By Zoë Balkwell, Global Head of Pre-Trade and Trading
For a long time, decision-making on a desk has sat in one of two places: in people’s heads, or in an algorithm. Neither is finding much that’s new anymore.
But speed only does one thing. It helps a trader act faster on a decision they’ve already
made. It does nothing for the decision they never knew they needed to make.
A desk runs on the experience of its people. The instincts that are built upon years of
relationships, half-remembered patterns, a feel for who needs what and when are real and
valuable. But instinct only covers what that desk has personally seen or heard secondhand.
A structural shift moving through a market they don’t cover, or a counterparty pattern sitting
quietly in someone else’s book three desks over – no amount of speed closes that gap. Only
visibility does.
The knowledge in people’s heads is real, but narrow, and it doesn’t transfer when they leave
the desk. And most of the algorithms running today were built on data the whole market has had access to for years. Whatever edge that data once had has mostly been arbitraged away by now; everyone’s looking at the same thing, which means it stopped being an edge a while ago.
What’s changed is that the kind of insight that used to only exist as someone’s hunch – a
pattern a trader half-notices after a decade on a desk – can now be surfaced from data no
individual has the time or reach to review manually. That’s the shift AI brings to this market.
The insight that used to require years of instinct to find is now something a sufficiently broad dataset can find on its own.
Pirum TradeConnect, plus our post-trade and collateral platforms, carries this market-wide
data: pre-trade availability, indicative pricing, matching, settlement, collateral, regulatory
reporting, all of it. That’s the asset: a view of the market that no single desk’s history or
algorithm was ever built to see.
That asset is what powers the pre-trade AI we’re building: it puts an opportunity in front of a
desk before that desk has gone looking for it: a borrower whose demand pattern says it’ll need a security next week, a lender sitting on inventory that’s gone quietly unused for a month, a rate that’s drifted out of line with where the rest of the network is pricing the same name.
I get the instinct. Desks are hiring for coding and AI fluency, and a good desk should build
its own tooling. But a model is only as good as the data under it, and a tool trained on one
desk’s own trade history sees only what that desk has always seen – which brings us right
back to the same bounded, stale-data problem.
Pirum’s data spans the whole network, which offers a unique layer for building your own
tooling on top of ours. Or let us handle it end to end. Either way, your own models get
sharper when the data feeding them is broader than your own book
This is how a desk experiences the problem, so it’s the frame I keep coming back to.
Diagnose – know a liquidity gap or settlement risk exists before it costs you anything. Fix –
resolve it with appropriate human oversight and as little unnecessary manual intervention
as possible. Prevent – use what the network has already seen to stop the same gap opening
again next time around.
What we’re building for pre-trade fits nicely into these buckets: surfacing what a desk would
otherwise have missed, using AI to resolve with oversight, and cutting down how often it
gets missed at all.
In physics, speed is a scalar: one number, how fast something’s moving, nothing else.
Velocity is the vector – speed and direction together. A signal that tells a desk where to look
is the direction. Execute as fast as you like once you’ve got it; I’d encourage it.
But direction is the part that’s been missing, and it’s the part worth building. All the speed
in the world just gets you to the wrong place faster if nothing told you where to go. Get the
direction right, and speed stops being the pitch – it’s just what you do next.
If you want to talk through what AI-readiness looks like for your own operation, get in touch.
Also included in the Quarterly Panorama:
Ben Challice, CEO, on Vision 2030, and why scale and connectivity are now table stakes, with the real advantage coming from turning network-wide data into intelligence and automation.
Ray Wong, Director of Product, AI & Data, on why there has never been a better time for Pirum to launch AI, and why infrastructure, people, platform, deep securities finance knowledge and client-validated use cases make Pirum unusually ready for this moment.
The Panorama also includes a regulatory and events calendar.