MarketAxess Holdings Inc. MKTX

Key data

The business

MarketAxess runs the largest electronic marketplace for US corporate bonds. Asset managers, insurers and hedge funds meet dealers there to buy and sell credit, and the company earns a commission on each bond that changes hands, a few basis points of face value, plus a smaller and steadier stream from selling pricing data and post-trade processing. The moat is a network: its all-to-all pool lets any participant face any other, so more users mean better prices mean more users. That pool dragged corporate credit from phone-and-fax to screens over two decades and made MarketAxess the incumbent in US high-grade.

The tension sits on top of that. Tradeweb has grown fast in the newer protocols, portfolio baskets and dealer-run sessions, and has been taking share. The easy part of the shift, small liquid high-grade tickets, is largely done; the next leg, big blocks, less-liquid names and emerging markets, is harder and more contested. So the question is whether the incumbent still compounds or is now defending a mature, share-losing base.

The numbers

The five-year record is the whole argument, so start with the quarters that built it.

The quarters show a business that stopped growing. Q2 2026 revenue was a shade below the year before and net income fell to $68M. Reported EPS has come in above analyst estimates four quarters running, by 2% to 8%, but those are beats on a flat top line, not growth. Step back to the full years and the point hardens.

Revenue grew from $699M to $846M over five years, about 5% a year. Diluted EPS went the other way, $6.77 to $6.64. More revenue, less per-share earnings. The reason is the third table.

Revenue rose 21% across the span and operating income did not move, as mix shifted toward lower-fee protocols and rate normalization drained the new-issue rush. The one comfort: that margin, the number to watch, held at 40.5% over the last twelve months and 41.1% in Q2, so the flat earnings are a growth-and-mix problem, not a cost blowout. What the market disbelieves and I think is defensible is that a business still growing mid-single-digits, at 40% margins with net cash, sits at 13 times cash earnings, below any multiple it carried in twelve years. The monthly volume release, and the high-grade share number inside it, settles which read is right.

Management

Capital allocation is where this team places its own bet. In 2025 the company bought back $420M of stock, up from $75M in 2024 and nothing in 2023, and cash fell from $544M at end-2024 to $246M by June 2026. That is a deliberate lean into the buyback as the multiple compressed, retiring shares at what management evidently reads as a cheap price, and it is why EPS held flat while net income fell 10%. Insiders sent no such signal with their own money: eleven small sales over the past year totaling $717K, the largest about $299K in September 2025, none offset by a single purchase, plan status not disclosed. Pay leans on stock, about 4% of revenue in dilution a year. The guidance record is clean, four beats in four quarters.

How it fails or surprises you

Tradeweb keeps taking share. MarketAxess's edge was being the only deep all-to-all pool; Tradeweb now runs credible portfolio and dealer sessions and has grown high-grade share for several quarters. If MarketAxess's estimated US high-grade share slips another point or two through 2026, mid-single-digit growth turns to zero and 13x looks generous. First tell: the September and October volume releases.

The margin is the real leak. The fact this memo explains least well is that revenue rose 21% over five years while operating income stood still. If mix keeps drifting to lower-fee protocols and the 40% floor cracks toward the mid-30s, earnings fall outright and the cheapness was a warning, not a gift. The read is wrong if operating margin prints below 39% for two straight quarters.

A credit dislocation lights up the network (right tail). MarketAxess earns most when bonds are hardest to move, exactly when spreads gap and dealers step back and everyone needs the all-to-all pool. A real credit event, the kind the MOVE index front-runs, could push volumes and all-to-all adoption to records the market is paying nothing for at 13x. First sign: a volatility spike, then a monthly volume record.

Closing thoughts

This is the old question of whether cheap is real or a trap, and a named number resolves most of it. At 13 times cash earnings and 19 times reported, below any multiple in twelve years, the market is pricing a business in slow, permanent share loss. The evidence is split: revenue still grows mid-single-digits and the margin held at 40% last quarter, which argues the fear is overdone, while earnings have gone nowhere for five years and a well-run rival keeps taking the growth, which argues the market has it right. The left tail only opens if the margin breaks; hold 40% and the earnings base is durable, with net cash and a $420M buyback doing the rest, and the multiple becomes the whole story. What is at risk if share and margin both slip is another stretch of no earnings growth at a still-full price; what it is worth if they stabilize is a re-rating off a trough.

The bet is still that big investors keep buying and selling corporate bonds on a screen instead of over the phone, and keep doing it on MarketAxess, and that it holds its cut against Tradeweb. What breaks it is not volumes, which keep rising, but share and margin. The two numbers that tell you first are the estimated US high-grade share in the monthly release and the operating margin in the quarterly. Hold both and this is cheap; lose either and the price was honest all along.