Why streaming bond prices are back in fashion, and why RFQ has quietly outlasted every prediction of its demise.
I had a wry smile listening to the debate at FILS Europe about whether RFQ's days are numbered, and in the conversations I had afterwards with a couple of senior e-sales heads about where streaming might finally take the market. Back in 2004 I was at Barclays Capital, and we were already streaming executable US Treasury prices to clients through the BARX page on Bloomberg, which is over twenty years ago now. Either we were well ahead of our time, or fixed income has been in no great hurry to adopt something we already knew how to do, and I'd guess the truth is a bit of both, since looking back we were certainly a little optimistic about how fast the market would actually move once the technology existed.
That leaves two questions worth asking properly. If streaming executable prices have been around for two decades, why is RFQ still so dominant in institutional bond trading, and what, if anything, is genuinely different this time around?
RFQ: if it ain't broke…
Before electronic RFQ came along, getting a bond price meant phone calls, bilateral conversations and a fair amount of chasing dealers around, and electronic RFQ didn't so much change that process as make it a great deal less painful. Tradeweb put through its first electronic US Treasury trade in 1998 and then expanded into European government bonds from 2000, while Bloomberg's BondTrader went live around the same time, and all of a sudden clients could pull competing prices from several banks at once without losing half the day on the phone, which anyone who remembers the old way will tell you was a genuinely welcome change.
I think RFQ took off as much because of how the market was set up commercially as because of the underlying technology, since it suited pretty much everyone involved: clients got competing prices, quicker execution and a proper audit trail, while dealers could still weigh up who was asking, the size, the direction and the state of the market before deciding how hard to quote, or indeed whether they wanted the risk at all. The market kept doing business much as it always had, just a lot more efficiently, and that's probably why it has proved so hard to dislodge.
A bit of déjà vu from BARX
BARX was a single-dealer platform, so what we were doing was streaming our own prices to our own clients rather than aggregating several banks together, which is a real difference from today's multi-bank streaming services. What it did prove, though, is that the technology to price and execute continuously had been sitting there for decades. Barclays was also a heavy user of automated pricing and hedging in its own right, with the fixed income business one of the bank's biggest users of futures, hedging rate exposure automatically as client trades came through, so I don't really buy the idea that we've spent twenty years waiting around for someone to invent faster pricing technology. The more interesting question, to my mind, is whether banks actually want to change how they provide liquidity in the first place.
I was talking about this recently with a senior e-sales person in European government bonds, who had genuine reservations about streaming, mainly the speed it demands and the way it takes away the chance to look properly at each request before committing to a price. That's a fair point, even if I couldn't resist reminding him that we'd been automating pricing and hedging for a very long time by then. In fairness to him, being able to hedge quickly is a long way from committing to continuously executable prices, because you're carrying inventory risk, market risk and client risk all at once, and in an RFQ world, knowing who's asking, what they want and how long you've got to respond is genuinely valuable information. With streaming you simply put your prices out there and let clients trade against them within whatever limits and controls you've agreed, which is a different business altogether. My sense is the technology has probably been ready for years, and it's the economics, and the appetite to actually change, that are taking rather longer to catch up.
Follow the money
Which brings me to something that came up again and again at FILS: the economics of electronic trading. Banks provide liquidity because there's a commercial return in taking on risk, and venues have built substantial businesses around distributing that liquidity to clients. Depending on where and how a dealer trades, they can face transaction fees, subscriptions, connectivity charges and all sorts of negotiated arrangements, and those costs differ meaningfully from one platform to the next.
I heard this put rather more bluntly by another e-sales head, this one at a tier-1 bank, who told me the fees banks pay venues like Tradeweb, Bloomberg and BondVision have become a noticeably significant expense for liquidity providers, and that this is a good part of why the RateStream consortium exists in the first place. There's a real irony buried in that, because the dealer community has been here before. Tradeweb started life as a dealer-owned network before Thomson Financial bought the business outright in 2004, only for nine of the biggest dealers to pay $180 million three years later to buy their way back in at a valuation of around $1.5 billion. Some of that same group went on to do rather well out of Tradeweb's 2019 IPO, before LSEG's takeover of Refinitiv left whatever stakes remained diluted further still under yet another owner entirely. Two decades on, some of those same institutions are now backing their own streaming venue, which, whatever else is driving it, will hardly hurt if it also means fewer fees flowing to a platform the dealer community has already built, sold, bought back into and sold down once before. And those fees are not small: Tradeweb's own Rates business alone generated around $1 billion in revenue in 2025, and almost every cent of that is paid by liquidity providers rather than their clients, since on Tradeweb's model it's the dealers who pick up the transaction costs while the buy side trades for free.
If the same bank quotes the same bond to the same client on two different venues, should the prices come out the same? Not necessarily, because venue costs are only one input into a dealer's pricing and they don't flow straight through into every quote, but they are very much part of the overall sum the dealer is doing in their head. The banks need to make money, the venues need to make money, and the buy side, quite reasonably, wants the best execution it can get, and keeping all three of those things happy at once is never going to be straightforward.
RateStream: what's different this time?
This is why RateStream caught my attention. It comes from the people behind FXSpotStream and brings a model that's already proven itself in FX into government bonds, starting with US Treasuries, with European government bonds expected to follow later on: several banks streaming relationship-based executable prices through a single connection.
The commercial model is the part I find most interesting of all. Liquidity takers pay no transaction fees whatsoever, and the participating banks pay a flat fee instead of being charged per trade, which could genuinely change the economics of distributing liquidity. Lower marginal costs might encourage dealers to price more keenly, clients would be able to compare executable streams from several banks side by side, and the established venues would suddenly face some real competition, which all seems quite plausible to me. That said, lower venue costs won't automatically translate into better prices for clients, and we'll only really know once we can look at the actual execution outcomes.
There's also the harder problem of persuading institutional clients to change how they trade in the first place. They've spent years building workflows around the platforms they already use: order management, execution controls, allocations, analytics, post-trade processing, and the ability to switch between protocols as the trade demands, and no trader is going to abandon all of that just because someone has turned up with a new way of showing prices, however attractive it might look on paper. Any new service has to slot into the way clients already trade, not ask them to rebuild their whole process around it.
Is RFQ finally heading for retirement?
I wouldn't go writing its obituary just yet. Coalition Greenwich research from early 2024 put direct streaming at around 9% of US Treasury volume, which is only a snapshot but still shows just how firmly RFQ has held on despite years of innovation in electronic pricing elsewhere. It's also worth remembering that a bank can answer an RFQ using exactly the same continuously updated algorithmic pricing it would use for streaming, so what really differs is how the trade gets initiated and executed, and an RFQ response can be every bit as firm and executable as a streamed price.
Different trades will always call for different approaches, and RFQ, RFM, RFS, streaming and voice all still have their place. Streaming may well be ideal for smaller tickets in liquid govies, while a larger or more sensitive order is probably better served by a targeted RFQ or something else entirely, and sometimes the most effective piece of kit on the desk is still, rather wonderfully, the telephone. Old-fashioned, I know, but it works.
Maybe we're asking the wrong question
At FILS, Adam Ferencz, head of US credit at Aurel Partners, suggested RFQ in its current form could become obsolete within the next few years. His point was really aimed at credit, where he'd like to see a hybrid RFQ built around more pre-trade data supporting all-to-all liquidity, but it's a fair challenge to carry over into Rates as well. Maile Robichaud of State Street Investment Management took a rather different angle, arguing that interoperability between systems, rather than any single protocol winning outright, is going to be the next real evolution, since buy-side traders need pre-trade data and access to whichever tools and protocols a given trade actually calls for. And on another panel that same week, buy-side traders pushed back in yet another direction, warning that platforms and electronic liquidity providers can't always be relied on once markets get genuinely stressed, because liquidity has a habit of concentrating right back with the usual banks when things turn difficult.
My own guess is that the answer sits somewhere between all three of those views. Rather than arguing endlessly over which protocol eventually wins, we could be making it easier for traders to pick whichever one suits the trade in front of them, without having to hop between a pile of disconnected applications to do it, and without pretending that any single venue or model is going to hold up in every market condition that comes along. That's a good deal of the thinking behind our work on Ignite at RocketFin, where we want to bring different liquidity sources and execution methods together and leave the market itself to decide how it ultimately wants to trade.
Back to 2004
Looking back, I don't think we were wrong to believe in streaming at BARX; we were just a bit ambitious about how quickly the market would come round to it. Twenty-two years on, streaming is getting fresh attention again, backed by better technology, newer connectivity models and some genuinely different commercial approaches, while RFQ has proved remarkably durable throughout and will, I suspect, be around for a good while yet. I don't think the future of Rates trading is going to be settled by whichever protocol turns out to be fastest or cleverest, but by whichever combination delivers the best execution at the right cost, without making the trader's life any harder, and without quietly costing the banks a fortune in venue fees along the way.
If streaming does finally turn out to be the next big thing, those of us who were around in 2004 can at least say we saw it coming, even if it took rather longer than we expected, and came with a rather larger platform fee bill than any of us bargained for.
I'd be genuinely keen to hear what others think. Is streaming finally going to change the economics of government bond trading, or will RFQ prove just as difficult to shift over the next twenty years?