needs to be. There was also a nuanced caveat: the tape will not solve everything, as large block trades will remain subject to deferred publication, preserving some information asymmetry at the top end of the market.
©Markets Media Europe 2025
