MarketAxess Paid Dealers To Show Up First

In 1999 a J.P. Morgan sales boss asked rival banks to fund a company designed to squeeze their own bond desks. Several of them wrote checks.

The pitch that should have been laughed out of the room

Rick McVey ran North American fixed income sales at J.P. Morgan. He put the idea into the bank's Lab Morgan program, an internal fund for web projects, and then spent months talking to trading heads, the syndicate desk, and the bank's own asset managers before he trusted the design. He has described the incubation as an inside job at a bank that had every reason to say no.

Then came the odd part. He took the same pitch to competitors and asked them for money. Three banks joined J.P. Morgan in the early rounds, and four more followed soon after. By the time the company filed to go public, its IPO prospectus disclosed that nine broker-dealer clients were also shareholders, among them Bear Stearns, Deutsche Bank, Lehman Brothers, and UBS. Most of the revenue came from those same nine.

The company spun out in 2000 into a crowded field. McVey counted the competition. There were about 50 other electronic bond-trading startups at the time. Only one of them is still trading today.

Somebody has to be paid to arrive first

Every marketplace opens with the same trap. Investors show up when dealers are quoting. Dealers quote when investors are asking. Nobody arrives first for free.

MarketAxess decided before launch that the dealers were the harder side and paid them in equity to be there. That is the whole trick. A bank that owns a slice of the venue puts prices on the venue, and prices pull in asset managers, and asset managers pull in the next bank.

Cash would not have worked. Cash runs out, and a dealer takes the check and goes back to the phone. Ownership sticks, and it converts a competitor into someone who wants your screen on their traders' desktops.

The first year produced no trading at all. It went into building the platform and getting client trading documents through legal review, which McVey remembers as painfully slow. Trading started in late 2000 with US corporate bonds, then eurobonds, then high yield, then emerging markets.

What it sold before it had liquidity to sell

The first product was a request-for-quote screen. An investor asks several dealers at once for a price on a bond and picks the best one. The investor stops making four phone calls. The dealer sees order flow; it would otherwise never have been shown.

That is worth noticing if you are building anything two-sided. Before the network is dense enough to be valuable, the product has to be useful to one participant working alone. Workflow first, network second. The liquidity arrives later and looks like a bonus.

Volumes grew through 2002 and 2003, helped by list trading that let investors rebalance whole baskets of bonds at once. The bank shareholders pushed for an IPO, which happened in November 2004 at $11 a share.

The rule change that opened up the other side

The 2008 crisis and the bank capital rules that followed did something the company could not have done for itself. Dealers were required to hold less bond inventory, which meant the side MarketAxess had originally paid to show up could no longer carry the whole market on its balance sheet.

The answer was Open Trading, an all-to-all model where any participant can supply the price, including an asset manager selling to another asset manager. A joint venture with BlackRock in 2013 helped seed it. Raj Paranandi, then chief operating officer for the firm in Europe and Asia-Pacific, told The Banker in October 2022 that "the world of monolithic liquidity provision through single protocols is behind us."

The subsidy bought the start. The rule change made the start permanent, because now the platform had two independent sources of price and did not depend on the goodwill of the nine.

By McVey's own accounting, the stock ran from $13 to $354 over the decade to 2020. In the second quarter of 2025 the company passed $1 trillion in quarterly credit trading volume for the first time.

Then somebody ran the same play on them

Tradeweb started corporate bonds in 2004 and got serious around 2014. Bloomberg pressed hard in Europe. Both attacked the parts of the market where MarketAxess was thin, dealer-to-dealer trading and portfolio trading, and share slipped from late 2023. In November 2025 its US high-grade share was 18.9%.

In July 2026, Intercontinental Exchange agreed to buy the company for $167 a share in cash, about $6 billion and a 33% premium to the previous close.

Subsidising one side gets you a market. It does not get you a moat that renews itself. The moat has to come from something the subsidy pays for on the way through, in this case a second, independent liquidity pool.

Few things to try before the end of the week

If you own a business or plan to own one in the future, here are a few things to note:

• Name your scarce side in writing. List both sides of your marketplace and score each on two questions: how many of them exist, and how much does it cost them to show up? The smaller, more expensive side is the one you pay. Time cost: 45 minutes. You know it worked when your growth spending stops being split evenly between the two sides.
• Find the non-cash currency. Equity is one option and usually the wrong one for a small company. Others: exclusivity for a window, category ownership, first access to data, a fee holiday that converts to a lower permanent rate. Draft three offers and price them against what a cash subsidy of the same size would cost you over two years. Time cost: two hours. Success signal is a scarce-side partner who asks to extend the arrangement rather than renegotiate it.
• Make the product work for one person alone. Write down what a single participant gains on day one with no counterparty present. If the honest answer is nothing, you have a network, not a product, and it will not survive the eighteen months before density arrives. Time cost: one afternoon with your first ten customers. Success signal is a customer who keeps using it during a week when volume is near zero.
• Check whether your subsidy has bought you anything durable. Ask what would survive if every subsidized partner left tomorrow. Data, workflow habits, a second supply source, and compliance plumbing. If the honest answer is nothing, your economics reset the day a better-funded competitor copies you. Time cost: one leadership meeting. Success signal is naming at least one asset the subsidy paid for that is now independent of it.

ICE is buying a company that solved the hardest problem in market design twice, once with equity in 2000 and once with a protocol in 2012, and then got caught out in the products it had not yet had to solve.