My Kingdom for an Axe!

Byron Cooper-Fogarty

With apologies to The Bard for the headline, you can imagine buy-side traders and portfolio managers paraphrasing Shakespeare’s Richard III, when they have a block to trade in Credit markets. Every seasoned buy-side trader or PM knows the ritual. Dealer axes populate your OMS, EMS, data aggregator – a list of securities a bank is motivated to buy or sell – and the first question isn’t “what’s the level?” It’s “how real is this?”

That instinctive scepticism tells you much about axe data quality in today’s bond markets. Despite the amount of money invested in fixed income technology on both sides of the Atlantic (and the considerable efforts of Neptune Networks as an axe distribution specialist), axe reliability – one of the most fundamental signals of dealer intent and liquidity – remains below where it should be.

The Axe as a Liquidity Signal — In Theory

The dealer axe is a powerful tool. When a bank is axed to sell a bond, it usually signals inventory overhang, potential price concession, and genuine liquidity. When axed to buy, it suggests demand and a tighter execution risk. For buy-side traders at asset managers and other buy-side firms, a clean, timely axe feed is as close to a real-time window into dealer balance sheets as they’re ever likely to get.

The problem is that in practice, axe data is too frequently none of those things. It can reflect a position that was already worked off. It can be a phishing exercise by bad actors on the sell-side. It can be broadcast so widely — to hundreds of counterparties simultaneously — that its informational edge has already evaporated by the time it reaches you – more on tiering below.

A Tale of Two (way) Markets

The quality problem manifests differently in Europe and the United States, shaped by each market’s distinct structure and regulatory DNA.

In U.S. credit markets, the sheer volume of dealer-to-client flow and the dominance of a handful of bulge-bracket banks has created a relatively mature, if imperfect, electronic axe distribution ecosystem. There seems to be more tolerance of imperfect axe data, and continued reliance on clunkier datasets such as “Runs” which are inherently “two-way”.

Europe presents a different set of challenges. The fragmentation of the European bond market — across currencies, jurisdiction. Whilst I’d argue that axes are more structured in Europe, they remain a challenge. MiFID II, for all its ambitions around transparency, did not directly address the axe distribution problem. Pre-trade transparency requirements were largely carved out for fixed income, leaving axe data in a grey zone where quality and timeliness remain highly dealer-dependent, but are also influenced by buy-side behaviour.

The Trading Venues aggregate and distribute axe data at scale, however, quantity doesn’t equal quality – our mantra at Neptune. The use of axes in these venues auto-ex  rules engines for “e” size orders and portfolio trading has led to a proliferation of two-way axes as the sell-side look to be selected regardless of which side they are truly axed. For the electronic sized trades, statistically this seems to benefit the buyside, however it is the primary use of axes, “voice” trading of larger or block sizes, that is suffering. As Charlie Munger said “Show me the incentives, and I will show you the outcome.”

There remains debate around what an axe is, it seems some interpret to suit what they are doing at the time. At Neptune we strongly feel that a real axe displays elements of these three attributes:

• It’s directional – the dealer has a preferred side (buy OR sell)
• Motivated pricing – the client can usually extract better terms than a standard bid or offer
• Time-sensitive – once the position is cleared, the axe disappears

As such, a two-way axe is simply not an axe, despite what some trading venues believe (see the Munger quote). It is a market and has a legitimate place, just not mis-labelled as an axe.

Having said that, some buy-side also need to look at their behaviour, and how it has led to this proliferation of two-way axes (again, see the Munger quote). We have had several dealers report back to us that certain buy-side firms have told them they miss auto-ex or PTs due to “not sending enough axes”!

Reporting is a key part of Neptune, and to that end we must share some of the blame as they naturally tend to be quantitative. To counter this we have produced a Data Quality Report for dealers which will be rolled out more broadly in the future. Whilst two-way axes are less of an issue on Neptune, in EUR IG we still saw a material reduction over Q1 as a result of these initial reports. The aim, which many dealers already abide by, is zero.

Whilst it is easy to observe and pick on two-way axes, the greater issue at hand are the “fake news” axes. We have all experienced them, they appear to have the above attributes and are only discovered once a buy-side trader has enquired on them and leaked information about their order. These are harder to observe and police ahead of that enquiry. This is where Neptune and linking axes to trading, to capture these instances will be able to make a difference. Measuring “no response” & price slippage may not necessarily identify individual “fake” axes, it will allow a buy-side client to be able to form a quantifiable picture of data quality for particular sell-side firms and specific traders. That can then form part of a robust trading and best execution process.

What Good Looks Like

The technology to solve much of this exists. Timestamped, machine-readable axe distribution with standardised FIX messages and the ability to tier axes per client based on size (“real size, real time” as a Neptune client often states) and/or level is achievable – much of it happens today, just not broadly or deeply enough. We observe on Neptune that the more progressive dealer desks are already applying this. By way of example, in EUR IG, there are less than 10% of axes that are “two-way” and this is heavily influenced by two dealers. These progressive dealers recognise that better data quality drives more “voice” flow their way.

Several buy-side firms have asked for more granular axe flags, that indicate more about the axe and inform the client on how best to engage. Algo flags already exist, however other flags focused on axe provenance may help weed out some of the “less real” axes. Having said that, the tech on the buy-side is often lacking, particularly the OMS & EMS vendors. Too frequently these systems come from an equities perspective and are still catching up on fixed income. Today several of these systems would not be able to display these flags, and increasing numbers of clients access axe data this way, aggregated with orders, and decreasingly through data vendor terminals or trading venue GUIs.

In future, AI-assisted axe validity scoring – flagging when a position is likely already worked based on real-time trading patterns (using TRACE and, eventually, CTP data) – is moving from concept to early deployment at a small number of institutions. Along the lines of the slippage or “no response” plans mentioned above.

Regulatory nudging could accelerate the shift. A relatively modest extension of MiFID II’s transparency framework to cover pre-trade axe data standards in Europe, or FINRA guidance on axe data disclosure practices in the U.S., could establish a baseline that raises the floor for the whole market. I’ll not hold my breath on either of those occurrences in the near future, however.

The Bottom Line

Bond markets have made enormous strides in electronification and transparency over the past decade. But axe data quality is one of the places, given how important that data is, that seems to have suffered over recent years. By all means, dealers can keep sending two-way axes to the major trading venues, if that works for them, their clients and the venue (as several claim it does based on their data); however, at Neptune Networks, we continue to focus on quality over quantity on behalf of your clients. Instead, utilise the simple technology to tier axes and support adding more granular information via axe tags. Likewise, those OMS and EMS vendors that service the buy-side invest in your fixed income offer to capture more granular information.

Until the industry — dealers, clients and vendors/platforms (we can’t wait on regulators) – treats axe data with the same rigour applied to post-trade reporting, fixed income participants will keep paying a silent tax on orders routed by information that may or may not reflect reality.

In a market where basis points matter, that’s not a rounding error. It’s a structural inefficiency hiding in plain sight.

The author is the Global Head of Neptune, and, as CEO, led the 70% sale to TP ICAP Group in 2025.