European fixed-income market structure focus
Trading venue consolidation, the consolidated tape launch, EMS adoption, and AI everything. What a great time for me to have joined Crisil Coalition Greenwich to analyze what this all means and where the market is headed.
Consolidation
First the acquisitions: ICE going all-in on bond trading in buying MarketAxess, and Tradeweb targeting grey market bonds, dealer workflows and new protocols via Trumid. The big question on everyone’s mind is what does this do to trading costs?
The fight for market share in U.S. credit trading has been brutal. Bloomberg, MarketAxess, Tradeweb, Trumid, and others have worked to attract users and keep them captive via new tools, data, execution methods, and lower costs in the form of fee holidays, rebates and more. Dealers have priced against platforms that charge them higher fees to encourage their buy-side clients to trade on lower-cost venues. That has been partially successful, particularly for some of the biggest banks.
That dynamic is already playing out in Europe, with MTS offering a subscription model to reduce trading costs for dealers. Major U.S. banks are publicly backing that approach at industry events, while both U.S. and European banks are pricing to reflect their lower costs of trading on the platform.
For trading desks on both sides of the Street, getting a clearer understanding of the costs of trading—both explicit and implicit—is material to efficient execution. Dealers are posting tight bid-ask spreads, asset managers are facing tighter margins, cost management is essential. How much can be trimmed from trading fees in Europe, and how much further bid-ask spreads can tighten, remains to be seen.
Transparency in European bonds?
Buy-side traders are expressing a need for clean, consistent data to be surfaced inside their trading workflows so that it can be shared with portfolio managers (PMs) before an order is raised. In June 2026, our Market Data Study found that 69% of European firms were increasing their data budget by 1-5%, with 16% of firms upping it by more than 5% this year. “Accuracy” was cited by 88% of respondents as “Very important/important” as a factor in valuing market data, with 91% making it the driver that investment firms consider when changing data providers, although cost is near the top as well. Trading desks are welcoming artificial intelligence (AI) where it removes manual analysis, and treats it with caution where it could replace judgement.
Buy-side desk behavior has also revealed a few surprises. The consolidated tape (CT) has been demanded for some time and is widely seen as important, but limited numbers of desks have it directly in front of traders so far. Real-time reporting of corporate bond trades has reportedly risen from around 5% to 75%, and government bonds from 49% to 80%. While impressive coverage and minimal reporting issues have been noted, the basic tape, managed by Etrading Software, has a reported 1.6 million users globally. Yet, a poll of the buy-side audience at a trade show in October reported limited use to date, with 34% having no plans to use it directly at all. Possible reasons include:
- Historically, changing buy-side fixed-income desk behavior can take time, and this early pulse may prove to be more about timing than utility.
- Where desks use specialist traders, common in fixed income, the level of information provided by the tape can be of limited value to affecting trading decisions, given existing information and expertise.
- Existing data providers are intermediating between trading desks and the CT, because CT data can be used with a republishing license, making direct engagement unnecessary.
- Mismatch between expected output and real output has not been a concern cited publicly, and this seems unlikely.
Data quality and usability is of greater interest to traders than access to more data. Consuming fragmented and non-standard sets of data from liquidity scores to dealer axes is a challenge. This can lead to use of a single market-data source for consistency, which may also reduce the direct CT engagement (see above).
Axes particularly have been consistently noted as “noisy,” a long-standing issue that is hard to resolve, given the need on the sell side to advertise prices which may later move, and value of axes to the buy side as a firm signal of which dealers to approach, in order to take down risk.
Building stronger desks
Despite a plethora of new tools and data sources that can be used to improve trader workflows, finding tools that fit existing activity is the key to adoption, not inventing better ways of trading. Vendors with some experience sitting on a desk are getting preference over pure technology providers and taking a consultative approach of “what keeps the desk awake,” rather than pitching a product that’s favorably viewed by traders.
Pilots can often stall due to limits on desktop real estate, muscle memory, data-sharing agreements, and legal queues. Desks often ask vendors to plug into existing platforms, accept portfolio holdings and not add another screen. Pilots can reportedly lose momentum after six to nine months, as barriers limit adoption speed.
Budgets are tight and so the decision whether to build technology, buy it or hire somebody can fall under a single collective budget. With fee compression limiting total spend, hiring a trader or a quant may be set against acquiring an EMS or spending on AI tokens.
The value of a service or tool is being judged at the point of trade. For example, one litmus test for continued evaluation has been whether a service or tool would materially change the outcome of the desk’s next 10 trades.
From EMS to MCP
Two technology trends cited on multiple desks are the adoption of an execution management system (EMS) overlay on top of an order management system (OMS). This is becoming increasingly common, as two streams of liquidity—highly liquid and illiquid—bifurcate, allowing a low-to-no-touch approach for liquid bonds and a choice of protocols, including voice for less-liquid order flow, depending on the instrument.
Our recent report, How EMS providers stand out in fixed income found roughly 29% of buy-side professionals from larger firms use an EMS for fixed-income trading and strongly prefer third-party solutions. Market-wide, we estimate around 15% of total firms are users, as smaller and midsize firms have lower uptake.
Adoption of large language models (LLMs) is broadly consistent, with natural-language queries improving transaction cost analysis (TCA) and trade data analysis, and improved trend-spotting across brokers, venues and protocols. Vendors are placing data behind MCP servers and using machine learning for pricing-confidence metrics. Desks are cognisant that they must challenge AI output.
The impact of new tools and data on buy-side trading workflows has been nothing short of amazing. The story for 2027 is one of increased adoption. For instance:
Smarter selection and dealer rules: Feeding pre-trade inputs, including cleaned and standardised axe and liquidity metrics, into dealer selection logic/smart-order router to optimize lower-touch trading.
The single-trader-facing data layer: Putting pre-trade, post-trade, internal, and external data in one place so a trader can see recent prints and likely buyers/sellers for a bond on order entry. EMSs are playing a role in this, reflecting the better quality of data available relative to three years ago, when many European buy-side desks saw no value in an EMS for trading fixed income.
Conversational analytics: Uploading digital and primary-market trade data to ask questions on broker, venue and protocol performance, and receive answers in minutes.
Pre-trade analysis shared with PMs: Tools that tell PMs before the order, how hard a trade is, a range pricing for holdings to be sold, and how long accumulating a position may take. This may be something PMs need to budget for rather than the trading desk, given the upside to their time. Improving their decision-making reduces the back-and-forth between the trading desk and PMs which can a) waste time by making corrections based on information the trading desk already had and b) contribute to friction if there is miscommunication or push back.
Looking ahead
2027 is shaping up to be the year of the request-for-quote (RFQ). Sell-side investment in pricing algos has continued to grow for both direct streams and responding to RFQs. They’re doing this to both serve clients better and operate more efficiently as profit margins in credit trading decline.
Conversations with market participants tell us that as dealers are investing in their algos to price RFQs better and tighter, RFQS are becoming more rather than less efficient for the buy side. RFQ is still the top D2C protocol. In U.S. corporate bonds, it has handled nearly a quarter of volume in 2026 (or nearly $15 billion per day), while portfolio trading is holding at 13% of total volume. Further RFQ innovations could see it steal share back from PTs in the year ahead.
